2024 IJSS SurveyFuzzyControlMechatronics
2024 IJSS SurveyFuzzyControlMechatronics
2024 IJSS SurveyFuzzyControlMechatronics
net/publication/376377595
CITATIONS READS
0 41
3 authors:
Saso Blazic
University of Ljubljana
147 PUBLICATIONS 2,220 CITATIONS
SEE PROFILE
All content following this page was uploaded by Anh-Tu Nguyen on 09 December 2023.
a
Department of Automation and Applied Informatics, Politehnica University of
Timisoara, Timisoara, Romania
b
Center for Fundamental and Advanced Technical Research, Romanian Academy –
Timisoara Branch, Timisoara, Romania
c
Laboratory LAMIH UMR CNRS 8201, Université Polytechnique Hauts-de-France,
Valenciennes, France
d
INSA Hauts-de-France, Valenciennes, France
e
Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Slovenia
*
Corresponding author, Bd. V. Parvan 2, 300223 Timisoara, e-mail address:
[email protected]
A Survey on Fuzzy Control for Mechatronics Applications
Abstract. Fuzzy control has become one of the most effective tools for dealing
with complex engineering processes. Over the years, research on fuzzy control
systems has continuously evolved, witnessing numerous theoretical contributions
and successful real-world achievements. The concept of model-free or data-driven
fuzzy control was initially introduced with specific heuristics incorporated into the
design. Due to the lack of a systematic framework for stability analysis in model-
free fuzzy control, the significance of model-based fuzzy control has grown
extensively. This approach ensures systematic design based on precise fuzzy
models of the process. This survey focuses on the fundamental aspects of three
prominent classes of fuzzy control. First, the paper commences with a review of
Takagi-Sugeno fuzzy control systems. This includes discussions on stability
analysis and controller design, exploring techniques to derive less conservative
and/or complex results from a numerical burden perspective. Second, various
aspects of data-driven fuzzy control are analyzed in detail including a
classification of the most popular data-driven control techniques and their
combination with fuzzy control; a representative Iterative Feedback Tuning-based
fuzzy controller is described. Third, this survey explores the fundamental aspects
of evolving fuzzy control, with a particular emphasis on the significance of
stability and control laws, which are not usually the primary focus of evolving
intelligent systems research. For each discussed class of fuzzy control, the paper
provides a selective list of mechatronics applications to illustrate their
performance effectiveness, emphasizing research papers published after 2011.
Finally, drawing from recent advances in fuzzy control theory and mechatronics
applications, future research directions and associated challenges are discussed.
& Fodor, 2010). The incorporation of advanced hardware and software architectures in
dynamics, steady-state and robust performance under various operating conditions, and
mechatronics applications are now ubiquitous in daily life, including aircraft, spacecraft,
and trains. Along with the pressing demands for high performance, reliability and
affordability, the need to make these mechatronics applications safer, greener and
intervention. To achieve this goal, new sensing, actuation and control technologies have
challenges justify the study and research interest drawn towards fuzzy control for
controllers. If the controller design is carried out systematically, these challenges are
properly solved and thus transformed into advantages of using fuzzy control for
mechatronics.
As shown in (Precup, & Preitl, 1999a), next (Precup, & Hellendoorn, 2011) and
“average temperature”, “high humidity”, “low pressure”, “very high speed”, referring to
the variables specific to the behaviour of a controlled process, are subjected to relatively
variables and information that are processed with well-specified numerical values. In
this respect, the design and tuning of the controller and its subsequent implementation
in the control system require a quantitative modelling of the controlled process that is as
structures and/or values of the parameters that characterize these (parametric) models.
Lotfi A. Zadeh established the foundations of fuzzy set theory (Zadeh, 1965),
science in the 1970s opened up the first prospects for practical applications of the theory
in the field of automatic control, and these first applications belong to Ebrahim H.
Mamdani and his co-author Sedrak Assilian (Mamdani, 1974; Mamdani, & Assilian,
1975). The reference application of fuzzy control involves some “special” controllers
based on fuzzy set theory, called fuzzy controllers, in cement kiln control (Holmblad, &
Ostergaard, 1982). In the 1980s, the so-called fuzzy boom took place in Japan, the USA
and later in Europe in the field of fuzzy control applications, covering several domains
ranging from the electrical household industry to the control of vehicles, transportation
systems and robots. This is partly due to the spectacular development of electronic
technologies and computer systems, which made it possible to manufacture circuits with
very high information processing speed, dedicated (by design and use) to a specific
efficiently use a large amount of information about the controlled process and the
control equipment.
The applications of fuzzy control reported until now point out two important
aspects related to this control strategy (Precup, & Preitl, 1999a; Precup, & Hellendoorn,
• In certain situations (such as the control of nonlinear processes that are difficult
(conventional) control.
• Compared to conventional control, fuzzy control can rely heavily on and focus
on the experience of a human operator, and a fuzzy controller can model this
The main features of fuzzy control are discussed in (Precup, & Preitl, 1999a;
Precup, & Hellendoorn, 2011; Precup, Roman, & Safaei, 2021) as follows:
• Fuzzy control uses the so-called fuzzy controllers or fuzzy logic controllers that
• Fuzzy control can handle multiple variables from the controlled process.
Therefore, fuzzy control structures are considered to belong to the class of Multi
Input-Multi Output (MIMO) systems with interactions, and fuzzy control can be
feedback controllers.
• Fuzzy controllers are basically without dynamics. However, the applications and
into fuzzy controller structures, resulting in the so-called fuzzy controllers with
in industrial applications.
the performance specifications imposed on the control systems and then elaborating the
control signal in terms of “natural” IF-THEN rules belonging to the set of rules
...
IF (antecedent ) THEN (consequent ), (1)
...
where the antecedent (the premise) refers to the current situation concerning the
controlled process evolution (usually compared to the desired evolution), and the
consequent (the conclusion) refers to the measures that should be taken – in the form of
the control signal u – in order to fulfil the desired evolution of the control system
imposed by the performance specifications. The set of these rules forms the rule base of
fuzzy control system considered as a single-input system with respect to the reference
input r and a single-output system with respect to the controlled output y is shown in
Figure 1 (a). The reference input is also referred to as the set-point or the desired output
y d . The second input fed to the controlled process / the fuzzy control system is the
disturbance input d .
Figure 1. Fuzzy control system structure (a) and fuzzy controller structure (b) (Precup,
& Preitl 1999a; Precup, & Hellendoorn, 2011; Precup, Roman, & Safaei, 2021).
Figure 1 (b) also highlights the operating principle of a fuzzy controller in its
variables and modules: (1) the crisp inputs, (2) the fuzzification module, (3) the
fuzzified inputs, (4) the inference module, (5) the fuzzy conclusions, (6) the
defuzzification module, and (7) the crisp output. In this regard, the operating principle
of the Mamdani fuzzy controller includes the sequence of operations (a), (b) and (c)
(Precup, & Preitl, 1999a; Precup, & Hellendoorn, 2011; Precup, Roman, & Safaei,
2021):
(a) The crisp input information, expressed as measured variables, reference input
(or set-point) and control error e , is converted into a fuzzy representation. This
Sugeno (T-S) fuzzy controllers or Sugeno fuzzy controllers. In addition to the control
error, the vector y a of additional variables (shown in Figure 1) can be used as crisp
input information applied to the fuzzy controller; such variables, with an important
(b) The fuzzified information is processed using the rule base, which consists of
fuzzy IF-THEN rules referred to as fuzzy control rules illustrated in (1), that must be
control system. The principles for evaluating and processing the rule base represent the
inference mechanism/engine, and the result is the “fuzzy” form of the fuzzy control
(c) The fuzzy control signal must be converted into a crisp formulation with a
structure of a fuzzy controller shown in Figure 1 (b), namely the fuzzification module
(2), the inference module (4) and the defuzzification module (6). All these modules are
The structure shown in Figure 1 (b) becomes more complex in the context of
uncertainties in terms of type-2 fuzzy sets with additional parameterization of the input
defuzzification module. This leads to interval type-2 fuzzy controllers, with different
applications besides the general ones of type-2 fuzzy logic ones reviewed in (Mittal,
Jain, Vaisla, Castillo, & Kacprzyk, 2020). As outlined in (Wu, 2012; Precup, David,
Roman, Szedlak-Stinean, & Petriu, 2021), the main differences between the interval
• adaptivity, i.e., changing the embedded type-1 fuzzy sets when computing the
• novelty, i.e., the upper and lower membership functions of the same interval
type-2 fuzzy set can be used simultaneously in computing each bound of the
type-reduced interval,
These differences have been exploited and transformed into the main advantage
of interval type-2 fuzzy controllers pointed out in (Lam & Seneviratne, 2008), namely
interval type-2 fuzzy model-based control directly handles the uncertainties in nonlinear
controller is used for direct feedback control or at the low level in hierarchical control
system structures. However, a fuzzy controller can also be used at the supervisory level,
such as in adaptive control system structures. Nowadays, fuzzy control is not only used
to directly express the knowledge about the controlled process, which was originally
called model-free fuzzy control. A fuzzy controller can be designed from a fuzzy model
controller tuning. The most commonly used are (Precup, & Hellendoorn, 2011):
• Mamdani fuzzy controllers, also called linguistic fuzzy controllers, with either
belong to type-II fuzzy systems. These fuzzy controllers are usually used as
direct controllers. These notations of the fuzzy systems type concern their
structure and are different from those of the fuzzy sets type, which concern the
definitions of the fuzzy sets. The notations of the fuzzy sets type are next
(Sugeno, 1999), especially when affine consequents are employed, and typically
Several surveys and position papers highlight specific topics in fuzzy control,
make characterizations, and present valuable points of view, and a part of the most
control is provided in (Babuška, & Verbruggen, 1996). The stability analysis methods
for type-II fuzzy control systems are analysed in detail in (Sugeno, 1999). An overview
(Mitra, & Hayashi, 2000). The fusion of computationally intelligent methods, including
fuzzy logic and sliding mode control, is discussed in (Kaynak, Erbatur & Ertugrul,
Tanaka, 2015) with quasi-linear parameter varying (LPV) model-based control design,
major applications and improvements. A survey on the analysis and design methods of
through some milestones and key applications is carried out in (Feng, 2006). Other
or high-fidelity simulations can be found in (Hoffmann, & Werner, 2014; Li, Nguyen,
Du, Wang, & Zhang, 2021). A survey on type-I, type-II, and type-III fuzzy systems
than type-1 fuzzy controllers, their systematic design is more difficult. Therefore, their
algorithms in the optimal tuning of type-2 fuzzy controllers have been reported in
(Castillo, Melin, Garza, Montiel, & Sepúlveda, 2011; Castillo, & Melin, 2012; Castillo,
& Melin, 2014; Valdez, Castillo, Cortes-Antonio, & Melin, 2020), and this issue is also
discussed in other reviews that deal with both type-1 and type-2 fuzzy controllers,
including (Hagras, 2008; Precup, Angelov, Costa, & Sayed-Mouchaweh, 2015; Hamza,
fuzzy control reported until 2011 (Precup, & Hellendoorn, 2011), nature-inspired
Mouchaweh, 2015), and the status of fuzzy control systems and perspectives in 2019
(Nguyen, Taniguchi, Eciolaza, Campos, Palhares, & Sugeno, 2019), and provides an
However, papers published before 2011 are considered here if they are representative of
the topics discussed. Some of these applications are industrial, while others are
results, but significant results with strong theory and rich simulation studies are also
included. The authors are aware of the fact that the number of publications on the topic
list is impossible. Selected papers are listed in the References section of this paper.
Unfortunately, many excellent papers have been omitted, and we hope that their authors
will accept the apologies of the authors of this paper. In addition, this survey paper is
not able to cover all categories of fuzzy control applications of mechatronic systems,
which may include manufacturing, robotics, automotive and process industries, servo
different control system structures, divided into model-based fuzzy control, data-
driven fuzzy control, and evolving systems. Conventional, state feedback and
This section first reviews some main features and research mainstreams related to fuzzy
the classical T-S fuzzy control systems with linear consequents and the T-S fuzzy
control systems with nonlinear consequents. Then, this section discusses some notable
developed with the Lyapunov stability method and convex optimization techniques
under LMI constraints. Other complementary and related theoretical and application
results of fuzzy model-based techniques with other design tools, for instance, sliding
mode control, adaptive control, predictive control, etc., can be found in (Feng, 2006;
(Sugeno, 1999; Sala, Guerra, & Babuška, 2005; Nguyen, Taniguchi, Eciolaza, Campos,
Palhares, & Sugeno, 2019). The primary reason was the absence of stability analysis for
fuzzy control during that period. In response to these critiques, Sugeno’s research group
Model rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (2)
THEN x (t ) = A i x(t ) + B i u(t ), i = 1...r ,
where x(t ) n is the state vector, u(t ) m is the control input vector; y (t ) q is
the output vector, z(t ) = [ z1 (t ) z 2 (t ) ... z p (t )]T p is the vector of premise variables or
fuzzy set, specifically the linguistic term, and r is the number of fuzzy model rules. The
constant subsystem matrices A i nn and B i nm , for i = 1...r , are known. The
overall fuzzy model of the system is obtained by fuzzy “blending” of the linear
subsystem models as
r
x (t ) = hi (z(t ))[ Ai x(t ) + Biu(t )], (3)
i =1
where
i (z (t )) p
hi (z (t )) = r
, i (z (t )) = M ij ( z j (t )). (4)
(z(t ))
i =1
i
j =1
The term M ij ( z j (t )) represents the grade of membership of z j (t ) in M ij . The
h (z(t )) = 1,
i =1
i
(5)
0 hi (z (t )) 1, i = 1...r.
The discrete-time T-S fuzzy model is constructed in the same way by replacing
x (t ) with x(t + 1) in (2) and (3), with t indicating the continuous time argument and
the discrete time argument, respectively. However, for the sake of simplicity and
illustration, continuous-time T-S fuzzy models are mainly discussed hereafter.
Among various analytical methods to construct a T-S fuzzy model from a given
nonlinear system, the sector nonlinearity approach (Wang, Tanaka, & Griffin, 1996) has
received the most attention since it provides a systematic method to derive a strictly
equivalent T-S fuzzy representation within a compact set of the state space. However,
for existing T-S fuzzy results based on the sector nonlinearity approach, the numerical
exponentially grows with respect to the number of premise variables (Tanaka, & Wang,
2004). This limits the applicability of these results to systems with only few
nonlinearities (Li, Xie, Zhao, Gao, Hu, & Wong, 2021). To overcome this major
practical issue, several approaches have been proposed to reduce the numerical
complexity of T-S fuzzy systems. Based on a singular value decomposition method, the
authors in (Yam, Baranyi, & Yang, 1999) proposed an approach to reduce the fuzzy
rules, yielding approximate T-S fuzzy models. In (Taniguchi, Tanaka, Ohtake, & Wang,
2001), some nonlinearities were transformed into system uncertainties to reduce the
number of local linear sub-models of T-S fuzzy systems. Exploiting the linear
dependencies between the T-S fuzzy local sub-models obtained with the sector
However, the reduction approaches in (Taniguchi, Tanaka, Ohtake, & Wang, 2001;
Dehak, Nguyen, Dequidt, Vermeiren, & Dambrine, 2020) can lead to over-conservative
stability analysis and control design results. To avoid this drawback, a reduced-
representation of a given nonlinear system within a compact set of the state space.
(Dehak, Nguyen, Dequidt, Vermeiren, & Dambrine, 2022) only grows proportionally,
rather than exponentially, with the number of premise variables. In particular, this
approach allows introducing some specific slack variables at the modelling step to
State feedback fuzzy control has been widely applied to stabilize a T-S fuzzy
system. The most popular state feedback fuzzy control scheme is based on the concept
of parallel distributed compensation (PDC) (Tanaka, & Sugeno, 1992; Tanaka, Ikeda, &
Wang, 1996; Wang, Tanaka, & Griffin, 1996). The key idea is to construct the fuzzy
controller sharing the same premise membership functions and the same number of
rules as the T-S fuzzy model (Tanaka, Ikeda, & Wang, 1996). Then, the fuzzy controller
Control rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (6)
THEN u(t ) = K i x(t ), i = 1...r.
r
u(t ) = hi (z(t ))K i x(t ). (7)
i =1
The goal of the fuzzy controller design is to determine the local feedback (matrix) gains
K i , for i = 1...r , in the linear consequent parts such that the closed-loop T-S fuzzy
system with the models (3) and (7) is asymptotically stable and verifies some predefined
performance specifications.
For illustrations, let us consider the stability of the closed-loop system with the
models (3) and (7). Substituting (7) into (3), the state-space equation of the closed-loop
r r
x (t ) = hi (z (t ))h j (z (t ))( A i + B i K j )x(t ). (8)
i =1 j =1
The most basic stabilization result for system (8) is based on the quadratic Lyapunov
V (x) = xT P x, P 0, (9)
stability of the closed-loop system (8), the time derivative of V (x) along its trajectory is
r r
(BMI) due to the coupling between the Lyapunov matrix P and the control gains K j ,
for j = 1...r , which cannot be solved effectively with standard numerical solvers (Boyd,
El Ghaoui, Feron, & Balakrishnan, 1994). To convexify the control design conditions, a
congruence transformation can first be performed by pre- and post-multiplying (10)
r r
With a simple change of variable M j = K j Q , for j = 1...r , the stability condition (11)
becomes
r r
make the condition for checking the closed-loop stability feasible and tractable, the
membership functions should be dropped out from (12) to obtain a finite set of LMI
control design conditions. To this end, numerous relaxation results have been proposed
to drop out the membership functions with different degrees of conservativeness and
numerical complexities, which can be categorized into two classes: i) without requiring
slack variables such as Tanaka’s relaxation (Tanaka, & Wang, 2004), Tuan’s relaxation
(Tuan, Apkarian, Narikiyo, & Yamamoto, 2001); ii) involving slack variables such as
Liu and Zhang’s relaxation (Liu, & Zhang, 2003), Pólya-based relaxation (Sala &
Ariño, 2007; Montagner, Oliveira, & Peres, 2009). In particular, Pólya-based relaxation
results offer asymptotically necessary and sufficient LMI-based conditions for the
stability and performance of T-S fuzzy systems (Sala, & Ariño, 2007). Nevertheless,
approaches based on Pólya's theorem are more conceptual than feasible, as the
computational load escalates rapidly, often leading to crashes in most numerical solvers
(Nguyen, Taniguchi, Eciolaza, Campos, Palhares, & Sugeno, 2019). Although the PDC
concept provides a natural, effective and systematic framework for T-S fuzzy controller
and observer designs using LMI-based techniques, the obtained results can be very
conservative (Lam, 2018). Therefore, most of the research effort has been focussed on
reducing the design conservatism of T-S fuzzy model-based approaches. Apart from
finding an effective way to drop the membership functions from the design conditions
as discussed above, there are two main directions to reduce the conservativeness of the
design conditions (Nguyen, Taniguchi, Eciolaza, Campos, Palhares, & Sugeno, 2019):
i) using different families of Lyapunov function candidates, ii) exploit better the
Lyapunov function, where a single Lyapunov matrix must be used for stability analysis
of all local sub-models of the T-S fuzzy systems. To relax this constraint, more general
classes of Lyapunov candidate functions have been leveraged for T-S fuzzy model-
based approaches, for instance, piecewise Lyapunov functions (Johansson, Rantzer, &
Arzen, 1999; Feng, 2004), line integral Lyapunov functions (Rhee, & Won, 2006;
Mozelli, Palhares, & Avellar, 2009), fuzzy Lyapunov functions depending on the
membership functions (Tanaka, Hori, & Wang, 2003; Guerra, & Vermeiren, 2004;
Zheng, Xie, Nguyen, & Qu, 2023), polynomial Lyapunov functions depending on the
membership functions with an arbitrary degree (Zhang, & Xie, 2011), multidimensional
fuzzy Lyapunov functions (Lee, Joo, & Tak, 2014), etc. In particular, together with the
use of fuzzy Lyapunov functions and slack decision variables, the non-PDC control
concept has been proposed to further reduce the design conservatism of T-S fuzzy
approaches (Guerra, & Vermeiren, 2004; Xie, Ma, Zhao, Ding, & Wang, 2012; Zheng,
Xie, Nguyen, & Qu, 2023). More insightful discussions on the advantages and
drawbacks of each type of Lyapunov candidate functions can be found in (Nguyen,
Second, the membership functions are used to interconnect the local sub-models
of the T-S fuzzy system, which represent the nonlinearities of the original plant to be
controlled. However, only their convex sum property (5) has been exploited in most of
the existing works based on various types of Lyapunov candidate functions. As far as
the shape characteristics of the membership functions, i.e., the intrinsic nonlinear nature
of the plant, are not exploited for stability analysis, the concerns related to the
conservativeness issue still remain (Lam, 2018; Sala, & Arino, 2008). To reduce the
staircase and/or piecewise linear membership functions (Lam, & Narimani, 2009;
functions for stability analysis (Sala, & Arino, 2008; Li, Xie, Zhao, Gao, Hu, & Wong,
convex stability constraints and a high number of involved decision variables, thus a
high degree of computational complexity. Then, the resulting T-S fuzzy control and
estimation results can be impractical, especially for complex processes with a high
number of premise variables (Dehak, Nguyen, Dequidt, Vermeiren, & Dambrine, 2022).
Research effort has been also devoted to reducing the numerical design complexity.
Two notable approaches can be distinguished. First, the numerical complexity reduction
of the T-S fuzzy results can be done by “controlling” the number of slack decision
variables introduced in the control design for relaxation purposes (Sala, & Arino, 2008;
Montagner, Oliveira, & Peres, 2009; Xie, Lu, & Yue, 2022a). Second, researchers have
tried to find alternative T-S fuzzy modelling methods to reduce the excessive number of
fuzzy rules caused by the classical sector nonlinearity approach as discussed above.
Apart from the previously mentioned modelling reduction methods, T-S fuzzy
especially N-TS fuzzy systems (Dong, Wang, & Yang, 2009) and fuzzy polynomial
N-TS fuzzy systems. Using similar techniques as for the classical T-S fuzzy
modelling, i.e., the sector nonlinearity approach, a N-TS fuzzy model can be derived
Model rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (13)
THEN x (t ) = A i x(t ) + B i u(t ) + G i φ(t ), i = 1...r ,
where the same system notations as in (2) are used. The vector of nonlinearities
where the matrix E = [E1T ... ETh ]T nh is given, and Λ hh is a positive definite
diagonal matrix. Note that for engineering applications, since the states of the systems
always possible (Khalil, 2002). Since some nonlinearities, i.e., premise variables, of the
original complex plant can be retained in the consequent parts with φ (t ) of the T-S
fuzzy models, this N-TS fuzzy modelling generally requires less fuzzy rules, i.e., local
sub-models, than the classical T-S fuzzy modelling with linear consequents (Takagi, &
Sugeno, 1985). Using the PDC concept, a fuzzy state feedback controller of the N-TS
system (13) is given in the following form, where the nonlinearity φ (t ) is incorporated
Control rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (15)
THEN u(t ) = K xi x(t ) + K i φ(t ), i = 1...r.
where the control gains K xi mn and K i mh , for i = 1...r , are to be determined.
It is clear that using the same fuzzy inference method, the N-TS fuzzy model (13) and
r
x (t ) = hi (z(t ))[ A i x(t ) + B i u(t ) + G i φ(t )] (16)
i =1
and
r
u(t ) = hi (z(t ))[K xi x(t ) + K i φ(t )]. (17)
i =1
i = 1...r , then the N-TS fuzzy system (13) (respectively the fuzzy controller (15))
reduces to the classical T-S fuzzy system (2) (respectively the fuzzy controller (6)),
widely studied in the fuzzy control literature. Therefore, N-TS fuzzy models encompass
the classical T-S fuzzy models as a special case. Then, one can follow the same
approach as Chapter 14 in (Tanaka, & Wang, 2004) to demonstrate that N-TS fuzzy
models can be used as a universal approximator for smooth nonlinear control systems.
functions as the case of the classical T-S fuzzy systems, N-TS fuzzy-model-based
control approaches have been actively developed for nonlinear control of complex
systems (Dong, Wang, & Yang, 2010; Coutinho, Araujo, Nguyen, & Palhares, 2020;
Araujo, Coutinho, Nguyen, & Palhares, 2021; Xie, Yang, Wan, Xia, & Shi, 2022). In
particular, an N-TS fuzzy static output feedback controller was developed in (Nguyen,
Coutinho, Guerra, Palhares, & Pan, 2021) for nonlinear systems subject to state and
reduce the control design conservatism. As illustrated in these references, N-TS fuzzy
model-based control approaches can offer less conservative design results and less
using N-TS fuzzy modelling, an effective solution has been recently proposed for the
challenging T-S fuzzy observer design with unmeasured premise variables, where all
the unmeasurable premise variables of the original plant are isolated in the nonlinear
consequents (Pan, Nguyen, Guerra, & Ichalal, 2020; Nguyen, Pan, Guerra, & Wang,
2020; Quadros, Leite, & Palhares, 2022; Peixoto, Nguyen, Guerra, & Palhares, 2023).
Model rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (18)
THEN x (t ) = A i (x(t ))xˆ (x(t )) + B i (x(t ))u(t ), i = 1...r ,
where A i (x(t )) nn and B i (x(t )) nm are polynomial matrices in x(t ) , and
xˆ (x(t )) n is a column vector whose entries are all monomials in x(t ) . Following a
similar defuzzification procedure as the case of the classical T-S fuzzy system, the
Inspired by the PDC control concept, the following polynomial fuzzy controller can be
Control rule i :
IF z1 (t ) IS M i1 AND ... AND z p (t ) IS M ip (20)
THEN u(t ) = K i (x(t ))xˆ (x(t )), i = 1...r ,
r
u(t ) = hi (z(t ))K i (x(t ))xˆ (x(t )). (21)
i =1
The closed-loop polynomial fuzzy system can be formed from (19) and (21) as
r r
x (t ) = hi (z (t ))h j (z (t ))[ A i (x(t )) + B i (x(t ))K j (x(t ))]xˆ (x(t )). (22)
i =1 j =1
For stability analysis and control design of the fuzzy model (19), the following
where P(x) nn is a symmetric positive definite polynomial matrix. The control
problem related to the polynomial fuzzy system (22) is to determine the feedback gains
K j (x(t )) , for j = 1...r , and the polynomial Lyapunov matrix P (x) such that x(t ) → 0 ,
when t → , with the standard assumption that xˆ (x(t )) = 0 if and only if x(t ) = 0 . The
control design conditions are recast in the form of the sum of squares (SOS) constraints,
which can be effectively solved using the well-known Matlab SOSTOOLS toolbox
xˆ (x(t )) = x(t ) , and A i (x(t )) , B i (x(t )) , K i (x(t )) and P (x) are constant matrices, for
i = 1...r , then the polynomial fuzzy model (19) reduces to the T-S fuzzy model (3), and
the polynomial Lyapunov function (23) is the same as the classical quadratic Lyapunov
function. Hence, the SOS-based approaches to polynomial fuzzy models encompass the
expected that polynomial fuzzy model-based approaches using SOS tools can provide
significantly more relaxed stability results than the classical LMI-based T-S fuzzy
approaches (Tanaka, Yoshida, Ohtake, & Wang, 2008; Sala, & Arino, 2009).
Similar to classical T-S fuzzy model-based approaches, the main research issues
to be considered when investigating the stability of the polynomial fuzzy system (22)
are related to the types of Lyapunov candidate functions used for stability analysis, the
analysis conditions, and the convexification techniques used for stability analysis and
control design. These research issues are directly related to the design conservatism, the
systems can be found in the recent survey (Lam, 2018). It is important to note that
despite their appealing theoretical interests, there are only very few successful
Lam, & Althoefer (2016) presented a notable kinematic polynomial fuzzy model-based
in the current literature are essentially illustrated with simple academic examples and
simulation results.
have been largely applied to the control and estimation problems of engineering
of T-S fuzzy control techniques for aerospace and marine applications, robotics and
hereafter.
Aerospace and marine applications. A T-S fuzzy integral sliding mode control
method was proposed in (Hu, Wu, Hu, & Gao, 2013) for the control problem of flexible
The authors in (Sun, Xu, & Li, 2017b) dealt with the T-S fuzzy model-based finite-time
controller was developed under the presence of stochastic actuator failures and sampled-
data control inputs. The authors in (Aslam, Tiwari, Pandey, & Band, 2023) investigated
an H∞ attitude control method for sustainable hypersonic vehicles with Markovian jump
parameters, whose dynamics were approximated using Taylor’s expansion and T-S
fuzzy linearization approaches. A coupled Lyapunov function was used to derive the
conditions for stochastic T-S fuzzy controller design while guaranteeing a predefined
H∞ performance level. A fuzzy adaptive sliding mode control method for attitude
tracking control of a flexible spacecraft in (Li, Liu, & Shi, 2020), where the authors
scheme was developed in (Jiang, Gao, Shi, & Xu, 2010) for a near-space vehicle,
represented by a T-S fuzzy model. Note that the proposed fault-tolerant tracking control
method does not depend on any fault detection and isolation mechanism, making it
distributed attitude control for spacecraft formation flying systems was investigated in
(Zhang, Zhang, & Zhang, 2017), where T-S fuzzy modelling was used to represent the
In (Li, Dai, Song, Wang, & Du, 2019), the authors presented a T-S fuzzy fault-tolerant
attitude tracking control method for Mars entry vehicles subject to disturbances and
actuator failures, where a fuzzy line-integral Lyapunov function was used to guarantee
the closed-loop stability with satisfactory disturbance attenuation level under actuator
controller was proposed in (Ma, Nie, Yu, Hu, & Peng, 2020) for underactuated
Fei, & Peng, 2018a), where the asynchronous difference of the normalized membership
functions between the marine vehicle and its controller was explicitly considered in the
control design. In (Zhang, Ye, Feng, & Li, 2021), the authors developed an event-based
T-S fuzzy dynamic positioning control method for unmanned marine vehicles under
learning algorithm to reduce the communication loads, and the observer and controller
design conditions were derived via a piecewise Lyapunov function to guarantee the
functions, was proposed in (Hao, Zhang, Li, Lin, & Chen, 2021). The thruster faults
were dealt with using a switching mechanism, while the time-varying delay effects were
fuzzy event-triggered sliding mode control was also proposed in (Zhang, Yao, Xing, &
Feng, 2022) for unmanned underwater vehicles under multiple practical constraints,
practical sliding mode and the stability of the sliding motion while avoiding the Zeno’s
effect in the proposed event-triggering mechanism. The authors in (Cheng, Zhang, Xie,
Zhang, & He, 2023) designed an adaptive T-S fuzzy dynamic positioning control of
mechanism was proposed to dynamically adjust the event-triggered thresholds for data
transmission improvements. T-S fuzzy modelling was considered in (Shi, Sun, & Hou,
external disturbances and actuator faults. Then, a disturbance observer and an extended
state observer were developed to estimate external disturbances, unmeasured states and
actuator faults for sampled-data dynamic positioning output feedback controller design.
Another sampled-data fuzzy controller was designed in (Kim, Lee, & Joo, 2021) for a
decentralized T-S fuzzy model. The LMI-based sampled-data tracking control design
conditions were derived using a time-dependent Lyapunov-Krasovskii functional and a
reference model. A fault-tolerant T-S fuzzy controller was developed in (Li, Xu, & Yu,
2022) for underwater vehicles, where the event-triggering mechanism was designed
using multiple past sampled data to decide the next release instant. Based on the
canonical Bessel-Legendre inequality, the proof for the asymptotic stability of the T-S
Due to the high complexities of aerospace and marine systems, most of the
Ning, Xing, & Li, 2016) developed a state-observer-based T-S fuzzy control method for
Experimental tests were performed with a quarter-car test rig under different road
consider simultaneously the varying sprung and unsprung masses, the unknown actuator
nonlinearity, and the suspension performances in the control design, an adaptive sliding-
mode controller was developed in (Li, Yu, Hilton, & Liu, 2012), whose design
conditions are expressed in the form of an optimization problem. Other H∞ T-S fuzzy
controllers were also proposed for active suspension systems with random actuator
delay in (Han, Zhong, Chen, & Tang, 2019), and in the presence of sprung mass
variation while reducing the motor wear in (Shao, Naghdy, & Du, 2017). A disturbance-
observer-based T-S fuzzy controller was proposed in (Ning, Sun, Zhang, Du, Li, &
Zhang, 2017) for an active seat suspension. The disturbance observer was used to
compensate the disturbances caused by frictions, and modelling errors. Moreover, the
T-S fuzzy technique was leveraged to improve the control performance by considering
different drivers’ weights in the control design. Extensive simulations and experiments
were carried out to validate the effectiveness of the proposed T-S fuzzy controller. A
switched T-S fuzzy method was proposed in (Qing, Hongliang, Songlin, Weiwei, &
Yongfeng, 2023) for continuous damping control of semi-active suspensions, where the
asymmetry saturation of the control current was explicitly considered in the PDC-based
state feedback controller design. Extensive simulation and experimental results were
improvements of the proposed fuzzy control method in terms of ride comfort and road
handling. The authors in (Jeong, & Choi, 2021) proposed a magnetorheological damper
fault diagnosis algorithm using only two accelerometers, which are commonly available
for commercial vehicles. To this end, a T-S fuzzy unknown input observer was
developed for the estimation of the vehicle suspension system subject to damper
algorithm was used to generate a fault flag, which not only minimizes design efforts but
quarter-car test rig. A fuzzy reduced-order observer was developed in (Zhu, & Li, 2019)
the T-S fuzzy modelling was used to deal with the system nonlinearity caused by the air
drag torque. As shown by the simulation results, the transient estimation performance
can be enhanced with the pole placement technique, while the robustness with respect to
the road condition variations was guaranteed with a robust H∞ filtering approach. A T-S
fuzzy unknown input observer was proposed in (Li, Liu, & Shi, 2020) to estimate the
transmission input-output shaft torque and the drive wheel speed for a hybrid
another unknown input T-S fuzzy observer scheme, the authors in (Losero, Lauber, &
Guerra, 2018) developed a virtual strain gauge, where the observer design was
performed in the angular domain. This fuzzy observer scheme was applied to estimate
the engine torque and the clutch torque via the angular deflection of a dual-mass
flywheel, which is crucial for powertrain management in the automotive industry. T-S
fuzzy control techniques have been successfully applied to various robust path tracking
(Nguyen, Sentouh, & Popieul, 2018; Guo, Wang, Luo, & Li, 2020; Nguyen, Sentouh,
Zhang, & Popieul, 2019b; Liang, Feng, Lu, Yin, Zhuang, & Mao, 2023). In particular,
scheme was recently investigated in (Zhang, Hu, Zhang, Bian, Nguyen, & Ding, 2023)
for path-tracking control of autonomous vehicles. The vehicle lateral dynamics were
described using T-S fuzzy framework and the uncertainties of the cornering stiffness
approach. The proposed event-triggered control method was validated with a high-
expensive on-board vehicle sensors for control purposes, T-S fuzzy output feedback
control with or without using observer structures has been investigated while
considering the actuator saturation constraints (Hu, Chen, & Wang, 2020; Nguyen,
Rath, Guerra, Palhares, & Zhang, 2020) or the effects of friction forces (Alcalá, Puig, &
Quevedo, 2019). T-S fuzzy model-based observers have been proposed for the
estimation of vehicle states and various related unknown variables. For instance, based
on a N-TS fuzzy representation of the vehicle nonlinear dynamics, fuzzy unknown input
observers were developed and real-time validated in (Nguyen, Campos, Guerra, Pan, &
Xie, 2021; Nguyen, Dinh, Guerra & Pan, 2021) to cope with nonlinear descriptor
systems and simultaneously estimate the lateral speed, steering input, and effective
engine torque, which are key components influencing vehicle handling, stability control,
and fault diagnosis in autonomous ground vehicles (Pan, Nguyen, Guerra, Sentouh,
Wang, & Popieul, 2022). To further improve the estimation performance in the case
where the vehicle system is subject to modelling uncertainties and unknown inputs, a
Nguyen, & Delprat, 2023). The effectiveness of the proposed neural-network-based T-S
fuzzy observer was experimentally validated with an autonomous vehicle on a real test
track. To deal with the time-varying driver characteristics while considering the human-
machine interaction involved in the driving process, T-S fuzzy state feedback shared
controllers were developed in (Nguyen, Sentouh, & Popieul, 2016; Benloucif, Nguyen,
Sentouh, & Popieul, 2019; Fang, Wang, Wang, Liang, Liu & Yin, 2023) for driver-
automation cooperative path tracking control. To avoid the use of costly vehicle
sensors, fuzzy output feedback shared controllers have been also investigated for co-
driving control with and without requiring an observer, see for instance (Li, Xie, Zhao,
Gao, Hu, & Wong, 2021) and (Nguyen, Sentouh, & Popieul, 2017; Ding, Shan, Han,
Jiang, Peng, & Liu, 2022), respectively. The driver-in-the-loop test results, obtained
with these fuzzy human-machine cooperative control methods, showed that the path
tracking performance, the vehicle stability and the driving comfort are significantly
improved while the potential driver-automation conflicts can be mitigated when the
controller designs. Apart from automotive and intelligent vehicles applications, T-S
fuzzy control techniques were also successfully explored in railways engineering, e.g.,
for high-speed train control with experimental tests (Tasiu, Liu, Yan, Chen, Hu, & Wu,
2020; Tasiu, Wang, Liu, Zhang, Zhang, Meng, & Zhao, 2023).
networked nonlinear systems is a hot topic. As stated in (Precup, Preitl, Petriu, Bojan-
Dragos, Szedlak-Stinean, Roman, & Hedrea, 2020), the presence of time delay in both
the control signal and the sensor measurement transmission in networked control
systems creates challenging control problems. One such problem is the transcendental
characteristic equation of the control system, which is not simple even if linear
controlled processes and controllers are assumed, and numerical problems have to be
exemplified with telesurgical applications (Precup, Haidegger, Preitl, Benyó, Paul, &
Kovács, 2012), event-triggered path tracking control considering roll stability under
Lauber, Boada, & Boada, 2023), fuzzy model-based nonlinear networked control
systems (Qiu, Gao, & Ding, 2016), and event-triggered control with unreliable
communication links (Li, Chen, Wu, & Lam, 2017). The up-to-date literature on this
topic includes the representative results dealing with the distributed PI state estimation
problem for nonlinear systems over sensor networks (Wang, Wang, Zou, Chen, & Yue,
2023), T-S fuzzy systems under FlexRay communication protocol (Wang, Wang, Zou,
Ma, & Dong, 2023), PID-fuzzy control of nonlinear systems subjected to controller
parameter disturbances over mixed fading channels (Wang, Wang, Zou, & Dong, 2022),
PDC concept, a T-S fuzzy controller was proposed in (Guechi, Lauber, Dambrine,
measurements. A nonlinear predictor observer was designed to estimate the robot state
variables while accounting for the constant time-delayed outputs. Both simulation and
experimental tests were performed to illustrate the effectiveness of the proposed T-S
fuzzy tracking-error model-based approach. Sun et al. developed a switched T-S fuzzy
control method for wheeled mobile robots in (Sun, Chen, Wang, & Huang, 2017a),
where the visual odometry was leveraged for robot localization purposes. Moreover, the
ensured using Lyapunov stability theory as illustrated with suitable experimental tests.
Using a T-S fuzzy descriptor control approach, PDC-based controllers were investigated
Vermeiren, & Dambrine, 2019) for the tracking control of planar parallel robots and
serial robots, respectively, where parametric uncertainties were considered in the fuzzy
investigated the fuzzy model-based control problem for flexible joint robots with
improve the control performance of the proposed PDC-based state feedback controller,
the Hybrid-Taguchi genetic algorithm was employed to select the control parameters. A
T-S fuzzy model-based controller was developed in (Wen, Hu, Lv, Wang, & Peng,
2019) for the trajectory tracking of a humanoid robot NAO manipulator, where a Q-
learning reinforcement learning algorithm was used for robot trajectory planning with
the possibility of obstacle avoidance. Both simulation and experimental results were
presented to illustrate the tracking control effectiveness of the proposed T-S fuzzy
proportional-integral (PI) observer was proposed to real-time estimate the joint torques
and angular velocities in human stance from angular positions. By experimental results,
the authors showed that the proposed T-S fuzzy observer can outperform the well-
known inverse dynamics joint torques estimation method in terms of robustness with
Gabrielli, 2018; Guerra, Blandeau, Nguyen, Srihi, & Dequidt, 2020) to study the sitting
control strategies of persons living with spinal cord injury (SCI). Despite some
thoroughly validate the proposed model-based approaches for these challenging SCI
control the knee joint angle movement of paraplegic patients through electrical
Sanches, De Carvalho, Biazeto, & Teixeira, 2020), where a simple method was used for
system discretization and then T-S fuzzy modelling. Experiments with a paraplegic
volunteer and a healthy person showed that predefined design specifications (stability,
decay rate, and input constraints) can be achieved under small sampling periods. T-S
including two-wheel inverted pendulum systems (Huang, Wang & Chiu, 2010),
overhead crane systems (Aguiar, Leite, Pereira, Andonovski, & Škrjanc, 2021), variable
speed wind turbines (Bououden, Chadli, Filali, & El Hajjaji, 2012; Schulte, & Gauterin,
2015), and hydraulic turbines (Ma, & Wang, 2021; Tian, Wang, Chen, & Yang, 2021).
design conservatism due to the classical norm-bounded approach (Tanaka, & Wang,
2004) complex nonlinear systems, is based on the interval type-2 fuzzy modelling
(Lam, & Seneviratne, 2008). Different from the classical T-S fuzzy system, whose
grades of membership are fixed, those of the interval type-2 fuzzy system may vary
within a certain range to capture the system uncertainties. Considerable research efforts
have been devoted to developing a systematic control framework for interval type-2
fuzzy systems (Lam, 2018). However, despite their theoretical interests, the application
techniques with experimental validations can be mentioned such as the robust control of
a bolt-tightening tool mounted on a robot arm (Lam, Li, Deters, Secco, Wurdemann, &
Althoefer, 2013), a planar snake robot (Bhandari, Raj, Pathak, & Yang, 2022), and a
vehicle sensor fault estimation application (Liu, Chen, Na, Luo, & Zhang, 2020).
directly from data without identifying a system (or process) model. For this reason,
controller tuning. In addition, Li, Yuan, Li, & Zhu (2022) consider that model-free
control achieves model-free operation such that the dynamics and the stability
performance of a control system are robust to process model variations. This paper uses
the term “data-driven” instead of the term “model-free” because there are data-driven or
closed-loop control system models expressed in the time domain or the frequency
domain. Nevertheless, the book (Precup, Roman, & Safaei, 2021) considers both “data-
this book, a useful discussion on model-based versus data-driven control, which has
inspired and continues to inspire future research directions, is provided in (Hou, &
their model-based design discussed in the previous section, in order to benefit from the
advantages of data-driven control and fuzzy control and, if possible, to mitigate their
fuzzy controllers was first proposed and applied in (Preitl, Precup, Fodor, & Bede,
2006; Preitl, Precup, Preitl, Vaivoda, Kilyeni, & Tar, 2007; Precup, Preitl, Rudas,
Tomescu, & Tar, 2008) and continued in (Roman, Precup & David, 2018; Roman,
Precup, Bojan-Dragos, & Szedlak-Stinean, 2019; Precup, Roman, & Safaei, 2021;
Roman, Precup & Petriu, 2021; Precup, Preitl, Bojan-Dragos, Hedrea, Roman, & Petriu,
2022), mainly dealing with hybrid data-driven and fuzzy controllers through structures
that combine data-driven control and fuzzy control to incorporate model-free features in
fuzzy control system structures. The direct data-driven fuzzy control ensures the data-
these controllers in fuzzy control. Both indirect and direct data-driven fuzzy control
techniques are discussed. A classical indirect data-driven fuzzy control technique will
optimization problems where the variables are represented by the parameters of the
fuzzy controllers. These optimization problems are then solved in various one-shot or
iterative formulations, which are briefly discussed in the context of the data-driven
control techniques discussed below. The robustness and stability guarantees of data-
driven fuzzy systems are challenging. These will be discussed in Section 5, and the
information provided in (Precup, Roman, & Safaei, 2021) can be used for the stability
control techniques. For example, the structure of the control system is considered in
(Hou, & Zhou, 2013), which leads to two categories. In the first category, the controller
structure with one or more unknown parameters is assumed to contain the optimal
structure; the controller design is then transformed into a direct identification problem
process, such as neural networks, fuzzy models, or Taylor series approximation; the
(which plays the role of performance criterion) using the input-output data, including
both offline and online data. This section will discuss, as carried out in (Precup, Roman,
& Safaei, 2021), the techniques associated with these controllers, divided into two
categories that aim their implementation, namely iterative ones and non-iterative or one-
shot ones. The most successful data-driven control techniques considered in (Precup,
Roman, & Safaei, 2021) will be briefly discussed below, supporting them with classical
along the gradient direction of an objective function. IFT is applicable when an initial
function is assumed to be known (Preitl, Precup, Preitl, Vaivoda, Kilyeni, & Tar, 2007;
Jung, Jeon, Kang, & Oh, 2021). IFT is applied in (Precup, Preitl, Rudas, Tomescu, &
Tar, 2008; Precup, Radac, Tomescu, Petriu, & Preitl, 2013) in the indirect data-driven
fuzzy control approach for tuning PI fuzzy controllers for servo systems with
experimental validation. IFT is also applied in (Precup, Tomescu, Radac, Petriu, Preitl,
& Dragos, 2012) in the same indirect approach to the tuning of state feedback fuzzy
output data collected from the process (Hou, & Huang, 1997; Yu, Wang, Bu, & Hou,
2020). The MFAC structures are based on local dynamic linearized models of the
process, and the control algorithms are formulated in a similar manner to model-based
predictive control. Three types of dynamic linearization data models are included in
MFAC structures: Compacted Form Dynamic Linearization (CFDL) (Hou, & Jin,
2011a; Hou & Jin, 2011b), Partial Form Dynamic Linearization (PFDL) (Hou, & Jin,
2011b) and Full Form Dynamic Linearization (FFDL) (Hou, 1999). A fuzzy logic-based
adjustment strategy is utilized in (Wang, Yang, Liang, Liu, Wang, & He, 2018b) to
select either MFAC or data-driven Iterative Learning Control (ILC) and appropriate
spray fluidized bed granulation process with simulated validation. MFAC is modelled
as a fuzzy relational model in (Kadri, & Hussain, 2010) and applied to cooling coil
control with simulated validation. The performance of MFAC with CFDL is improved
Hedrea, Bojan-Dragos, & Radac, 2019) by replacing the linear PD component of the
control law with a PD-fuzzy component with experimental validation on tower crane
systems. Hybrid model-free adaptive fuzzy controllers with tower crane systems
According to (Preitl, Precup, Preitl, Vaivoda, Kilyeni, & Tar, 2007), Iterative
Learning Control (ILC) ensures that the performance of control systems that repeatedly
perform the same tasks can be improved using the experience gained from previous
for two reasons. First, ILC can be formulated as the iterative solution of a parametric
optimization problem that ensures the minimization of an objective function to meet the
performance indices imposed on the control system; this can be viewed in the general
context of learning. Second, since ILC generally does not act on the controller
parameter but on the control system structure outside the controller, it can be applied to
beneficial effects on the control system behaviour with respect to both reference and
disturbance inputs.
ILC has been treated recently in the framework of data-driven control, and a
thorough analysis of both theory and applications is conducted in (Chi, Hui, & Hou,
2022). Other relevant results in this regard are optimal data-driven ILC (Chi, Hou,
Huang, & Jin, 2015), constrained data-driven optimal ILC (Radac, Precup, & Petriu,
2015; Chi, Liu, Zhang, Hou, & Huang, 2018), MIMO ILC (Bolder, Kleinendorst, &
Oomen, 2018), data-driven terminal ILC (Bu, Zhu, Hou, & Liang, 2020), and the
fuzzy ILC algorithm with learning capabilities offered by the fuzzy logic part is
designed in (Yu, & Hou, 2021) and applied to the control of high-speed trains with
simulated validation. A fuzzy model is included in (Liu, Illian, Leonhardt, & Misgeld,
2023) in the ILC algorithm for rehabilitation exoskeletons with compliant joints with
applied in (Yan, Guo, Zhang, Yan, & Liu, 2020) to air-conditioning control systems
embedded in (He, Li, Li, Liu, & Wu, 2022) in an adaptive robust ILC scheme that also
includes a certain neural network architecture, and the control system structure is
validated by digital simulation results. Three ILC schemes are applied in (Precup, Preitl,
Tar, Tomescu, Takács, Korondi, & Baranyi, 2008) in the indirect data-driven fuzzy
control approach to the tuning of PI-fuzzy controllers for servo systems with
experimental validation.
a data-driven and machine learning (ML) technique whose specific feature is the use of
solve the optimization problem that ensures optimal reference tracking. An RL agent
performs actions in the environment and adjusts its knowledge about itself and the
environment based on the reward it receives. This process is applied iteratively and
incrementally, so that the RL agent gets better and better at choosing actions that
maximize or minimize rewards (Sutton, & Barto, 2017). As pointed out in (Sutton,
Barto, & Williams, 1992), RL is a viable technique that solves optimal reference
tracking problems and bridges the gap between ML and control. In this context, the RL
agent is the controller that automatically learns how to modify its parameters and how
to control a process based on the feedback (i.e., reward) it receives from it (Busoniu, de
Bruin, Tolić, Kober, & Palunko, 2018). A neural fuzzy controller is designed in (Guo,
Lama, Jiang, & Zhou, 2014) to combine the strengths of fast online learning and self-
adaptation of neural networks and fuzzy control for automated server parameter tuning
is proposed in (Zhang, Zhang, Cai, & Han, 2019) to combine an RL technique and a
fuzzy-enhanced model, and applied to a single-link robotic arm system and pitch-rate
combined with fuzzy logic in (Chen, Hu, Tang, & Cheng, 2022) to address autonomous
driving with experimental validation. The inputs of a Q-learning scheme are fuzzified in
(Yin, & Li, 2022) to ensure generation control of power systems with simulated
Zhang, Mu, & Sun, 2019) and applied to a single-link robotic arm system with
(Tooranjipour, & Vatankhah, 2018) to reduce the number of rules using the quaternion
backpropagation concept and using RL to evaluate the output value produced by the
critical neural network; this controller is applied to chaotic spinning disk control with
simulated validation.
Model-Free Control (MFC) (Fliess, & Join, 2009; 2013) combines the popular
and widely used PI and PID controllers, as well as Proportional (P) controllers, with an
intelligent term that compensates for the effects of nonlinear dynamics, disturbances,
PI, and intelligent PID controllers. An integral local model of the process is used and its
controllers with tower crane systems is reported in (Precup, Roman, & Safaei, 2021).
Merah, Hartani, Yazid, & Chikouche (2022) combined a hybrid fuzzy system to control
the suspension damper with an intelligent PID-fuzzy controller to manage the in-wheel
system aimed at jointly improving ride comfort and road holding. To enhance passenger
comfort and vehicle stability regardless of road conditions, fuzzy logic was applied in
system of a large van. A type-1 fuzzy attitude controller was developed to mitigate loop
interactions and to address limitations in optimizing control gains between heave and
pitch with roll motions. In the inner loop, a Mamdani interval type-2 fuzzy logic
controller was used for ride control to accommodate system uncertainties and
nonlinearities.
the system model with an additional and fictitious state variable that models the
unmodeled dynamics of the controlled process. This virtual state is estimated online by
an extended state observer and then used in the control signal to decouple the system
from the actual disturbance acting on the controlled process. Fuzzy control replaces PID
control in (Feng, Zhang, Gao, & Li, 2021) in ADRC applied to hypersonic aircraft with
simulated validation. Fuzzy control adjusts online the parameters of the state error
nonlinear control law in (Shen, Xu, Chen, & Xia, 2023) in ADRC applied to an
unmanned helicopter with simulated validation. Fuzzy logic is used in (Li, Sung, Guo,
& Liu, 2022) to smoothly correlate ADRC and MFAC in the control of two-degrees-of-
ADRC is improved in (Roman, Precup, & Petriu, 2021) by replacing the linear PD
component of the control law with a PD-fuzzy component with experimental validation
the identified fuzzy model of the controlled process is proposed in (Dombi, & Hussain,
2019) and applied to vehicle lateral dynamics with simulated validation. A direct data-
driven fuzzy controller based on interval type-2 distending function and the identified
fuzzy model of the controlled process is proposed in (Dombi, & Hussain, 2023) and
proposed in (Yuan, Nguyen, & Zhou, 2021) for semi-active air suspensions of heavy
trucks, where the fuzzy control rules are optimized by the genetic algorithm. Based on
the optimal fuzzy control scheme and the data map of the random road surfaces, an
adaptive network-based fuzzy inference system was developed to enhance the ride
driven technique, which aims to minimize objective functions J that are specific to
N
1
J (χ ) = E{{[Ly (q −1 )y(k , χ )]2 + [ Lu (q −1 )u (k , χ )]2 }}, (24)
2 N k =1
where χ is the parameter vector of the controller, L y (q −1 ) and Lu (q −1 ) are weighting
filters that penalize the output error (or the tracking error) y and the control input (or
the control signal) u to give importance to certain frequency regions, q −1 is the unit
delay operator, referred to also as the backward shift operator, is the control signal
weighting parameter, k is the discrete time index, and N is the number of samples or
the length of the experiment (the trial). The mathematical expectation E{} is taken
with respect to the stochastic probability distribution of the disturbance inputs applied to
the process and thus affects the control system behaviour. The disturbance inputs are
assumed to be zero mean discrete-time stochastic processes, and it is also assumed that
The expression of the vector variable of the objective function in (24), which is
where l , l = 1...n , are the controller tuning parameters. The expression J (χ ) of the
Lu (q −1 ) and the weighting parameter , such that the minimization of the objective
function will ensure the fulfilment of the performance specifications imposed on the
control system. The expression of the output error (or the tracking error) y ( k , χ ) is
where y ( k , χ ) is the controlled output and y d (k ) is the output of the reference model,
i.e., the desired output of the control system to be tracked by the controlled output.
The objective of IFT is to compute the optimal parameter vector χ * as the
which minimizes the objective function J expressed in (2). The expression of the
where *l , l = 1...n , are the optimal tuning parameters of the controller.
As pointed out in (Precup, Roman, & Safaei, 2021), the major hint in solving the
computation of the gradient of the objective function with respect to the controller
parameters. The specific feature of IFT is that an estimate of the gradient of the
gradient-based search algorithms as, for example, the Gauss-Newton scheme expressed
as the following IFT algorithm, which represents the parameter update law
J
χ [i +1] = χ [i ] − [i ] (R[i ] ) −1 est (χ [i ] ), (29)
χ
J
certain scalar or vector or matrix at iteration i , the vector est (χ [i ] ) is the estimated
χ
gradient, the parameter [i ] , [i ] 0 , is the step size of the current iteration, and R [i ] is
The Single Input-Single Output (SISO) control system structure with the IFT
objective function in (2), RM – the reference model, r – the reference input or the set-
e = r − y, (30)
u – the control signal or the control input, and d – the disturbance input. As illustrated
addition, all signals in the control system are assumed to be differentiable with respect
to χ .
Figure 2. SISO control system structure with IFT algorithm (Precup, Roman, & Safaei,
2021).
function in (24) is formulated and its parameters are set such that the minimization of
the objective function by solving the optimization problem defined in (27) ensures the
parameters specific to the parameter update law given in (29) are set. An initial
using the reference input r{1} = r applied to the control system, where the notation { j} ,
j {1,2} , is used for the subscript that is inserted to certain variables in order to specify
the number of the experiment conducted with the control system. The normal
experiment is conducted in terms of the control system structure given in Figure 2. The
control signal u{1} , the controlled output y{1} and the control error e{1} are measured.
conducted using the reference input r{2} = e{1} applied to the control system in terms of
the control system structure given in Figure 2. The control signal u{2} and the controlled
y u
Step IFT4. The estimated expressions of (k , χ ) and (k , χ ) , namely
χ χ
y u
est (k , χ ) and est (k , χ ) , respectively, are computed using (Precup, Roman,
χ χ
y 1 C −1
est (k , χ ) = −1
(q , χ ) y{2} (k , χ ), (31)
χ C (q , χ ) χ
u 1 C −1
est (k , χ ) = −1
(q , χ ) u{2} (k , χ ), (32)
χ C (q , χ ) χ
C −1
in (25). The gradient of the controller transfer function, (q , χ ) , is a column matrix
χ
expressed as
C −1 C −1 C −1 C −1
(q , χ ) = [ (q , χ ) (q , χ ) ... (q , χ )]T , (33)
χ 1 2 n
where all derivatives are assumed to exist. The two estimates in (31) and (32) are next
J
objective function, i.e., est (χ ) as in (Precup, Roman, & Safaei, 2021):
χ
J 1 N y
est (χ ) = {L y (q −1 )y (k , χ )est (k , χ )
χ N k =1 χ
(34)
u
+ L y (q −1 )u (k , χ )est (k , χ )}.
χ
The values of y ( k , χ ) and u (k , χ ) are taken from the normal experiment and next used
in (34).
Step IFT5. The next parameter vector χ [ i +1] is computed in terms of applying the
Step IFT1 is carried out once. Steps IFT2 to IFT5 are repeated for a predefined
number of iterations set in step IFT1. The above steps clearly show that the controller
parameter tuning is carried out without using the process model in terms of executing
two experiments per iteration and additional computation, thus ensuring the iterative
The steps IFT1 to IFT5 are applied to linear controllers. However, in the indirect
IFT approach applied to fuzzy controllers, the linear controllers are replaced by fuzzy
controllers are next briefly discussed because they can be extended relatively easily to
versions of PI-fuzzy controllers are next considered (Precup, & Preitl, 1999a): the PI-
fuzzy controller with integration of controller output (PI-FC-OI), and the Proportional-
As shown in Figure 3, the two PI-fuzzy controllers are built around the Two
without dynamics. In addition, the input variables are also scheduling variables. The
dynamics are inserted in PI-FC-OI by the numerical differentiation of the control error
e(k ) , leading to the increment of control error e(k ) = e(k ) − e(k − 1) , and the numerical
that the nonlinear scaling factors of the input and output variables specific to TISO-FC
As specified in (Precup, & David, 2016; Precup, & David, 2022), the
fuzzification in TISO-FC that belongs to PI-FC-OI is done in terms of the input (and
also scheduling) and output membership functions illustrated in Figure 4 for Mamdani
terms of the same membership functions as those specific to PI-FC-OI, but the input
variable e I (k ) is used instead of e(k ) , and the output variable u (k ) is used instead of
fuzzy controllers make use of only the input membership functions illustrated in Figure
4.
Figure 4. Input and output membership functions of Mamdani PI-fuzzy controllers with
integration on controller output (Precup, & David, 2016; Precup, & David, 2022).
2022):
and the centre of gravity method for singletons is used in the defuzzification module of
PI-fuzzy controllers are based on the SUM and PROD operators assisted by the
following rule base, exemplified for PI-FC-OI (Precup, & David, 2016; Precup, &
David, 2022):
Using the PI-fuzzy controller structures descried above, the rule bases given in
(35) and (36) make these controllers behave as bumpless interpolators between
K P1 = K P2 = K P4 = K P5 = K P6 = K P8 = K P9 = K P , K P3 = K P7 = K P ,
(37)
1 = 2 = 3 = 4 = 5 = 6 = 7 = 8 = 9 = .
Step FC1. The continuous-time design and tuning of the linear PI controller with
k c (1 + T i s) 1
C ( s) = = k C (1 + ), k C = k c Ti (38)
s Ti s
is carried out, and it leads to the controller gain k c (or k C depending on the expression
Step FC2. The sampling period T s is set according to the requirements of quasi-
continuous digital control. Tustin’s method is next applied to discretize the continuous-
time linear PI controller, and the recurrent equation of the incremental digital PI
controller is
where the expressions of the parameters K P and that appear in (37) and (39) are
the overshoot of the fuzzy control system when both inputs have the same sign. These
PI-fuzzy controllers can also be applied to the control of non-minimum phase systems
with right half-plane zeros, where such rule bases produce the alleviation of the
downshoot as well.
Step FC3. The modal equivalence principle (Galichet & Foulloy, 1995) is
applied to map the linear controller parameters onto the PI-fuzzy controller ones. The
Be = Be , (41)
and the application to the Mamdani PI-FC-OI leads to the tuning conditions (Precup, &
David, 2016)
The tuning conditions for the Mamdani PI-FC-II are (Precup, & Preitl, 1999b)
1
Be = Be , Bu = K P Be , (43)
and the tuning condition for the Takagi-Sugeno PI-FC-II is (Precup, & David, 2016)
1
Be = Be . (44)
As specified in (Precup, & David, 2016; Precup, & David, 2022), the value of
the parameter Be in (41) – (44) must be set by the designer. This can be carried out
according to the designer’s experience, and the stability analysis of the fuzzy control
system can be conducted in this regard in terms of the results presented in the previous
section, but the data-driven feature of the fuzzy controller could be lost. The systematic
tuning of the parameter Be can be done in terms of defining and solving optimization
problems, where this parameter is one of the elements of the vector variables of the
objective functions.
Step FCIFT1. The steps IFT1 to IFT5 of the IFT technique are applied to the
linear PI controllers.
Step FCIFT2. The parameters of the linear controllers are mapped on to the
This approach has been applied successfully in (Precup, Preitl, Rudas, Tomescu,
& Tar, 2008; Precup, Radac, Preitl, Petriu, & Dragos, 2009) to the IFT-based
the application of the step FCIFT1 are given in (Roman, Precup, Hedrea, Preitl,
Zamfirache, Bojan-Dragos, & Petriu, 2022). Although this indirect approach affects the
this context clearly show the performance improvement of the control system as a direct
Fuzzy control systems have proven countless times that they can effectively control
complex nonlinear systems, as shown in Sections 2 and 3. The field has matured over
time and has established links with several related research areas. The introduction of
the Takagi-Sugeno fuzzy model, which has similarities with linear systems, has
systems. The adaptive approach was one of the first to be adopted. Adaptive fuzzy
control offers several advantages because it is known for its adaptability, allowing it to
adjust parameters in real time as system conditions change. This adaptability leads to
operation in the presence of uncertainties. Additionally, it reduces the need for manual
However, Figure 1 (b) shows that there are several points in the control loop
or, equivalently, into the parameters of the fuzzy rule consequents. This is
functions. This approach led to adaptive fuzzy control, for which there exist
many direct and indirect schemes in the literature (Wang, 1993; Yin, & Lee,
1995; Blažič, Škrjanc, & Matko, 2003). It soon became clear that the full
• The next step towards evolving systems is to introduce the ability to adjust
can vary across different regions of the input space. Consequently, a more
represent input data is similar to clustering techniques, which group data points
based on similarity, revealing inherent clusters, and the latter has often been
refer to such a system as an evolving intelligent control (EIC) system. Like any
learn from data streams, but the specific task of an EIC is to adapt and evolve
the control strategy accordingly. The main difference with adaptive control is
that not only the parameters but also the structure of the control systems are
information. In particular, this refers to the case where the number of rules in the
inference module adapts over time. New rules may be added, obsolete rules may
be removed, and existing rules may be combined or split. Not all EICs have all
corresponding outputs. One of the key steps in the implementation of the evolving fuzzy
defuzzification, and the inference system. As a result, the approximation of the EIS
output is expressed as a function of the inputs and the initially unknown parameters.
The learning capabilities of the EIS are introduced through adaptive laws that govern
gradual parameter updates and a mechanism that allows changes in the structure by
adjusting the number of fuzzy rules. If properly designed, the evolving nature of the EIS
allows it to continuously refine its input-output mapping to the desired one with an
generally straightforward, as these parameters are often linear with respect to the output.
Established adaptive laws from adaptive fuzzy systems derived by gradient descent of
the cost function or using Lyapunov stability theory can be applied. This typically
evolving algorithms, with various options available for calculating the degree of
fulfilment of fuzzy rules. In the field of evolving fuzzy systems, these approaches can
current data to the cluster centres (closer data points have higher
position. In the case of clustering techniques, the parameters that define the position and
fuzzy model structure, requiring a significant improvement in the cost function before
introducing changes. Fuzzy rules can also be deleted. This is typically done when a rule
contradicts other rules in its vicinity. In addition, rules may be deleted if they have had a
consistently low degree of fulfilment in the past, and/or if they have been created and
updated with “old” data. Some evolving fuzzy systems also provide mechanisms for
merging and splitting rules. However, it is important to note that similar results can
often be achieved by using adding and pruning mechanisms alone. Because there is so
much freedom in designing an EIS, almost every paper proposes some variation of an
and performance over time. Feedback, a key concept in control theory that makes
control so powerful, can induce system instability if not appropriately designed. The
within linear control systems. This can lead to significant stability challenges when
seemingly small deviations from the initial assumptions occur. The initial excitement
about the practical applications of adaptive control waned after the tragic accident of the
X-15-3 aircraft in 1967. Early indications of the vulnerabilities of adaptive control led
Lyapunov functions for stability analysis. Robust adaptive control aims to develop
control strategies and techniques that can effectively deal with uncertainties, unmodeled
performance. Many ideas borrowed from the theory of robust adaptive control have
been introduced into adaptive fuzzy control to ensure system stability for a certain class
of nonlinear systems.
stability analysis increases. In the context of EICs, stability analysis becomes more
complex due to potential changes in control structure and strategies. Another important
aspect of EICs is the quality of available training data, which is essential to ensure
that have undergone at least some validation may be preferable to introducing poor
models obtained during transients, where the quality of the data may be questionable.
contexts where input/output data are accessible during learning (Angelov, & Buswell,
2001; Kasabov, & Song, 2002) with rare early papers dealing with control (Angelov,
2004). The evolving approach has been successfully used in several applications where
stability is not the primary concern, such as prediction and forecasting, fault detection
and diagnosis, data mining, pattern recognition, and classification, and can be extended
to other types of fuzzy models, such as type-2 fuzzy models (Juang, & Tsao, 2008).
In control, however, the fundamental presence of a control loop requires that the
control design remains based on stability considerations. Looking at this problem from a
during learning. However, the control must be applied to the actuators at each sampling
time, and a poor choice of control at successive sampling times will result in degraded
performance. Therefore, the use of EIS for control purposes is inherently more
Various approaches are used to address this issue, but most often the control is
built around a relatively simple and robust control structure that does most of the work
to achieve stabilization, while EISs are used to reduce modelling errors. In the
following, the approaches in the literature are classified according to this fixed-structure
Some early papers dealing with EIC did not really solve the fundamental problem of
simultaneous control and controller evolving, but somehow simplified it. For instance,
in (Juang, & Lin, 1998), the structure of the evolving neural fuzzy inference network
was initially pre-trained using data collected during open-loop operation. Afterwards,
complete adaptation, with or without evolving mechanisms, was initiated to guide the
system toward the control objective. A similarly conservative approach was taken in
(Cara, Pomares, Rojas, Lendek, & Babuška, 2010), where the data are first kept in the
membership function and on which variable. The modification of the structure of the
evolving controller is therefore based on the analysis of the error surface and the
determination of the input variables that are most responsible for the error. The
historical data. However, some systems that incorporate evolutionary algorithms as part
example can be found in (Lee, & Hallam, 1999) where the controller structure evolves
in real time based on a growing tree of logical operators and the fitness function
assigned to individual controllers. The approach has been successfully tested on a real
Khepera miniature robot performing various tasks such as obstacle avoidance, box
A major problem with many evolving controllers is that the antecedent part of
the rule, especially when the membership functions are added, is completely
uninterpretable. This often leads to overlapping of the rules. In (Cara, Herrera, Pomares,
& Rojas, 2013), this problem is addressed by using a scatter partitioning of the input
space. Together with a singleton in the consequent of the fuzzy controller rules, this
leads to a nice interpretable surface of the controller mapping. The structure of the
controller is thus much more controlled and sudden loss of stability is unlikely, although
Perhaps the most common approach to solving the EIC design problem is to integrate an
evolving fuzzy system to emulate the “ideal controller” and a sliding mode controller
(SMC) to ensure system stability. Being stability-oriented, this method is based on the
to the control error states e and a term that is quadratic with respect to the parameter
estimation errors (deviations from the "ideal" parameterization of the evolving fuzzy
control system). The task of control design is to find adaptive laws for parameter
estimates and a control law that together ensure that the derivative of the Lyapunov
function is negative (at least everywhere except in the region where the errors are
relatively small). The problem is that the effect of modelling errors (fuzzy modelling
the derivative of the Lyapunov function becomes less than a negative quadratic form of
e , ensuring that the error states e converge to 0 along the so-called sliding surface.
The drawback of this approach is that the control signal exhibits undesirable chattering.
All the methods discussed in this section guarantee closed-loop stability because the
In (Chang, 2010) the combination of adaptive fuzzy control and sliding mode
control was proposed where the number of fuzzy rules is kept constant. The approach
actuators. In an early implementation of EIC (Park, Park, Kim, & Moon, 2005), the
approach follows similar lines to other papers dealing with adaptive fuzzy control, i.e.,
the algorithms are designed within the Lyapunov stability framework. However, unlike
the other approaches at the time, this paper introduced the mechanism of creating new
triangular membership functions, and thus new fuzzy rules, as needed. The stability is
achieved by adding the control term sgn( k T e) , where k is a constant vector and is
determined by the adaptive law. In the case of (Lu, Chang, & Tsai, 2011), the following
where e(t ) is the control error while the constants k i , i = 1...n , are chosen so that all
roots of the polynomial s n + k n s n−1 + ... + k1 are located in the open left half-plane.
Similarly as before, the additive term takes the form of sgn( s (t )) where s (t ) is
defined by (45). This is a very common solution that is used in many works discussed in
this section. Gaussian membership functions are utilized in (Lu, Chang, & Tsai, 2011),
and fuzzy rules can be either added or deleted from the rule base. The approach was
system.
In (Lin & Li, 2013), the evolving fuzzy system is used to replicate an ideal
control that results in perfect tracking of the reference model. Gaussian membership
functions with adaptive parameters are used, and a method for generating and
eliminating fuzzy rules is introduced. In the SMC term, soft switching is used
(implemented with the tanh function). The approach has been evaluated on a simulated
performance under different road conditions and transitions between them. In (Hsu &
Wong, 2016), the evolving fuzzy system is again used to tune the primary controller,
while the additive SMC term includes the sign function. The number of fuzzy rules can
control algorithm proposed in (Lin, Ramarao, & Gopalai, 2019) again gradually
constructs the main controller by adding and pruning the rules and adapting their
parameters, while an additive term contains another three-rule fuzzy system that serves
as a compensator to ensure the robust stability. The approach was successfully tested on
two complex simulated mechatronic systems, namely a double inverted pendulum
The approach of (Han, Zhou, Qiao, & Feng, 2015) uses a linear controller to
stabilize the linearized dynamics, the EIC with changing structure is used to account for
the nonlinearity, while the third term again includes the sign function of the linear
The method proposed in (Ferdaus, Pratama, Anavatti, & Garratt, 2019; Ferdaus,
Pratama, Anavatti, Garratt, & Pan, 2020) introduces an EIS with the addition and
acting as an auxiliary robustifying control. The parameters of the sliding surface also
language to facilitate easy transition to dedicated hardware. While testing has been
limited to simulated autonomous vehicles like bio-inspired flapping wing micro air
vehicles, quadcopters, and hexacopters, future plans involve experiments with real
flying devices.
Anavatti, & Garratt, 2020), where they also introduce a dedicated database to store the
record of deleted nodes for future retrieval. The algorithm is used to control the angular
position of the NXT DC motor. A similar approach is used in (Huynh, Lin, Le, Le, Vu,
& Chao, 2022), where a self-organising structure can automatically add or prune the
layers to achieve an efficient network structure, while adapting the parameters of the
simulated nine-link biped robot and a real magnetic levitation device. Another version
of the EIC supported by an SMC is proposed in (Ngo, Hoang, Tran, Nguyen, Nguyen,
& Le, 2022), where they also show the results of experimental validation on the
Quanser 2-DOF robot. In (Wang, & Fei, 2022), they also propose the combination of
the EIC and an SMC but the latter is a fractional-order SMC. As a result, the size of the
chattering is smaller than that of an ordinary integer-order SMC. The approach was
The approach in (Le, Lin, & Huynh, 2018) uses type-2 fuzzy functions. Similar
to the above approaches, an evolving fuzzy system with an adaptive number of fuzzy
rules plays the main role in the control. The additive term with the sign function again
helps to achieve stability. It is worth noting that the learning rates for the adaptive laws
are optimized using the PSO algorithm. A type-2 evolving fuzzy controller has also
been proposed in (Al-Mahturi, Santoso, Garratt, & Anavatti, 2023). It also uses an
evolving system to approximate the ideal controller, while robustness is ensured by the
term using the saturation function applied to s (t ) defined by (45). The algorithm was
introducing the mechanisms to adjust the number of rules. In the case of a direct
adaptive fuzzy controller with fuzzy blending of individual control rules designed in the
Lyapunov stability framework, the extension with more fuzzy rules does not introduce
changes in the control law. Consequently, the stability of the whole system is not
compromised if the new rules are based on a sufficient amount of data. This statement is
an oversimplification of a very complex system and can only serve as a basic idea. A
In an early implementation of this approach (Phan, & Gale, 2008), the direct
number is limited by the user. Adaptive laws use parameter projection to prevent
parameter drift. The stability of the control system was proven and tests on an inverted
The EIC proposed in (Blažič, Škrjanc, & Matko, 2014) can be seen as an
extension of the direct fuzzy model reference adaptive control proposed in (Blažič,
Škrjanc, & Matko, 2003). It uses the same control law and the same adaptive law for the
control gains. The evolving part introduced in (Blažič, Dovžan, & Škrjanc, 2014) uses
the eFuMo method based on the Gustafson-Kessel clustering algorithm. This EIS is the
mechanisms for adding new rules. The stability of the proposed EIC has been
demonstrated. In (Dovžan, Blažič, & Škrjanc, 2014), new evolving mechanisms were
introduced, allowing not only the addition of new rules but also their deletion, splitting
into two, or merging of two rules. The method was applied to the control of a detailed
simulated model of the helio-crane laboratory pilot device (Blažič, & Zdešar, 2017).
Research papers dealing with EIC often aim to simplify the core control stability
feedback loop. Meanwhile, the second controller, which evolves over time, deals with
The above approach was used in (Gao, & Er, 2003). This work applies to a
certain class of MIMO systems. The evolving part is based on a fuzzy network with
Gaussian membership functions and allows the addition and deletion of rules. The
stability of the system is proven in the paper. The approach has been used to control
If it is looked carefully at the control law in (Gao, & Er, 2003), it becomes clear
that the inputs to both the linear (constant gain) auxiliary controller and the evolving
controller consist of error signals and their derivatives. Consequently, both controllers
can be considered as PD-like controllers. It will be shown that nonlinear PID control is
In (Chen, & Lin, 2011), a linear PD controller was used in the feedback, while
the evolving interval type-2 neural fuzzy network was used in the feedforward part.
Thus, the stability problem of the EIC was not an issue because the PD controller was
chosen by the designer. The proposed approach was used for high-precision motion
a stabilizing controller is also proposed in (Ngo, & Phuong, 2015). However, in their
applied in the feedback. The stability of the proposed solution is shown, and the control
proposed for control. The wavelet base functions are used at the input of the network.
The evolving algorithm can generate additional hidden neurons. Stability is proven in
the paper. The proposed control is applied to a DC motor. The control part of the
t
s(t ) = e(t ) + k 2 e(t ) + k1 e()d. (46)
0
The sliding surface s (t ) can also be interpreted as the output of the PID
controller. The signal s (t ) is fed into the supervisory compensator, which is simply an
adaptive gain. The signal s (t ) also serves as the only input to the evolving part, which
shows that the total control input is the linear combination of the three components of
A very similar approach from the control law perspective which can also be
interpreted as a nonlinear PID, is proposed in (El-Sousy, 2014), while the evolving part
control computer and experimentally tested on the torque control of the rotor position
H2/H∞ controller, an evolving recurrent fuzzy wavelet neural network controller, and a
robust controller (again a PD). It is important to note, however, that the vector entering
all three parts is the same, namely a filtered tracking error vector. The paper includes
stability proof. The proposed approach has been implemented on a two-axis motion
control system (X-Y table) driven by two permanent-magnet linear synchronous motor
servo drives.
The following approaches use fuzzy rules with PID control actions in the rule
consequents. When these rules are combined, a nonlinear PID is again obtained. The
main question is how to define the control objective. This influences the design of the
introduce the reference model, i.e., the output of the plant should follow the reference
signal filtered through a reference model (typically a first-order linear system). In other
words, the evolving PID attempts to linearize the control system. However, in most
cases, the linearization is only approximate due to the limited degrees of freedom of the
adaptive fuzzy control, where each fuzzy rule consists of a PID controller with adaptive
control gains in the consequent. The structure of the controller evolves over time based
on the distances in the input space where so-called data clouds are defined. The
approach also introduces several techniques from robust adaptive control (dead zone in
the adaptation law, parameter projection, leakage in the adaptation law, interruption of
the data and changes in the adaptation laws of the consequent parameters (Andonovski,
Blažič, Angelov, & Škrjanc, 2015; Andonovski, Angelov, Blažič, & Škrjanc, 2016), the
approach was also implemented on the plate heat exchanger device. A similar approach
where structure evolving is based on data clouds is used in (Andonovski, Mušič, Blažič,
& Škrjanc, 2016), where a new mechanism for removing the "less active" and "less
outliers, or at least to help remove existing clouds with little information. Another
approach where evolving is based on data clouds is presented in (Angelov, Škrjanc, &
Blažič, 2013). The approach also features PID controllers in the rule consequents, but
new rules are added not only regarding the local density, but also considering the global
density, which indicates how representative a measurement is with respect to the entire
data distribution. The approach has been successfully applied to the laboratory pilot
plant of two tanks (Costa, Škrjanc, Blažič, & Angelov, 2013). In (Chen, & Liu, 2017),
feature is that the membership functions are asymmetric (different “standard deviations”
are used on either side of a Gaussian function). The algorithm includes the addition and
pruning of fuzzy rules. From a control point of view, the controller is a PD with
adaptive and nonlinear gains. The approach was implemented for position servo control
of a permanent magnet linear synchronous motor servo drive system with two position
Another option in the feedback loop, when combined with an EIC, is model predictive
control (MPC). It has been applied in (Zdešar, Dovžan, & Škrjanc, 2014), where it
fuzzy model. This approach does not require explicit optimization of the cost function,
on the eFuMo model, which has the mechanisms for adding, deleting, merging and
pilot system.
In (Han, Zhang, Hou, & Qiao, 2016), a nonlinear MPC is developed based on an
evolving recurrent radial basis function neural network. The optimization problem is
formulated as minimizing the cost function over the horizon, which is achieved by
this paper. The following classification of control systems was proposed and used for
this purpose: (i) model-based fuzzy control, (ii) data-driven fuzzy control, and (iii)
evolving fuzzy control. The applications highlighted in this paper cover several areas
beginning in 2011. However, papers published before 2011 have also been considered if
the authors believe they are representative of the topics discussed. There are several
challenges that deserve more study when considering fuzzy control from the point of
been directed towards mitigating the conservatism in stability analysis and control
design. As previously discussed, this has been achieved by exploring more versatile
reformulate the stability analysis and control design problems within an LMI
functions during the design procedure. In spite of recent advances, these two research
directions still have the potential to yield highly impactful results in the field of fuzzy
T-S fuzzy control and estimation are primarily limited to academic demonstrations, with
no notable industrial applications. This is primarily due to the fact that the numerical
complexity of the related stability analysis, observation and control design conditions
exponentially grows with respect to the number of premise variables (Tanaka, & Wang,
2004). Hence, T-S fuzzy model-based results are mostly suitable for systems with only
a few nonlinearities, which is generally not the case in complex industrial plants. As
previously discussed, several approaches have been developed to overcome this major
reduce the dimensionality of the premise variables, for instance, principal component
analysis (Kwiatkowski, & Werner, 2008) and deep neural networks (Koelewijn, & Tóth,
2020). However, these modelling approaches also suffer from significant drawbacks
strongly believe that only a T-S fuzzy control/estimation framework with a low level of
numerical complexity, reasonable design conservatism, and simple tuning methods can
facilitate its practical application. The quest for such a T-S fuzzy framework remains an
open-ended question for the fuzzy control community, and it will undoubtedly inspire a
analysis to mitigate the design conservatism associated with T-S fuzzy model-based
approaches. This is possible because the information from the triangular membership
analysis (Sugeno, & Taniguchi, 2004). Moreover, Mamdani-like fuzzy models are
particularly useful within industrial contexts to study complex plants with no acceptable
However, despite some recent advances to achieve convex LMI-based stability analysis
conditions for continuous-time case (Nguyen, Sugeno, Campos, & Dambrine, 2016) as
well as discrete-time counterpart (Nguyen, Dehak, Guerra, & Sugeno, 2023), Mamdani-
like fuzzy model-based approaches still remain computationally heavy, especially when
the number of piecewise regions becomes large. Moreover, the existing control design
(Nguyen, Taniguchi, Eciolaza, Campos, Palhares, & Sugeno, 2019), which poses
additional challenges for real-time applications. Therefore, identifying pertinent
research directions aimed at reducing the computational load and establishing convex
control design conditions for Mamdani-like fuzzy models would constitute significant
standpoints.
As stated in (Precup, Roman, & Safaei, 2021), the main shortcoming of data-
driven control is the difficulty of systematic stability and robustness analyses. In other
words, tuning to ensure tracking of the reference trajectory does not guarantee robust
stability and robust performance. This is normal, since these analyses require detailed
tuning process here is model-free, the use of the term “robustness” sounds ambiguous
and should be clarified. The stability analysis of a data-driven fuzzy control system is
challenging and the information given in (Precup, Roman, & Safaei, 2021) can be used
in this regard.
There is still a lot of room to develop various direct data-driven fuzzy controllers
following three data-driven control techniques that have the potential to be merged with
fuzzy control in direct and indirect approaches are presented next, the rest of them are
1998; Wang, & Spall, 1998) is supported by the fact that, unlike deterministic
they must be estimated from the noisy measurements of the objective function
using finite difference approximations around the current operating point. SPSA
Mišković, Karimi, Bonvin, & Gevers, 2007) operates in the model reference
control framework. The relationship between the reference input and the
tracking error is highlighted in the correlation function of the two signals when
The objective function depends on the correlation function of the two signals
(i.e., the reference input and the tracking error) and is minimized in an iterative
fashion. The initial controller in iterative CbT is usually linear and appropriately
Bruyne, & Bitmead, 1999; Kammer, Bitmead, & Bartlett, 2000) makes use of a
Linear-Quadratic (LQ) objective function that penalizes the tracking error in the
expressed in the frequency domain using Parseval’s theorem and next spectral
models. Thus, the derivatives of the objective function with respect to the
Júnior, & Rezende Barros, 2018). Stability between two consecutive iterations is
(2-DOF) fuzzy controllers, which have been proposed in (Precup & Preitl, 1999b;
Precup, & Preitl, 2003) as fuzzy controllers with non-homogeneous dynamics with
respect to the input channels, and then applied to servo systems and electric drives
(Precup, Preitl, Petriu, Tar, Tomescu, & Pozna, 2009; Preitl, Stinean, Precup, Preitl,
Petriu, Dragos, & Radac, 2012). These structures can provide very good performance in
structure fuzzy control systems is a challenging task, which becomes even more
difficult when the evolving nature of the controller is considered. Considering that the
controller evolving depends on the streaming data taken from different sources in the
complex control system, we realise that this process should not be left to run without
that lead to instability of adaptive control (Ioannou, & Kokotovic, 1984). The
proposed solutions to improve the robustness of the system, such as the use of a
dead zone, parameter projection, leakage, etc., should also be tailored to the
systems have moved away from the initial concept of early fuzzy control
systems, where linguistic rules were translated into fuzzy logic rules.
that are obviously flawed and should not be there, but which produce
typically created from short sequences of bad data or data taken during
continuously learn and adapt over long periods of time. This includes
These challenges will attract both researchers and practitioners. The authors are
convinced that the number of successful mechatronics applications of fuzzy control will
continue to grow.
Acknowledgements. This work was supported by the grant of the Romanian Ministry
of Education and Research, CNCS - UEFISCDI, project number PN-III-P4-ID-PCE-
2020-0269, within PNCDI III. Anh-Tu Nguyen was supported in part by the French
Ministry of Higher Education and Research, in part by the National Center for Scientific
Research (CNRS); in part by the European Community, the Délégation Régionale à la
Recherche et à la Technologie; in part by the Hauts-de-France region under the project
RITMEA. This work has been done within the framework of the CoCoVeIA project
(ANR-19-CE22-0009-01) and the HM-Science project (ANR-21-CE48-0021), funded
by the Agence Nationale de la Recherche. The financial support from the Slovenian
Research Agency (Research Program P2-0219) is also acknowledged.
Data sharing is not applicable to this article as no new data were created or analyzed in
this study.
References
Aguiar, C., Leite, D., Pereira, D., Andonovski, G., & Škrjanc, I. (2021). Nonlinear
modelling and robust LMI fuzzy control of overhead crane systems. Journal of
The Franklin Institute, 358 (2), 1376–1402.
Alcalá, E., Puig, V., & Quevedo, J. (2019). TS-MPC for autonomous vehicles including
a TS-MHE-UIO estimator. IEEE Transactions on Vehicular Technology, 68 (7),
6403–6413.
Al-Mahasneh, A. J., Anavatti, S. G., & Garratt, M. A. (2020). Self-evolving neural
control for a class of nonlinear discrete-time dynamic systems with unknown
dynamics and unknown disturbances. IEEE Transactions on Industrial
Informatics, 16 (10), 6518–6529.
Al-Mahturi, A., Santoso, F., Garratt, M. A., & Anavatti, S. G. (2023). A novel evolving
type-2 fuzzy system for controlling a mobile robot under large uncertainties.
Robotics, 12 (2), paper 23.
Andonovski, G., Angelov, P., Blažič, S., & Škrjanc, I. (2016). A practical
implementation of robust evolving cloud-based controller with normalized data
space for heat-exchanger plant. Applied Soft Computing, 48, 29–38.
Andonovski, G., Blažič, S., Angelov, P., & Škrjanc, I. (2015). Robust Evolving Cloud-
based Controller in normalized data space for heat-exchanger plant. Proceedings
of 2015 IEEE International Conference on Fuzzy Systems, Istanbul, Turkey, 1–
7.
Andonovski, G., Mušič, G., Blažič, S., & Škrjanc, I. (2016). On-line Evolving Cloud-
based Model Identification for Production Control. IFAC-PapersOnLine, 49 (5),
79–84.
Angelov, P. (2004). A fuzzy controller with evolving structure. Information Sciences,
161 (1–2), 21–35.
Angelov, P., & Buswell, R. (2001). Evolving rule-based models: A tool for intelligent
adaptation. Proceedings of Joint 9th IFSA World Congress and 20th NAFIPS
International Conference, Vancouver, BC, Canada, 2, 1062–1067.
Angelov, P., Škrjanc, I., & Blažič, S. (2013). Robust evolving cloud-based controller for
a hydraulic plant. Proceedings of 2013 IEEE Conference on Evolving and
Adaptive Intelligent Systems, Singapore, 1–8.
Araujo, R. F., Coutinho, P. H., Nguyen, A.-T., & Palhares, R. M. (2021). Delayed
nonquadratic L2-stabilization of continuous-time nonlinear Takagi–Sugeno
fuzzy models. Information Sciences, 563, 59–69.
Aslam, M. S., Tiwari, P., Pandey, H. M., & Band, S. S. (2023). Robust stability analysis
for class of Takagi-Sugeno fuzzy with stochastic process for sustainable
hypersonic vehicles. Information Sciences, 641, paper 119044.
Babuška, R., & Verbruggen, H. B. (1996). An overview on fuzzy modelling for control.
Control Engineering Practice, 4 (11), 1593–1606.
Benloucif, A., Nguyen, A.-T., Sentouh, C., & Popieul, J. C. (2019). Cooperative
trajectory planning for haptic shared control between driver and automation in
highway driving. IEEE Transactions on Industrial Electronics, 66 (12), 9846–
9857.
Bhandari, G., Raj, R., Pathak, P. M., & Yang, J. M. (2022). Robust control of a planar
snake robot based on interval type-2 Takagi–Sugeno fuzzy control using genetic
algorithm. Engineering Applications of Artificial Intelligence, 116, paper
105437.
Blandeau, M., Estrada-Manzo, V., Guerra, T. M., Pudlo, P., & Gabrielli, F. (2018).
Fuzzy unknown input observer for understanding sitting control of persons
living with spinal cord injury. Engineering Applications of Artificial
Intelligence, 67, 381–389.
Blažič, S., Dovžan, D., & Škrjanc, I. (2014). Robust evolving fuzzy adaptive control
with input-domain clustering. IFAC Proceedings Volumes, 47 (3), 5387–5392.
Blažič, S., Škrjanc, I., & Matko, D. (2003). Globally stable direct fuzzy model reference
adaptive control. Fuzzy Sets and Systems, 139, 3–33.
Blažič, S., Škrjanc, I., & Matko, D. (2014). A robust fuzzy adaptive law for evolving
control systems. Evolving Systems, 5 (1), 3–10.
Blažič, S., & Zdešar, A. (2017). An implementation of an evolving fuzzy controller.
Proceedings of 2017 Conference on Evolving and Adaptive Intelligent Systems,
Ljubljana, Slovenia, 1–7.
Bolder, J., Kleinendorst, S., & Oomen, T. (2018). Data-driven multivariable ILC:
enhanced performance by eliminating L and Q filters. International Journal of
Adaptive Control and Signal Processing, 28 (12), 3728–3751.
Bououden, S., Chadli, M., Filali, S., & El Hajjaji, A. (2012). Fuzzy model based
multivariable predictive control of a variable speed wind turbine: LMI approach.
Renewable Energy, 37 (1), 434–439.
Boyd, S., El Ghaoui, L., Feron, E., & Balakrishnan, V. (1994). Linear Matrix
Inequalities in System and Control Theory. Society for Industrial and Applied
Mathematics, Studies in Applied Mathematics Series, vol. 15, Philadelphia, PA,
1994.
Bu, X.-H., Zhu, P.-P., Hou, Z.-S., & Liang, J.-Q. (2020). Finite-time consensus for
linear multi-agent systems using data-driven terminal ILC. IEEE Transactions
on Circuits and Systems II: Express Briefs, 67 (10), 2029–2033.
Busoniu, L., de Bruin, T., Tolić, D., Kober, J., & Palunko, I. (2018). Reinforcement
learning for control: performance, stability, and deep approximators. Annual
Reviews in Control, 46 (2), 8–28.
Cara, A. B., Herrera, L. J., Pomares, H., & Rojas, I. (2013). New online self-evolving
neuro fuzzy controller based on the TaSe-NF model. Information Sciences, 220,
226–243.
Cara, A. B., Pomares, H., Rojas, I., Lendek, Z., & Babuška, R. (2010). Online self-
evolving fuzzy controller with global learning capabilities. Evolving Systems, 1
(4), 225–239.
Castillo, O., & Melin, P. (2012). A review on the design and optimization of interval
type-2 fuzzy controllers. Applied Soft Computing, 12 (4), 267–1278.
Castillo, O., & Melin, P. (2014). A review on interval type-2 fuzzy logic applications in
intelligent control. Information Sciences, 279, 615–631.
Castillo, O., Melin, P., Garza, A. A., Montiel, O., & Sepúlveda, R. (2011). Optimization
of interval type-2 fuzzy logic controllers using evolutionary algorithms. Soft
Computing, 15 (6), 1145–1160.
Chang, M. K. (2010). An adaptive self-organizing fuzzy sliding mode controller for a 2-
DOF rehabilitation robot actuated by pneumatic muscle actuators. Control
Engineering Practice, 18 (1), 13–22.
Chen, C. S., & Lin, W. C. (2011). Self-adaptive interval type-2 neural fuzzy network
control for PMLSM drives. Expert Systems with Applications, 38 (12), 14679–
14689.
Chen, L., Hu, X.-M., Tang, B., & Cheng, Y. (2022). Conditional DQN-based motion
planning with fuzzy logic for autonomous driving. IEEE Transactions on
Intelligent Transportation Systems, 23 (4), 2966–2977.
Cheng, P., Zhang, C., Xie, W., Zhang, W., & He, S. (2023). Network-based adaptive
multi-event-triggered fuzzy dynamic positioning controller design for unmanned
surface vehicles against denial-of-service attacks. IEEE Transactions on Control
of Network Systems, 10 (2), 612–624.
Chen, S. Y., & Liu, T. S. (2017). Intelligent tracking control of a PMLSM using self-
evolving probabilistic fuzzy neural network. IET Electric Power Applications,
11 (6), 1043–1054.
Chi, R.-H., Hou, Z.-S., Huang, B., & Jin, S.-T. (2015). A unified data-driven design
framework of optimality-based generalized iterative learning control. Computers
& Chemical Engineering, 77, 10–23.
Chi, R.-H., Hui, Y., & Hou, Z.-S. (2022). Data-Driven Iterative Learning Control for
Discrete-Time Systems. Springer, Singapore.
Chi, R.-H., Liu, X.-H., Zhang, R.-K., Hou, Z.-S., & Huang, B. (2018). Constrained
data-driven optimal iterative learning control. Journal of Process Control, 55,
10–29.
Costa, B., Škrjanc, I., Blažič, S., & Angelov, P. (2013). A practical implementation of
self-evolving cloud-based control of a pilot plant. Proceedings of 2013 IEEE
International Conference on Cybernetics, Lausanne, Switzerland, 7–12.
Coutinho, P. H., Araujo, R. F., Nguyen, A.-T., & Palhares, R. M. (2020). A multiple-
parameterization approach for local stabilization of constrained Takagi-Sugeno
fuzzy systems with nonlinear consequents. Information Sciences, 506, 295–307.
da Silva Moreira, J., Acioli Júnior, G., & Rezende Barros, G. (2018). Time and
frequency domain data-driven PID iterative tuning. IFAC-PapersOnLine, 51
(15), 1056–1061.
de Bruyne, F., & Kammer, L. C. (1999). Iterative feedback tuning with guaranteed
stability. Proceedings of 1999 American Control Conference, San Diego, CA,
USA, 5, 3317–3321.
Dehak, A., Nguyen, A.-T., Dequidt, A., Vermeiren, L., & Dambrine, M. (2020).
Reduced-complexity affine representation for Takagi-Sugeno fuzzy systems.
IFAC-PapersOnLine, 53 (2), 8031–8036.
Dehak, A., Nguyen, A.-T., Dequidt, A., Vermeiren, L., & Dambrine, M. (2022).
Reduced-complexity LMI conditions for admissibility analysis and control
design of singular nonlinear systems. IEEE Transactions on Fuzzy Systems, 31
(4), 1377–1390.
Ding, F., Shan, H., Han, X., Jiang, C., Peng, C., & Liu, J. (2022). Security-based
resilient triggered output feedback lane keeping control for human-machine
cooperative steering intelligent heavy truck under denial-of-service attacks.
IEEE Transactions on Fuzzy Systems, 31 (7), 2264–2276.
Dombi, J., & Hussain, A. (2019). Data-driven arithmetic fuzzy control using the
distending function. In T. Ahram, R. Taiar, S. Colson, and A. Choplin (Eds.),
IHIET: Human Interaction and Emerging Technologies (Advances in Intelligent
Systems and Computing, vol. 1018, pp. 215–221), Springer, Cham.
Dombi, J., & Hussain, A. (2023). Data-driven interval type-2 fuzzy inference system
based on the interval type-2 distending function. IEEE Transactions on Fuzzy
Systems, 31 (7), 2345–2359.
Dong, J., Wang, Y., & Yang, G. H. (2009). Control synthesis of continuous-time TS
fuzzy systems with local nonlinear models. IEEE Transactions on Systems, Man,
and Cybernetics, Part B (Cybernetics), 39 (5), 1245–1258.
Dong, J., Wang, Y., & Yang, G. H. (2010). Output feedback fuzzy controller design
with local nonlinear feedback laws for discrete-time nonlinear systems. IEEE
Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), 40 (6),
1447–1459.
Dovžan, D., Blažič, S., & Škrjanc, I. (2014). Towards evolving fuzzy reference
controller. Proceedings of 2014 IEEE Conference on Evolving and Adaptive
Intelligent Systems (EAIS), Linz, Austria, 1–8.
Dragos, C.-A., Preitl, S., Precup, R.-E., Cretiu, M., & Fodor, J. (2010). Modern control
solutions with applications in mechatronic systems. In I. J. Rudas, J. Fodor, and
J. Kacprzyk (Eds.), Computational Intelligence in Engineering (Studies in
Computational Intelligence, vol. 313, pp. 87–102), Springer, Berlin, Heidelberg.
El-Sousy, F. F. M. (2014). Adaptive hybrid control system using a recurrent RBFN-
based self-evolving fuzzy-neural-network for PMSM servo drives. Applied Soft
Computing, 21, 509–532.
El-Sousy, F. F. M. (2016). Intelligent mixed H2/H∞ adaptive tracking control system
design using self-organizing recurrent fuzzy-wavelet-neural-network for
uncertain two-axis motion control system. Applied Soft Computing, 41, 22–50.
Feng, G. (2004). Stability analysis of discrete-time fuzzy dynamic systems based on
piecewise Lyapunov functions. IEEE Transactions on Fuzzy Systems, 12 (1),
22–28.
Feng, G. (2006). A survey on analysis and design of model-based fuzzy control
systems. IEEE Transactions on Fuzzy Systems, 14 (5), 676–697.
Feng, T.-S., Zhang, Z.-Y., Gao, F., & Li, B.-Y. (2021). Application of adaptive fuzzy
ADRC for hypersonic flight vehicle. Proceedings of 2021 33rd Chinese Control
and Decision Conference, Kunming, China, 1–6.
Fang, Z., Wang, J., Wang, Z., Liang, J., Liu, Y., & Yin, G. (2023). A human-machine
shared control framework considering time-varying driver characteristics. IEEE
Transactions on Intelligent Vehicles, 8 (7), 3826–3838.
Ferdaus, M. M., Pratama, M., Anavatti, S. G., & Garratt, M. (2019). A generic self-
evolving neuro-fuzzy controller based high-performance hexacopter altitude
control system. Proceedings of 2018 IEEE International Conference on Systems,
Man, and Cybernetics, Miyazaki, Japan, 2784–2791.
Ferdaus, M. M., Pratama, M., Anavatti, S. G., Garratt, M. A., & Pan, Y. (2020). Generic
evolving self-organizing neuro-fuzzy control of bio-inspired unmanned aerial
vehicles. IEEE Transactions on Fuzzy Systems, 28 (8), 1542–1556.
Fliess, M., & Join, C. (2009). Model-free control and intelligent PID Controllers:
Towards a possible trivialization of nonlinear control?. IFAC Proceedings
Volumes, 42 (10), 1531–1550.
Fliess, M., & Join, C. (2013). Model-free control. International Journal of Control, 86
(12), 2228–2252.
Galichet, S., & Foulloy, L. (1995). Fuzzy controllers: Synthesis and equivalences. IEEE
Transactions on Fuzzy Systems, 3 (2), pp. 140–148.
Gaino, R., Covacic, M. R., Cardim, R., Sanches, M. A., De Carvalho, A. A., Biazeto, A.
R., & Teixeira, M. C. (2020). Discrete Takagi-Sugeno fuzzy models applied to
control the knee joint movement of paraplegic patients. IEEE Access, 8, 32714–
32726.
Gao, Y., & Er, M. J. (2003). Online adaptive fuzzy neural identification and control of a
class of MIMO nonlinear systems. IEEE Transactions on Fuzzy Systems, 11 (4),
462–477.
Guechi, E. H., Lauber, J., Dambrine, M., Klančar, G., & Blažič, S. (2010). PDC control
design for non-holonomic wheeled mobile robots with delayed outputs. Journal
of Intelligent & Robotic Systems, 60, 395–414.
Guelton, K., Delprat, S., & Guerra, T. M. (2008). An alternative to inverse dynamics
joint torques estimation in human stance based on a Takagi–Sugeno unknown-
inputs observer in the descriptor form. Control Engineering Practice, 16 (12),
1414–1426.
Gao, Z. (2006). Active disturbance rejection control: a paradigm shift in feedback
control system design. Proceedings of 2006 American Control Conference,
Minneapolis, MN, USA, 2399–2405.
Guerra, T. M., Blandeau, M., Nguyen, A.-T., Srihi, H., & Dequidt, A. (2020).
Stabilizing unstable biomechanical model to understand sitting stability for
persons with spinal cord injury. IFAC-PapersOnLine, 53 (2), 8001–8006.
Guerra, T. M., & Vermeiren, L. (2004). LMI-based relaxed nonquadratic stabilization
conditions for nonlinear systems in the Takagi–Sugeno’s form. Automatica, 40
(5), 823–829.
Guo, J., Wang, J., Luo, Y., & Li, K. (2020). Robust lateral control of autonomous four-
wheel independent drive electric vehicles considering the roll effects and
actuator faults. Mechanical Systems and Signal Processing, 143, paper 106773.
Guo, Y.-F., Lama, P., Jiang, C.-J., & Zhou, X.-B. (2014). Automated and agile server
parameter tuning by coordinated learning and control. IEEE Transactions on
Parallel and Distributed Systems, 25 (4), 876–886.
Hagras, H. (2008). Type-2 fuzzy logic controllers: a way forward for fuzzy systems in
real world environments. In J. M. Zurada, G. G. Yen, and J. Wang (Eds.),
Computational Intelligence: Research Frontiers. WCCI 200 (Lecture Notes in
Computer Science, vol. 5050, pp. 181–200), Springer, Berlin, Heidelberg.
Hamza, M. F., Yap, H. J., Choudhury, I. A., Chiroma, H., & Kumbasar, T. (2018). A
survey on advancement of hybrid type 2 fuzzy sliding mode control. Neural
Computing and Applications, 30 (2), 331–353.
Han, H., Zhou, W., Qiao, J., & Feng, G. (2015). A direct self-constructing neural
controller design for a class of nonlinear systems. IEEE Transactions on Neural
Networks and Learning Systems, 26 (6), 1312–1322.
Han, H. G., Zhang, L., Hou, Y., & Qiao, J. F. (2016). Nonlinear Model Predictive
Control Based on a Self-Organizing Recurrent Neural Network. IEEE
Transactions on Neural Networks and Learning Systems, 27 (2), 402–415.
Han, S. Y., Zhong, X. F., Chen, Y. H., & Tang, G. Y. (2019). Fuzzy guaranteed cost H ∞
control of uncertain nonlinear fuzzy vehicle active suspension with random
actuator delay. International Journal of Fuzzy Systems, 21(7), 2021–2031.
Hao, L. Y., Zhang, H., Li, T. S., Lin, B., & Chen, C. P. (2021). Fault tolerant control for
dynamic positioning of unmanned marine vehicles based on TS fuzzy model
with unknown membership functions. IEEE Transactions on Vehicular
Technology, 70 (1), 146–157.
He, Y., Li, F., Li, J.-K., Liu, J.-S., & Wu, X.-Y. (2022). An sEMG based adaptive
method for human-exoskeleton collaboration in variable walking environments.
Biomedical Signal Processing and Control, 74, paper 103477.
Hjalmarsson, H. (2002). Iterative feedback tuning - an overview. International Journal
of Adaptive Control and Signal Processing, 16 (5), 373–395.
Hjalmarsson, H., Gevers, M., Gunnarsson, S., & Lequin, O. (1998). Iterative feedback
tuning: theory and applications. IEEE Control Systems Magazine, 18 (4), 26–41.
Hjalmarsson, H., Gunnarsson, S., & Gevers, M. (1994). A convergent iterative
restricted complexity control design scheme. Proceedings of 33rd IEEE
Conference on Decision and Control, Orlando, FL, USA, 2, 1735–1740.
Hoffmann, C., & Werner, H. (2014). A survey of linear parameter-varying control
applications validated by experiments or high-fidelity simulations. IEEE
Transactions on Control Systems Technology, 23(2), 416–433.
Holmblad, L. P., & Ostergaard, J. J. (1982). Control of a cement kiln by fuzzy logic. In
M. M. Gupta and E. Sanchez (Eds.), Fuzzy Information and Decision Processes
(pp. 389–399), North Holland, Amsterdam.
Hou, Z.-S. (1999). Nonparametric Models and Its Adaptive Control Theory. Science
Press, Beijing.
Hou, Z.-S., & Jin, S. (2011a). A novel data-driven control approach for a class of
discrete-time nonlinear systems. IEEE Transactions on Control Systems
Technology, 19 (6), 1549–1558.
Hou, Z.-S., & Jin, S. (2011b). Data-driven model-free adaptive control for a class of
MIMO nonlinear discrete-time systems. IEEE Transactions on Neural
Networks, 22 (12), 2173–2188.
Hou, Z.-S., & Huang, W.-H. (1997). The model-free learning adaptive control of a class
of SISO nonlinear systems. Proceedings of 1997 American Control Conference,
Albuquerque, NM, USA, 343–344.
Hou, Z.-S., & Wang, Z. (2013). From model-based control to data-driven control:
Survey, classification and perspective. Information Sciences, 235, 3–35.
Hou, Z.-S., & Zhou, Y.-M. (2013). Model based control and MFAC, which is better in
simulation?. IFAC Proceedings Volumes, 46 (13), 82–87.
Hsu, C. F. (2013). A self-evolving functional-linked wavelet neural network for control
applications. Applied Soft Computing Journal, 13 (11), 4392–4402.
Hsu, C. F., & Wong, K. Y. (2016). On-line constructive fuzzy sliding-mode control for
voice coil motors. Applied Soft Computing Journal, 47, 415–423.
Hu, C., Chen, Y., & Wang, J. (2020). Fuzzy observer-based transitional path-tracking
control for autonomous vehicles. IEEE Transactions on Intelligent
Transportation Systems, 22 (5), 3078–3088.
Hu, X., Wu, L., Hu, C., & Gao, H. (2013). Adaptive fuzzy integral sliding mode control
for flexible air-breathing hypersonic vehicles subject to input nonlinearity.
Journal of Aerospace Engineering, 26 (4), 721–734.
Huang, C. H., Wang, W. J., & Chiu, C. H. (2010). Design and implementation of fuzzy
control on a two-wheel inverted pendulum. IEEE Transactions on Industrial
Electronics, 58 (7), 2988–3001.
Huynh, T. T., Lin, C. M., Le, T. L., Le, N. Q. K., Vu, V. P., & Chao, F. (2022). Self-
Organizing Double Function-Link Fuzzy Brain Emotional Control System
Design for Uncertain Nonlinear Systems. IEEE Transactions on Systems, Man,
and Cybernetics: Systems, 52 (3), 1852–1868.
Ioannou, P. A., & Kokotovic, P. V. (1984). Instability analysis and improvement of
robustness of adaptive control. Automatica, 20 (5), 583–594.
Jeong, K., & Choi, S. B. (2021). Takagi–Sugeno fuzzy observer-based
magnetorheological damper fault diagnosis using a support vector machine.
IEEE Transactions on Control Systems Technology, 30 (4), 1723–1735.
Jiang, B., Gao, Z., Shi, P., & Xu, Y. (2010). Adaptive fault-tolerant tracking control of
near-space vehicle using Takagi–Sugeno fuzzy models. IEEE Transactions on
Fuzzy Systems, 18 (5), 1000–1007.
Johansson, M., Rantzer, A., & Arzen, K. E. (1999). Piecewise quadratic stability of
fuzzy systems. IEEE Transactions on Fuzzy Systems, 7 (6), 713–722.
Juang, C. F., & Lin, C. T. (1998). An on-line self-constructing neural fuzzy inference
network and its applications. IEEE Transactions on Fuzzy Systems, 6 (1), 12–32.
Juang, C. F., & Tsao, Y. W. (2008). A self-evolving interval type-2 fuzzy neural
network with online structure and parameter learning. IEEE Transactions on
Fuzzy Systems, 16 (6), 1411–1424.
Jung, H., Jeon, K., Kang, J.-G., & Oh, S. (2021). Iterative feedback tuning of cascade
control of two-inertia system. IEEE Control Systems Letters, 5 (3), pp. 785–790.
Kadri, M. B., & Hussain, S. (2010). Model free adaptive control based on FRM with an
approach to reduce the control activity. Proceedings of 2010 IEEE International
Conference on Systems, Man and Cybernetics, Istanbul, Turkey, 2110–2115.
Kammer, L. C., Bitmead, R. R., & Bartlett, P. L. (2000). Direct iterative tuning via
spectral analysis. Automatica, 36 (9), 1301–1307.
Kammer, L. C., De Bruyne, F., & Bitmead, R. R. (1999). Iterative feedback tuning via
minimization of the absolute error. Proceedings of 38th IEEE Conference on
Decision and Control, Phoenix, AZ, USA, 1999, 5, 4619–4624.
Karimi, A., Mišković L., & Bonvin, D. (2004). Iterative correlation-based controller
tuning. International Journal of Adaptive Control and Signal Processing, 18 (8),
645–664.
Kasabov, N. K., & Song, Q. (2002). DENFIS: Dynamic Evolving Neural-Fuzzy
Inference System and its application for time-series prediction. IEEE
Transactions on Fuzzy Systems, 10 (2), 144–154.
Kaynak, O., Erbatur, K., & Ertugrul, M. (2001). The fusion of computationally
intelligent methodologies and sliding-mode control – a survey. IEEE
Transactions on Industrial Electronics, 48 (1), 4–17.
Khadraoui, S., Nounou, H., Nounou, M., Datta, A., & Bhattacharyya, S. P. (2014). A
model-free design of reduced-order controllers and application to a DC
servomotor. Automatica, 50 (8), 2142–2149.
Khalil, H. K. (2002). Nonlinear Systems, 3rd Edition. Prentice Hall, Englewood Cliffs,
NJ.
Kim, H. S., Lee, K., & Joo, Y. H. (2021). Decentralized sampled-data fuzzy controller
design for a VTOL UAV. Journal of the Franklin Institute, 358 (3), 1888–1914.
Lam, H. K. (2018). A review on stability analysis of continuous-time fuzzy-model-
based control systems: From membership-function-independent to membership-
function-dependent analysis. Engineering Applications of Artificial Intelligence,
67, 390–408.
Lam, H. K., Li, H., Deters, C., Secco, E. L., Wurdemann, H. A., & Althoefer, K. (2013).
Control design for interval type-2 fuzzy systems under imperfect premise
matching. IEEE Transactions on Industrial Electronics, 61 (2), 956–968.
Lam, H. K., & Narimani, M. (2009). Quadratic-stability analysis of fuzzy-model-based
control systems using staircase membership functions. IEEE Transactions on
Fuzzy Systems, 18 (1), 125–137.
Lam, H. K., & Seneviratne, L. D. (2008). Stability analysis of interval type-2 fuzzy-
model-based control systems. IEEE Transactions on Systems, Man, and
Cybernetics, Part B (Cybernetics), 38 (3), 617–628.
Le, T. L., Lin, C. M., & Huynh, T. T. (2018). Self-evolving type-2 fuzzy brain
emotional learning control design for chaotic systems using PSO. Applied Soft
Computing, 73, 418–433.
Lee, D. H., Joo, Y. H., & Tak, M. H. (2014). Local stability analysis of continuous-time
Takagi–Sugeno fuzzy systems: A fuzzy Lyapunov function approach.
Information Sciences, 257, 163–175.
Lee, W.-P., & Hallam, J. (1999). Evolving reliable and robust controllers for real robots
by genetic programming. Soft Computing, 3, 63–75.
Li, A., Liu, M., & Shi, Y. (2020). Adaptive sliding mode attitude tracking control for
flexible spacecraft systems based on the Takagi-Sugeno fuzzy modelling
method. Acta Astronautica, 175, 570–581.
Li, H., Xu, J., & Yu, J. (2022a). Discrete event-triggered fault-tolerant control of
underwater vehicles based on Takagi–Sugeno fuzzy model. IEEE Transactions
on Systems, Man, and Cybernetics: Systems, 53(3), 1841–1851.
Li, H., Yu, J., Hilton, C., & Liu, H. (2012). Adaptive sliding-mode control for nonlinear
active suspension vehicle systems using T-S fuzzy approach. IEEE Transactions
on Industrial Electronics, 60 (8), 3328–3338.
Li, H.-Y., Chen, Z., Wu, L.-G., & Lam, H.-K. (2017). Event-triggered control for
nonlinear systems under unreliable communication links. IEEE Transactions on
Fuzzy Systems, 25 (4), 813–824.
Li, P., Nguyen, A.-T., Du, H., Wang, Y., & Zhang, H. (2021). Polytopic LPV
approaches for intelligent automotive systems: State of the art and future
challenges. Mechanical Systems and Signal Processing, 161, paper 107931.
Li, T., Dai, Z., Song, G., Wang, L., & Du, H. (2019). Fault tolerant tracking of Mars
entry vehicles via fuzzy control approach. Fuzzy Sets and Systems, 371, 123–
135.
Li, W., Xie, Z., Zhao, J., Gao, J., Hu, Y., & Wong, P. K. (2021). Human-machine
shared steering control for vehicle lane keeping systems via a fuzzy observer-
based event-triggered method. IEEE Transactions on Intelligent Transportation
Systems, 23 (8), 13731–13744.
Li, W.-R., Yuan, H.-W., Li, S.-N., Zhu, J.-G. (2022). A revisit to model-free control.
IEEE Transactions on Power Electronics, 37 (12), 14408–14421.
Li, X.-A., Sung, K., Guo, C.-Q., & Liu, H. (2022). Hybrid adaptive disturbance
rejection control for inflatable robotic arms. ISA Transactions, 126, 617–628.
Liang, J., Feng, J., Lu, Y., Yin, G., Zhuang, W., & Mao, X. (2023). A direct yaw
moment control framework through robust TS fuzzy approach considering
vehicle stability margin. IEEE/ASME Transactions on Mechatronics, DOI:
10.1109/TMECH.2023.3274689.
Lin, C. M., & Li, H. Y. (2013). Intelligent hybrid control system design for antilock
braking systems using self-organizing function-link fuzzy cerebellar model
articulation controller. IEEE Transactions on Fuzzy Systems, 21 (6), 1044–1055.
Lin, C. M., Ramarao, R., & Gopalai, S. H. (2019). Self-Organizing Adaptive Fuzzy
Brain Emotional Learning Control for Nonlinear Systems. International Journal
of Fuzzy Systems, 21 (7), 1989–2007.
Liu, L., Illian, M., Leonhardt, S., & Misgeld, B. J. E. (2023). Iterative learning control
for cascaded impedance-controlled compliant exoskeleton with adaptive reaction
to spasticity. IEEE Transactions on Instrumentation and Measurement, 72,
paper 4008211.
Liu, X. D., & Zhang, Q. L. (2003). New approaches to H-infinity controller designs
based on fuzzy observers for TS fuzzy systems via LMI. Automatica, 39 (9),
1571–1582.
Liu, Y., Chen, R., Na, X., Luo, Y., & Zhang, H. (2020). Robust fault estimation of
vehicular yaw rate sensor using a type-2 fuzzy approach. IEEE Transactions on
Industrial Electronics, 68 (10), 10029–10039.
Ljung, L. (1999). System Identification: Theory for the User, 2nd Edition. Prentice Hall,
Englewood Cliffs, NJ.
Losero, R., Lauber, J., & Guerra, T. M. (2018). Virtual strain gauge based on a fuzzy
discrete angular domain observer: Application to engine and clutch torque
estimation issues. Fuzzy Sets and Systems, 343, 76–96.
Lu, H. C., Chang, M. H., & Tsai, C. H. (2011). Adaptive self-constructing fuzzy neural
network controller for hardware implementation of an inverted pendulum
system. Applied Soft Computing Journal, 11 (5), 3962–3975.
Koelewijn, P. J., & Tóth, R. (2020). Scheduling dimension reduction of LPV models-a
deep neural network approach. Proceedings of 2020 American Control
Conference, Denver, CO, USA, 1111–1117.
Kwiatkowski, A., & Werner, H. (2008). PCA-based parameter set mappings for LPV
models with fewer parameters and less overbounding. IEEE Transactions on
Control Systems Technology, 16 (4), 781–788.
Ma, T., & Wang, B. (2021). Disturbance observer-based Takagi-Sugeno fuzzy control
of a delay fractional-order hydraulic turbine governing system with elastic water
hammer via frequency distributed model. Information Sciences, 569, 766–785.
Ma, Y., Nie, Z., Yu, Y., Hu, S., & Peng, Z. (2020). Event-triggered fuzzy control of
networked nonlinear underactuated unmanned surface vehicle. Ocean
Engineering, 213, paper 107540.
Mamdani, E. H. (1974). Applications of fuzzy control for control of simple dynamic
plant. Proceedings of the IEE, 121 (12), 1585−1588.
Mamdani, E. H., & Assilian, S. (1975). An experiment in linguistic synthesis with a
fuzzy logic controller. International Journal of Man-Machine Studies, 7 (1),
1−13.
Merah, A., Hartani, K., Yazid, N. E. H., & Chikouche, T. M. (2022). New integrated
full vehicle suspension system for improvements in vehicle ride comfort and
road holding. SAE International Journal of Vehicle Dynamics, Stability, and
NVH, 6 (3), 267–281.
Mišković, L., Karimi, A., Bonvin, D., & Gevers, M. (2007). Correlation-based tuning of
decoupling multivariable controllers. Automatica, 43 (9), 1481–1494.
Mittal, K., Jain, A., Vaisla, K. S., Castillo, O., & Kacprzyk, J. (2020). A comprehensive
review on type 2 fuzzy logic applications: Past, present and future. Engineering
Applications of Artificial Intelligence, 95, 103916.
Mitra, S., & Y. Hayashi, Y. (2000). Neuro-fuzzy rule generation: survey in soft
computing framework. IEEE Transactions on Neural Networks, 11 (3), 748–
768.
Montagner, V. F., Oliveira, R. C., & Peres, P. L. (2009). Convergent LMI relaxations
for quadratic stabilizability and H-infinity control of Takagi–Sugeno fuzzy
systems. IEEE Transactions on Fuzzy Systems, 17 (4), 863–873.
Mozelli, L. A., Palhares, R. M., & Avellar, G. S. (2009). A systematic approach to
improve multiple Lyapunov function stability and stabilization conditions for
fuzzy systems. Information Sciences, 179 (8), 1149–1162.
Ning, D., Sun, S., Zhang, F., Du, H., Li, W., & Zhang, B. (2017). Disturbance observer-
based Takagi-Sugeno fuzzy control for an active seat suspension. Mechanical
Systems and Signal Processing, 93, 515–530.
Ngo, T. Q., Hoang, D. K., Tran, T. T., Nguyen, T. T., Nguyen, V. T., & Le, L. H.
(2022). A novel self-organizing fuzzy cerebellar model articulation controller
based overlapping gaussian membership function for controlling robotic system.
International Journal of Computers, Communications and Control, 17 (4), paper
4606.
Ngo, T. Q., & Phuong, T. V. (2015). Robust adaptive self-organizing wavelet fuzzy
cmac tracking control for deicing robot manipulator. International Journal of
Computers Communications & Control, 10 (4), 567–578.
Nguyen, A.-T., Campos, V., Guerra, T. M., Pan, J., & Xie, W. (2021). Takagi–Sugeno
fuzzy observer design for nonlinear descriptor systems with unmeasured premise
variables and unknown inputs. International Journal of Robust and Nonlinear
Control, 31 (17), 8353–8372.
Nguyen, A.-T., Coutinho, P., Guerra, T. M., Palhares, R., & Pan, J. (2021). Constrained
output-feedback control for discrete-time fuzzy systems with local nonlinear
models subject to state and input constraints. IEEE Transactions on Cybernetics,
51 (9), 4673–4684.
Nguyen, A.-T., Dehak, A., Guerra, T. M., & Sugeno, M. (2023). Convex stability
analysis of Mamdani-like fuzzy systems with singleton consequents. IEEE
Transactions on Fuzzy Systems, 31 (11), 3787–3798.
Nguyen, A.-T., Dinh, T. Q., Chong, J.-J., Iwasaki, M., Precup, R.-E., & Ruderman, M.
(2023). Guest editorial introduction to the special issue on “Emerging control
and automation technologies for advanced mechatronic systems”. Control
Engineering Practice, 136, paper 105532.
Nguyen, A.-T., Dinh, T. Q., Guerra, T. M., & Pan, J. (2021). Takagi–Sugeno fuzzy
unknown input observers to estimate nonlinear dynamics of autonomous ground
vehicles: Theory and real-time verification. IEEE/ASME Transactions on
Mechatronics, 26 (3), 1328–1338.
Nguyen, C. M., Nguyen, A.-T., & Delprat, S. (2023). Neural-network-based fuzzy
observer with data-driven uncertainty identification for vehicle dynamics
estimation under extreme driving conditions: theory and experimental results.
IEEE Transactions on Vehicular Technology, 72 (7), 8686–8696.
Nguyen, V. A., Nguyen, A.-T., Dequidt, A., Vermeiren, L., & Dambrine, M. (2019).
Nonlinear tracking control with reduced complexity of serial robots: A robust
fuzzy descriptor approach. International Journal of Fuzzy Systems, 21 (4),
1038–1050.
Nguyen, A.-T., Pan, J., Guerra, T. M., & Wang, Z. (2020). Avoiding unmeasured
premise variables in designing unknown input observers for Takagi–Sugeno
fuzzy systems. IEEE Control Systems Letters, 5 (1), 79–84.
Nguyen, A.-T., Rath, J., Guerra, T. M., Palhares, R., & Zhang, H. (2020). Robust set-
invariance based fuzzy output tracking control for vehicle autonomous driving
under uncertain lateral forces and steering constraints. IEEE Transactions on
Intelligent Transportation Systems, 22 (9), 5849–5860.
Nguyen, A.-T., Sentouh, C., & Popieul, J. C. (2016). Driver-automation cooperative
approach for shared steering control under multiple system constraints: Design
and experiments. IEEE Transactions on Industrial Electronics, 64 (5), 3819–
3830.
Nguyen, A.-T., Sentouh, C., & Popieul, J. C. (2017). Sensor reduction for driver-
automation shared steering control via an adaptive authority allocation strategy.
IEEE/ASME Transactions on Mechatronics, 23 (1), 5–16.
Nguyen, A.-T., Sentouh, C., & Popieul, J. C. (2018). Fuzzy steering control for
autonomous vehicles under actuator saturation: Design and experiments. Journal
of the Franklin Institute, 355 (18), 9374–9395.
Nguyen, A.-T., Sentouh, C., Zhang, H., & Popieul, J. C. (2019). Fuzzy static output
feedback control for path following of autonomous vehicles with transient
performance improvements. IEEE Transactions on Intelligent Transportation
Systems, 21 (7), 3069–3079.
Nguyen, A.-T., Sugeno, M., Campos, V., & Dambrine, M. (2016). LMI-based stability
analysis for piecewise multi-affine systems. IEEE Transactions on Fuzzy
Systems, 25 (3), 707–714.
Nguyen, A.-T., Taniguchi, T., Eciolaza, L., Campos, V., Palhares, R., & Sugeno, M.
(2019). Fuzzy control systems: past, present and future. IEEE Computational
Intelligence Magazine, 14 (1) 56–68.
Nikdel, P., Hosseinpour, M., Badamchizadeh, M. A., & Akbari, M. A. (2014).
Improved Takagi-Sugeno fuzzy model-based control of flexible joint robot via
Hybrid-Taguchi genetic algorithm. Engineering Applications of Artificial
Intelligence, 33, 12–20.
Pan, J., Nguyen, A.-T., Guerra, T. M., & Ichalal, D. (2020). A unified framework for
asymptotic observer design of fuzzy systems with unmeasurable premise
variables. IEEE Transactions on Fuzzy Systems, 29 (10), 2938–2948.
Pan, J., Nguyen, A.-T., Guerra, T. M., Sentouh, C., Wang, S., & Popieul, J. C. (2022).
Vehicle actuator fault detection with finite-frequency specifications via Takagi-
Sugeno fuzzy observers: Theory and experiments. IEEE Transactions on
Vehicular Technology, 72 (1), 407–417.
Park, J. H., Park, G. T., Kim, S. H., & Moon, C. J. (2005). Direct adaptive self-
structuring fuzzy controller for nonaffine nonlinear system. Fuzzy Sets and
Systems, 153 (3), 429–445.
Peixoto, M. L., Nguyen, A.-T., Guerra, T. M., & Palhares, R. M. (2023). Unknown
input observers for time-varying delay Takagi-Sugeno fuzzy systems with
unmeasured nonlinear consequents. European Journal of Control, 72, paper
100830.
Phan, P. A., & Gale, T. J. (2008). Direct adaptive fuzzy control with a self-structuring
algorithm. Fuzzy Sets and Systems, 159 (8), 871–899.
Prajna, S., Papachristodoulou, A., & Parrilo, P. (2002). SOSTOOLS - sum of squares
optimization toolbox, User’s Guide,
http://www.mit.edu/~parrilo/sostools/index.html.
Precup, R.-E., Angelov, P., Costa, B. S. J., & Sayed-Mouchaweh, M. (2015). An
overview on fault diagnosis and nature-inspired optimal control of industrial
process applications. Computers in Industry, 74, 75–94.
Precup, R.-E., & David, R.-C. (2016). Nature-inspired optimization of fuzzy controllers
and fuzzy models. In P. P. Angelov (Ed.), Handbook on Computational
Intelligence (Volume 2: Evolutionary Computation, Hybrid Systems, and
Applications, pp. 697−729), World Scientific, Singapore.
Precup, R.-E., & David, R.-C. (2022). Nature-inspired optimal tuning of fuzzy
controllers. In P. P. Angelov (Ed.), Handbook on Computer Learning and
Intelligence, 2nd Edition (Volume 2: Deep Learning, Intelligent Control and
Evolutionary Computation, pp. 775−808), World Scientific, Singapore.
Precup, R.-E., David, R.-C., Roman, R.-C., Szedlak-Stinean, A.-I., & Petriu, E. M.
(2021). Optimal tuning of interval type-2 fuzzy controllers for nonlinear servo
systems using slime mould algorithm. International Journal of Systems Science,
DOI: 10.1080/00207721.2021.1927236.
Precup, R.-E., Haidegger, T., Preitl, S., Benyó, B., Paul, A. S., & Kovács, L. (2012).
Fuzzy control solution for telesurgical applications. Applied and Computational
Mathematics, 11 (3), 378–397.
Precup, R.-E., & Hellendoorn, H. (2011). A survey on industrial applications of fuzzy
control. Computers in Industry, 62 (3), 213−226.
Precup, R.-E., & Preitl, S. (1999a). Fuzzy Controllers. Editura Orizonturi Universitare,
Timisoara.
Precup, R.-E., & Preitl, S. (1999b). Development of some fuzzy controllers with non-
homogenous dynamics with respect to the input channels meant for a class of
systems. Proceedings of 1999 European Control Conference, Karlsruhe,
Germany, 61–66.
Precup, R.-E., & Preitl, S. (2003). Development of fuzzy controllers with non-
homogeneous dynamics for integral-type plants. Electrical Engineering, 85 (3),
155–168.
Precup, R.-E., Preitl, S., Bojan-Dragos, C.-A., Hedrea, E.-L., Roman, R.-C., & Petriu,
E. M. (2022). A low-cost approach to data-driven fuzzy control of servo
systems. Facta Universitatis, Series: Mechanical Engineering, 20 (1), 21−36.
Precup, R.-E., Preitl, S., Petriu, E. M., Bojan-Dragos, C.-A., Szedlak-Stinean, A.-I.,
Roman, R.-C., & Hedrea, E.-L. (2020). Model-based fuzzy control results for
networked control systems. Reports in Mechanical Engineering, 1 (1), 10–25.
Precup, R.-E., Preitl, S., Petriu, E. M., Roman, R.-C., Bojan-Dragos, C.-A., Hedrea, E.-
L., & Szedlak-Stinean, A.-I. (2020). A center manifold theory-based approach to
the stability analysis of state feedback Takagi-Sugeno-Kang fuzzy control
systems. Facta Universitatis, Series: Mechanical Engineering, 18 (2), 189–204.
Precup, R.-E., Preitl, S., Petriu, E. M., Tar, J. K., Tomescu, M. L., & Pozna, C. (2009).
Generic two-degree-of-freedom linear and fuzzy controllers for integral
processes. Journal of The Franklin Institute, 346 (10), 980–1003.
Precup, R.-E., Preitl, S., Rudas, I. J., Tomescu, M. L., & Tar, J. K. (2008). Design and
experiments for a class of fuzzy controlled servo systems. IEEE/ASME
Transactions on Mechatronics, 13 (1), pp. 22–35.
Precup, R.-E., Preitl, S., Tar, J. K., Tomescu, M. L., Takács, M., Korondi, P., &
Baranyi, P. (2008). Fuzzy control system performance enhancement by iterative
learning control. IEEE Transactions on Industrial Electronics, 55 (9), 3461–
3475.
Precup, R.-E., Radac, M.-B., Preitl, S., Petriu, E. M., & Dragos, C.-A. (2009). Iterative
Feedback Tuning in Linear and Fuzzy Control Systems. In I. J. Rudas, J. Fodor
and J. Kacprzyk (Eds.), Towards Intelligent Engineering and Information
Technology (Studies in Computational Intelligence, vol. 243, pp. 179–192),
Springer-Verlag, Berlin, Heidelberg, New York.
Precup, R.-E., Radac, M.-B., Tomescu, M. L., Petriu, E. M., & Preitl, S. (2013). Stable
and convergent iterative feedback tuning of fuzzy controllers for discrete-time
SISO systems. Expert Systems with Applications, 40 (1), 188–199.
Precup, R.-E., Roman, R.-C., & Safaei, A. (2021). Data-Driven Model-Free
Controllers, 1st Edition. CRC Press, Taylor & Francis, Boca Raton, FL.
Precup, R.-E., Tomescu, M. L., Radac, M.-B., Petriu, E. M., Preitl, S., & Dragos, C.-A.
(2012). Iterative performance improvement of fuzzy control systems for three
tank systems. Expert Systems with Applications, 39 (9), 8288–8299.
Preitl, S., Precup, R.-E., Fodor, J., & Bede, B. (2006). Iterative feedback tuning in fuzzy
control systems. Theory and applications. Acta Polytechnica Hungarica, 3 (3),
81–96.
Preitl, S., Precup, R.-E., Preitl, Z., Vaivoda, S., Kilyeni, S., & Tar, J. K. (2007). Iterative
feedback and learning control. Servo systems applications. IFAC Proceedings
Volumes, 40 (8), 16–27.
Preitl, S., Stinean, A.-I., Precup, R.-E., Preitl, Z., Petriu, E. M., Dragos, C.-A., & Radac,
M.-B. (2012). Controller design methods for driving systems based on
extensions of symmetrical optimum method with DC and BLDC motor
applications. IFAC Proceedings Volumes, 45 (3), 264–269.
Qi, P., Liu, C., Ataka, A., Lam, H. K., & Althoefer, K. (2016). Kinematic control of
continuum manipulators using a fuzzy-model-based approach. IEEE
Transactions on Industrial Electronics, 63 (8), 5022–5035.
Qing, Y., Hongliang, Z., Songlin, C., Weiwei, M., & Yongfeng, J. (2023). Designing a
switched Takagi-Sugeno fuzzy controller for CDC semi-active suspensions with
current input constraint. Mechanical Systems and Signal Processing, 199, paper
110450.
Qiu, J.-B., Gao, H.-J., & Ding, S.X. 92016). Recent advances on fuzzy-model-based
nonlinear networked control systems: A survey. IEEE Transactions on
Industrial Electronics, 63 (2), 1207–1217.
Quadros, M. M., Leite, V. J., & Palhares, R. M. (2022). Robust fault hiding approach
for T–S fuzzy systems with unmeasured premise variables. Information
Sciences, 589, 690–715.
Radac, M.-B., Precup, R.-E., & Petriu, E. M. (2015). Constrained data-driven model-
free ILC-based reference input tuning algorithm. Acta Polytechnica Hungarica,
12 (1), 137–160.
Rhee, B. J., & Won, S. (2006). A new fuzzy Lyapunov function approach for a Takagi-
Sugeno fuzzy control system design. Fuzzy Sets and Systems, 157 (9), 1211–
1228.
Roman, R.-C., Precup, R.-E., Bojan-Dragos, C.-A., & Szedlak-Stinean, A.-I. (2019).
Combined model-free adaptive control with fuzzy component by virtual
reference feedback tuning for tower crane systems. Procedia Computer Science,
162, 267–274.
Roman, R.-C., Precup, R.-E., & David, R.-C. (2018). Second order intelligent
proportional-integral fuzzy control of twin rotor aerodynamic systems. Procedia
Computer Science, 139, 372–380.
Roman, R.-C., Precup, R.-E., Hedrea, E.-L., Preitl, S., Zamfirache, I. A., Bojan-Dragos,
C.-A., & Petriu, E. M. (2022). Iterative feedback tuning algorithm for tower
crane systems. Procedia Computer Science, 199, 157–165.
Roman, R.-C., Precup, R.-E., & Petriu, E. M. (2021). Hybrid data-driven fuzzy active
disturbance rejection control for tower crane systems. European Journal of
Control, 58, 373–387.
Roman, R.-C., Precup, R.-E., Petriu, E. M., Hedrea, E.-L., Bojan-Dragos, C.-A., &
Radac, M.-B. (2019). Model-free adaptive control with fuzzy component for
tower crane systems. Proceedings of 2019 IEEE International Conference on
Systems, Man and Cybernetics, Bari, Italy, 1400–1405.
Sala, A., & Ariño, C. (2007). Asymptotically necessary and sufficient conditions for
stability and performance in fuzzy control: Applications of Polya’s theorem.
Fuzzy Sets and Systems, 158(24), 2671-2686.
Sala, A., & Arino, C. (2008). Relaxed stability and performance LMI conditions for
Takagi-Sugeno fuzzy systems with polynomial constraints on membership
function shapes. IEEE Transactions on Fuzzy Systems, 16 (5), 1328–1336.
Sala, A., & Arino, C. (2009). Polynomial fuzzy models for nonlinear control: A Taylor
series approach. IEEE Transactions on Fuzzy Systems, 17 (6), 1284–1295.
Sala, A., Guerra, T. M., & Babuška, R. (2005). Perspectives of fuzzy systems and
control. Fuzzy Sets and Systems, 156 (3), 432–444.
Sala, A., Guerra, T. M., & Tanaka, K. (2015). Fuzzy control turns 50: 10 years later.
Fuzzy Sets and Systems, 281, 168–182.
Schulte, H., & Gauterin, E. (2015). Fault-tolerant control of wind turbines with
hydrostatic transmission using Takagi–Sugeno and sliding mode techniques.
Annual Reviews in Control, 40, 82–92.
Shao, X., Naghdy, F., & Du, H. (2017). Reliable fuzzy H∞ control for active suspension
of in-wheel motor driven electric vehicles with dynamic damping. Mechanical
Systems and Signal Processing, 87, 365–383.
Shen, D. (2018). Iterative learning control with incomplete information: A survey.
IEEE/CAA Journal of Automatica Sinica, 5 (5), 885–901.
Shen, S.-Y., Xu, J.-F., Chen, P., & Xia, Q.-Y. (2023). An intelligence attitude controller
based on active disturbance rejection control technology for an unmanned
helicopter. IEEE Transactions on Vehicular Technology, 72 (3), 2936–2946.
Shi, J., Sun, H., & Hou, L. (2023). Observer-based Takagi–Sugeno fuzzy sampled-data
dynamic positioning controller design for unmanned surface vehicle with
external disturbances and actuator faults. Proceedings of the Institution of
Mechanical Engineers, Part I: Journal of Systems and Control Engineering, 237
(2), 294–313.
Spall, J. C., & Cristion, J. A. (1998). Model-free control of nonlinear stochastic systems
with discrete-time measurements. IEEE Transactions on Automatic Control, 43
(9), 1198–1210.
Sugeno, M. (1999). On stability of fuzzy systems expressed by fuzzy rules with
singleton consequents. IEEE Transactions on Fuzzy Systems, 7 (2), 201–224.
Sugeno, M., & Taniguchi, T. (2004). On improvement of stability conditions for
continuous Mamdani-like fuzzy systems. IEEE Transactions on Systems, Man,
and Cybernetics, Part B (Cybernetics), 34 (1), 120-131.
Sun, C. H., Chen, Y. J., Wang, Y. T., & Huang, S. K. (2017a). Sequentially switched
fuzzy-model-based control for wheeled mobile robot with visual odometry.
Applied Mathematical Modelling, 47, 765–776.
Sun, G., Xu, S., & Li, Z. (2017b). Finite-time fuzzy sampled-data control for nonlinear
flexible spacecraft with stochastic actuator failures. IEEE Transactions on
Industrial Electronics, 64 (5), 3851–3861.
Sutton, R. S., & Barto, A. G. (2017). Reinforcement Learning: An Introduction, 2nd
Edition. MIT Press, Cambridge, MA, London.
Sutton, R. S., Barto, A. G., & Williams, R. J. (1992). Reinforcement learning is direct
adaptive optimal control. IEEE Control Systems Magazine, 12 (2), 19–22.
Škrjanc, I., Blažič, S., & Angelov, P. (2014). Robust evolving cloud-based PID control
adjusted by gradient learning method. Proceedings of 2014 IEEE Conference on
Evolving and Adaptive Intelligent Systems, Linz, Austria, 1–8.
Takagi, T., & Sugeno, M. (1985). Fuzzy identification of systems and its applications to
modeling and control. IEEE Transactions on Systems, Man, and Cybernetics,
SMC-15 (1), 116–132.
Tanaka, K., Hori, T., & Wang, H. O. (2003). A multiple Lyapunov function approach to
stabilization of fuzzy control systems. IEEE Transactions on Fuzzy Systems, 11
(4), 582–589.
Tanaka, K., Ikeda, T., & Wang, H. O. (1996). Robust stabilization of a class of
uncertain nonlinear systems via fuzzy control: quadratic stabilizability, H∞
control theory, and linear matrix inequalities. IEEE Transactions on Fuzzy
Systems, 4 (1), 1–13.
Tanaka, K., & Sugeno, M. (1992). Stability analysis and design of fuzzy control
systems. Fuzzy Sets and Systems, 45 (2), 135–156.
Tanaka, K., & Wang, H. O. (2004). Fuzzy Control Systems Design and Analysis: A
Linear Matrix Inequality Approach. John Wiley & Sons, New York, NY.
Tanaka, K., Yoshida, H., Ohtake, H., & Wang, H. (2008). A sum-of-squares approach
to modeling and control of nonlinear dynamical systems with polynomial fuzzy
systems. IEEE Transactions on Fuzzy systems, 17 (4), 911–922.
Taniguchi, T., Tanaka, K., Ohtake, H., & Wang, H. O. (2001). Model construction, rule
reduction, and robust compensation for generalized form of Takagi-Sugeno
fuzzy systems. IEEE Transactions on Fuzzy Systems, 9 (4), 525–538.
Tang, X., Du, H., Sun, S., Ning, D., Xing, Z., & Li, W. (2016). Takagi–Sugeno fuzzy
control for semi-active vehicle suspension with a magnetorheological damper
and experimental validation. IEEE/ASME Transactions on Mechatronics, 22 (1),
291–300.
Tasiu, I. A., Liu, Z., Yan, Q., Chen, H., Hu, K., & Wu, S. (2020). Fuzzy observer-based
control for the traction dual rectifiers in high-speed train. IEEE Transactions on
Vehicular Technology, 70 (1), 303–318.
Tasiu, I. A., Wang, H., Liu, Z., Zhang, Y., Zhang, Q., Meng, X., & Zhao, W. (2023).
Robust fuzzy stabilization control for the traction converters in high-speed train.
Control Engineering Practice, 132, paper 105423.
Tian, Y., Wang, B., Chen, P., & Yang, Y. (2021). Finite-time Takagi–Sugeno fuzzy
controller design for hydraulic turbine governing systems with mechanical time
delays. Renewable Energy, 173, 614–624.
Tooranjipour, P., & Vatankhah, R. (2018). Adaptive critic-based quaternion neuro-
fuzzy controller design with application to chaos control. Applied Soft
Computing, 70, 622–632.
Tuan, H. D., Apkarian, P., Narikiyo, T., & Yamamoto, Y. (2001). Parameterized linear
matrix inequality techniques in fuzzy control system design. IEEE Transactions
on Fuzzy Systems, 9 (2), 324–332.
Unnithan, A. R. R., & Subramaniam, S. (2022). Enhancing ride comfort and stability of
a large van using an improved semi-active stability augmentation system. SAE
International Journal of Vehicle Dynamics, Stability, and NVH, 6(10-06-04-
0026).
Valdez, F., Castillo, O., Cortes-Antonio, P., & Melin, P. (2020). A survey of type-2
fuzzy logic controller design using nature inspired optimization. Journal of
Intelligent & Fuzzy Systems, 39 (5), 6169–6179.
Vermeiren, L., Dequidt, A., Afroun, M., & Guerra, T. M. (2012). Motion control of
planar parallel robot using the fuzzy descriptor system approach. ISA
Transactions, 51 (5), 596–608.
Viadero-Monasterio, F., Nguyen, A.-T., Lauber, J., Boada, M. J. L., & Boada, B. L.
(2023). Event-triggered robust path tracking control considering roll stability
under network-induced delays for autonomous vehicles. IEEE Transactions on
Intelligent Transportation Systems, DOI: 10.1109/TITS.2023.3321415.
Wang, H. O., Tanaka, K., & Griffin, M. F. (1996). An approach to fuzzy control of
nonlinear systems: Stability and design issues. IEEE Transactions on Fuzzy
Systems, 4 (1), 14−23.
Wang, I.-J., & Spall, J. C. (2008). Stochastic optimisation with inequality constraints
using simultaneous perturbations and penalty functions. International Journal of
Control, 81 (8), 1232–1238.
Wang, L.-X. (1993). Stable Adaptive Fuzzy Control of Nonlinear Systems. IEEE
Transactions on Fuzzy Systems, 1 (2), 146–155.
Wang, Y.-J., Wang, Z.-D., Zou, L., Chen, Y., & Yue, D. (2023). Distributed
proportional-integral fuzzy state estimation over sensor networks under energy-
constrained denial-of-service attacks. IEEE Transactions on Cybernetics, DOI:
10.1109/TCYB.2023.3288829.
Wang, Y.-J., Wang, Z.-D., Zou, L., & Dong, H.-L. (2022). Nonfragile dissipative fuzzy
PID control with mixed fading measurements. IEEE Transactions on Fuzzy
Systems, 30 (11), 5019–5033.
Wang, Y.-J., Wang, Z.-D., Zou, L., Ma, L.-F., & Dong, H.-L. (2023). Ultimately
bounded PID control for T-S fuzzy systems under FlexRay communication
protocol. IEEE Transactions on Fuzzy Systems, DOI:
10.1109/TFUZZ.2023.3282044.
Wang, Y. L., Han, Q. L., Fei, M. R., & Peng, C. (2018a). Network-based T–S fuzzy
dynamic positioning controller design for unmanned marine vehicles. IEEE
Transactions on Cybernetics, 48 (9), 2750–2763.
Wang, Z., & Fei, J. (2022). Fractional-order terminal sliding-mode control using self-
evolving recurrent Chebyshev fuzzy neural network for MEMS gyroscope.
IEEE Transactions on Fuzzy Systems, 30 (7), 2747–2758.
Wang, Z.-S., Yang, L., Liang, Y., Liu, Q., Wang, Z.-Q., & He, D.-K. (2018b). Data-
driven particle quality control of spray fluidized bed granulation process.
Proceedings of 2018 Chinese Automation Congress, Xi’an, China, 3677–3611.
Wen, S., Hu, X., Lv, X., Wang, Z., & Peng, Y. (2019). Q-learning trajectory planning
based on Takagi–Sugeno fuzzy parallel distributed compensation structure of
humanoid manipulator. International Journal of Advanced Robotic Systems, 16
(1), paper 1729881419830204.
Wu, D.-R. (2012). On the fundamental differences between interval type-2 and type-1
fuzzy logic controllers. IEEE Transactions on Fuzzy Systems, 20 (5), 832−848.
Xie, X., Lu, J., & Yue, D. (2022a). Resilient stabilization of discrete-time Takagi-
Sugeno fuzzy systems: Dynamic trade-off between conservatism and
complexity. Information Sciences, 582, 181–197.
Xie, X., Ma, H., Zhao, Y., Ding, D. W., & Wang, Y. (2012). Control synthesis of
discrete-time T–S fuzzy systems based on a novel non-PDC control scheme.
IEEE Transactions on Fuzzy Systems, 21 (1), 147–157.
Xie, X., Yang, F., Wan, L., Xia, J., & Shi, K. (2022). Enhanced local stabilization of
constrained N-TS fuzzy systems with lighter computational burden. IEEE
Transactions on Fuzzy Systems, 31 (3), 1064–1070.
Yam, Y., Baranyi, P., & Yang, C. T. (1999). Reduction of fuzzy rule base via singular
value decomposition. IEEE Transactions on Fuzzy Systems, 7 (2), 120–132.
Yan, X.-Y., Guo, P.-J., Zhang, S.-X., Yan, X.-L., & Liu, C. (2020). Static pressure set-
point reset smoothing scheme based-on improved iterative learning control and
variable trajectory for air-conditioning system. Journal of Process Control, 91,
63–71.
Yin, L.-F., & Li, Y. (2022). Fuzzy vector reinforcement learning algorithm for
generation control of power systems considering flywheel energy storage.
Applied Soft Computing, 125, paper 109149.
Yin, T.-K., & Lee, C. S. G. (1995). Fuzzy model-reference adaptive control. IEEE
Transactions on Systems, Man, and Cybernetics, 25 (12), 1606–1615.
Yu, Q.-X., & Hou, Z.-S. (2021). Adaptive fuzzy iterative learning control for high-
speed trains with both randomly varying operation lengths and system
constraints. IEEE Transactions on Fuzzy Systems, 29 (8), 2408–2418.
Yu, W., Wang, R., Bu, X.-H., & Hou, Z.-S. (2020). Model free adaptive control for a
class of nonlinear systems with fading measurements. Journal of The Franklin
Institute, 357 (12), 7743–7760.
Yuan, H., Nguyen, V., & Zhou, H. (2021). Research on semi-active air suspensions of
heavy trucks based on a combination of machine learning and optimal fuzzy
control. SAE International Journal of Vehicle Dynamics, Stability, and NVH, 5
(2), 159−172.
Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8 (3), 338−353.
Zdešar, A., Dovžan, D., & Škrjanc, I. (2014). Self-tuning of 2 DOF control based on
evolving fuzzy model. Applied Soft Computing, 19, 403–418.
Zhang, C., Lam, H. K., Qiu, J., Liu, C., & Chen, Q. (2018). A new design of
membership-function-dependent controller for TS fuzzy systems under
imperfect premise matching. IEEE Transactions on Fuzzy Systems, 27 (7),
1428–1440.
Zhang, D., Ye, Z., Feng, G., & Li, H. (2021). Intelligent event-based fuzzy dynamic
positioning control of nonlinear unmanned marine vehicles under DoS attack.
IEEE Transactions on Cybernetics, 52 (12), 13486–13499.
Zhang, H., & Xie, X. (2011). Relaxed stability conditions for continuous-time T–S
fuzzy-control systems via augmented multi-indexed matrix approach. IEEE
Transactions on Fuzzy Systems, 19 (3), 478–492.
Zhang, H.-G., Zhang, K., Cai, Y.-L., & Han, J. (2019). Adaptive fuzzy fault-tolerant
tracking control for partially unknown systems with actuator faults via integral
reinforcement learning method. IEEE Transactions on Fuzzy Systems, 27 (10),
1986–1998.
Zhang, K., Zhang, H.-G., Mu, Y.-F., & Sun, S.-X. (2019). Tracking control
optimization scheme for a class of partially unknown fuzzy systems by using
integral reinforcement learning architecture. Applied Mathematics and
Computation, 395, 344–356.
Zhang, L., Hu, M., Zhang, H., Bian, Y., Nguyen, A.-T., & Ding, R. (2023). Channel-
level event-triggered communication scheme for path tracking control of
autonomous ground vehicles with distributed sensors. IEEE Transactions on
Vehicular Technology, 72 (10), 12553–12566.
Zhang, X., Yao, S., Xing, W., & Feng, Z. (2022). Fuzzy event-triggered sliding mode
depth control of unmanned underwater vehicles. Ocean Engineering, 266, paper
112725.
Zhang, Z., Zhang, Z., & Zhang, H. (2017). Distributed attitude control for
multispacecraft via Takagi–Sugeno fuzzy approach. IEEE Transactions on
Aerospace and Electronic Systems, 54 (2), 642–654.
Zheng, H., Xie, W. B., Nguyen, A.-T., & Qu, D. (2023). A model reconstruction
approach for control synthesis of Takagi-Sugeno fuzzy systems. Fuzzy Sets and
Systems, 469, paper 108640.
Zhu, X., & Li, W. (2019). Takagi–Sugeno fuzzy model-based shaft torque estimation
for integrated motor–transmission system. ISA Transactions, 93, 14–22.
Zou, L., Wang, Z.-D., Shen, B., Dong, H.-L., & Lu, G.-P. (2023). Encrypted finite-
horizon energy-to-peak state estimation for time-varying systems under
eavesdropping attacks: Tackling secrecy capacity. IEEE/CAA Journal of
Automatica Sinica, 10 (4), pp. 985–996.