Boost Converter PSO PID
Boost Converter PSO PID
Boost Converter PSO PID
(2)
ISSN Print (2314 7318) and Online (2314 730X) April 2016
I. INTRODUCTION
Boost converters are employed when the required output Recently the Genetic Algorithm (GA) [20], Ant Colony
voltage must be greater than its input voltage. Such converters Optimization (ACO) [21] and Particle Swarm Optimization
are largely used in battery charging, solar power applications, (PSO) [22-26] based approaches were also explored for the
fuel cell power converting systems, and battery driven closed-loop controller design of boost type DC-DC converters.
vehicles [1-3]. These converters, when operated in the open- Controllers designed through such techniques have been
loop mode, exhibit poor voltage regulation and unsatisfactory shown to guarantee high dynamic and static performance over
dynamic response. Hence, these converters are generally a wide range of operating points.
provided with a closed-loop controller to achieve good In this paper, a novel optimization model based on the
transient response and output voltage regulation [4]. The mode Particle Swarm Optimization (PSO) algorithm for the closed-
of operation of the converter varies from ON- to OFF-state of loop controller design of a converter is designed to obtain the
the main power switch and, traditionally, small-signal static and dynamic characteristics at all operating points. To
linearization techniques have been employed for controller test the methodology, the small-signal model of a boost type
design. DC-DC converter is used to evaluate the objective function to
Several conventional methods of closed-loop controller yield a robust closed-loop controller structure that is subjected
design for boost type DC-DC converters have been proposed to internal and external disturbances. Design equations are
to obtain the desired output voltage in the literature [5-8]. In derived and the system is modeled in MATLAB. Extensive
all of these works, the linearized transfer function model of a simulation tests are carried out with linear controller
boost type DC-DC converter is employed for the closed-loop parameters and the results are presented.
controller design. Linear PID and PI controllers are usually The rest of this paper is presented as follows: The
employed with boost type DC-DC converters, where these discussion on the small-signal averaged state-space modelling
controllers are designed using standard frequency response technique for DC-DC boost converters, with and without
techniques based on the small-signal model of the converter. parasitic elements, operating in continuous conduction mode
The design is based on linear control theory such as Ziegler- (CCM) is presented in Section II. Next, an overview of the
Nichol method [9], Root Locus technique [10], Circle Based PSO method and its application in boost converter controller
criterion [11], Hysteresis method [12], Bode plot, etc. These design is presented in Section III. Section IV presents the
control strategies that are based on the linearized small-signal simulation results and analysis of the proposed PSO based
model of the converter generally have good performance boost converter controller design, and finally the last section
around the operating point. However, a boost converters presents the conclusions.
small-signal model changes when its operating point changes.
The poles and the right-half-plane zero, as well as the II. SMALL SIGNAL AVERAGED MODEL FOR BOOST
magnitude of the frequency response, are all dependent on the CONVERTER
operating point. Thus, when the operating point changes, the The small-signal averaged state-space technique is a general
fixed gain controller designed at one typical operating point analysis tool which can be applied to either simple or complex
through conventional techniques does not guarantee a circuits [27, 28]. The linear averaged time-invariant models,
satisfactory dynamic response at other operating points. created by using this technique, are relatively simple, but need
Furthermore, DC-DC converters are switched mode circuits; a lot of mathematical computational effort to find the final
hence, the mode of operation of these converters varies from results. To get such models, the detailed generalized procedure
the ON- to the OFF-state of the power switch, and therefore, it explained in [27, 28] is applied to our case.
is difficult for the PID controller to respond well to changes in For modelling with the state-space technique, the desired
the operating point, and it often exhibits a poor performance state variables for any electric circuit are the energy storing
when the system is subjected to large load variations. elements (i.e. inductor current iL and capacitance voltage vC).
Thus, the control of a boost type DC-DC converter has been To apply the state-space technique to any complex circuit, it
a challenging task. The underlying fact of all of the above must first be converted into a piece-wise linearized simple
works is that the boost converter dynamics are strongly non- circuit where the circuit laws can be easily applied.
linear and the controller designed at a specific operating point
A. Case 1: Ideal Boost Converter
often fails to perform satisfactorily over a wide range of
operating points. The past few decades have witnessed a The boost converter is a non-isolated power converter, also
number of contributions for the controller design of the DC- known as a step-up power converter where the required output
DC converters [13-19]. The improved methods of controller voltage is greater than its input voltage. The output current is
design reported in the literature are Predictive Control [13], continuous because the output diode becomes forward biased
Sliding Mode Control [14], Adaptive Control [15], Passivity only during a portion of the switching cycle and the output
Based Control [16], H Control [17], Robust Control [18], capacitor gives the whole load current during the other portion
Nominal and Tolerance Design [19], etc. of the switching cycle.
State-space averaging techniques are employed to get a set B. Case 2: Parasitic Realization in Boost Converter
of equations that describe the system over one switching
cycle. After applying averaging technique to (5) and (6), the Due to the different difficulties found during the modelling
following expression can be found: procedure of the boost converter systems, parasitic elements
such as conduction voltages, conduction resistances, inductor
DC resistance and equivalent series resistances (ESR) of
capacitors are ignored. The idea of considering ideal/lossless
components and leaving out parasitic elements like we have
Thus, with the help of state-space equations, values of
done earlier is to simplify model development procedure and
matrices A1, B1, C1, D1 (state parameters of ON-state) and A2,
to understand the basic behavior of the switching system.
B2, C2, D2 (state parameters of OFF-state) are extracted and A,
However, the effects of parasitic elements and losses are
B, C, D state parameters can be obtained as follows:
important for improving model accuracy, studying efficiency
. Similarly B, C and D and dynamic performance of the system. The problem with
parameters are also obtained. including the parasitic elements is that they lead to nonlinear
current and voltage waveforms and hence result in
Thus, state-space average model for DC-DC boost complications in the modelling development procedure.
converter in CCM would be as follows: Fig. 5 shows a simplified equivalent circuit of the DC-DC
boost converter with parasitic elements. Inductor L and
capacitor C can jointly be considered as an output filter. The
capacitor equivalent series resistance (ESR), RC, and the
inductor DC resistance, RL, are included in the analysis.
Resistor R represents the load seen by the power supply.
can be described as , Stopping criteria: these are the conditions under which the
where xs are the optimized parameters and is the search process will terminate. In this study, the search will
position of the jth particle with respect to the kth dimension, terminate if one of the following criteria is satisfied:
i.e. the value of the kth optimized parameter in the jth a) The number of iterations since the last change of the
candidate solution. best solution is greater than a pre-specified number or
b) The number of iterations reaches the maximum
allowable number.
Population, : It is a set of n particles at time t,
Fig.8 illustrates the flowchart of the PSO method.
i.e. .
Swarm: It is an apparently disorganized population of In this work, the PSO algorithm has been developed as
moving particles that tend to cluster together while each follows:
particle seems to be moving in a random direction [24]. An annealing procedure has been incorporated in order to
make uniform search in the initial stages and very local
Particle velocity, : It is the velocity of the moving search in the later stages. A decrement function for
particles represented by an m-dimensional vector. At time t, decreasing the inertia weight given as ,
the jth particle velocity can be described where is a decrement constant smaller than but close to
1, is proposed in this work.
as , where is the Feasibility checks procedure of the particle positions has
velocity component of the jth particle with respect to the kth been imposed after the position updating to prevent the
dimension. particles from flying outside the feasible search space.
The particle velocity in the kth dimension is limited by
Inertia weight, : It is a control parameter that is used some maximum value, his limit enhances the
to control the impact of the previous velocities on the current local exploration of the problem space and it realistically
velocity. Hence, it influences the trade-off between the global simulates the incremental changes of human learning [23].
and local exploration abilities of the particles [24]. For initial
stages of the search process, large inertia weight to enhance
the global exploration is recommended while, for last stages,
the inertia weight is reduced for better local exploration.
For comparison, the controller design is also carried out Fig.12 Closed loop response of DC-DC boost converter
using standard PSO. The controller constants obtained through operating in CCM with 13% sudden load disturbance
standard PSO are listed below:
KP = 1.325, Ki = 15.8851, Kd = 0.000489
The computed closed loop response of the DC-DC boost
converter operating in CCM obtained by using traditional
method of designing a PID controller and PSO algorithm with
small signal model is given in Fig.10.
ZN-PID PSO-PID
Test Cases Transient Parameters
Controller Controller
Rising time (ms) 0.50 0.45
Starting case 100% Step Change Settling time (ms) 4.71 1.22
Peak Overshoot (%) 28.75 0.00
Settling time (ms) 3.34 0.25
-10% Step Change
Peak Overshoot (%) 12.15 0.00
SMALL Source Voltage Disturbance
Settling time (ms) 3.53 0.23
+10% Step Change
Peak Overshoot (%) 12.15 0.00
Settling time (ms) 3.84 1.22
-13% Step Change
Peak Overshoot (%) 7.51 1.53
SMALL Load Resistance Disturbance
Settling time (ms) 4.03 2.14
+13% Step Change
Peak Overshoot (%) 7.51 2.52
Settling time (ms) 6.55 0.25
-25% Step Change
Peak Overshoot (%) 15.02 0.00
LARGE Source Voltage Disturbance
Settling time (ms) 6.75 0.24
+25% Step Change
Peak Overshoot (%) 15.02 0.00
Settling time (ms) 7.24 1.25
-35% Step Change
Peak Overshoot (%) 12.53 1.38
LARGE Load Resistance Disturbance
Settling time (ms) 7.22 1.65
+35% Step Change
Peak Overshoot (%) 12.53 2.23
16, No. 6, pp. 752763, 2001. Transmission. His research focuses on the monitoring, control,
[21] Zhang, J., Chung, H. S.-H., Lo, A. W.-L., and Huang, T., and protection of power systems and on the integration of
Extended ant colony optimization algorithm for power renewable energy systems into the smart grid.
electronic circuit design, IEEE Transactions on Power Dr. Sood is a Registered Professional Engineer in the
Electronics, Vol. 24, No. 1, pp. 147162, 2009. province of Ontario. He is a Fellow of the IEEE, Engineering
[22] Kennedy, J., and Eberhart, R., Particle swarm Institute of Canada and the Canadian Academy of
optimization, IEEE International Conference on Neural Engineering. He served previously as a Director of the IEEE
Network., Vol. 4, pp. 19421948, 1995. Canadian Foundation and is a former Editor of the IEEE
[23] Kennedy, J., "The particle swarm: social adaptation of Transactions on Power Delivery, and Co-editor of the IEEE
knowledge, IEEE International Conference on Evolutionary Canadian Journal of Electrical and Computer Engineering.
Computation, pp.303-308, 13-16 Apr 1997.
[24] Shi, Yuhui, and Russell C. Eberhart. "Parameter selection
in particle swarm optimization." In Evolutionary programming
VII, pp. 591-600. Springer Berlin Heidelberg, 1998.
[25] Eberhart, R. C., and Shi, Y., Guest editorial, IEEE
Transaction on Evolutionary Computation (Special Issue on
Particle Swarm Optimization), Vol. 8, No. 3, pp. 201203,
June 2004.
[26] Kennedy, J., The behavior of particles, Proceedings of
the 7th Annual Conference on Evolutionary Programming, pp.
581589, San Diego, CA, 1998.
[27] R. W. Erickson, Fundamentals of Power Electronics, 1st
ed., New York: Chapman and Hall, 1997.
[28] J. G. Kassakian, M. F. Schlecht, and G. C. Verghese,
Principles of Power Electronics, 1st ed., Reading,
Massachusetts: Addison-Wesley, 1991.
[29] Viorel Nicolau, On PID Controller Design by
Combining Pole Placement Technique with Symmetrical
Optimum Criterion, Mathematical Problems in Engineering,
Vol. 2013, Article ID 316827, 8 pages, 2013.