Maximizing Energy Efficiency in Drones Through Accurate State of Charge Estimation Using Extended Kalman Filter
Maximizing Energy Efficiency in Drones Through Accurate State of Charge Estimation Using Extended Kalman Filter
Maximizing Energy Efficiency in Drones Through Accurate State of Charge Estimation Using Extended Kalman Filter
Corresponding Author:
Kamal Anoune
LMIET Laboratory, Physics Department, Faculty of Sciences and Technics FST-Settat
Hassan 1st University
Settat, Morocco
Email: [email protected], [email protected]
1. INTRODUCTION
In recent years, drones have emerged as one of the most extensively researched logistics technologies,
incorporating technical elements that align with modern transportation and societal developments, including
autonomy, adaptability, and agility. The first factor influencing a drone's battery life is the payload weight.
Heavier payloads require more energy to fly, resulting in shorter battery life. For instance, a few kilograms of
payload can significantly reduce the drone's flight time compared to flying without any payload. Additionally,
wind speed affects battery life, as flying against the wind consumes more energy while flying with the wind can
extend flight time. Temperature also plays a crucial role; high temperatures can lead to battery overheating, and
low temperatures reduce battery capacity and overall performance. Furthermore, flight height influences battery
life, as maintaining stability and continuous flight demands more energy, resulting in shorter battery life.
Considering these aspects becomes critical when planning drone operations, especially with payloads, to ensure
the drone has sufficient battery life to complete tasks safely and effectively. Drones find extensive use in remote
guidance and encompass a variety of vehicles, including submarines and land-based autonomous vehicles.
Notably, hybrid-wing drones, a recent addition to the market, possess fixed and rotary-wing capabilities,
enabling rapid destination reach through gliding and utilizing four rotors for hovering [1], [2]. Achieving
operational drones requires addressing three significant challenges through automation research: vehicle design,
positioning and routing, and vehicle coordination. The vehicle design aspect involves creating equipment that is
not only efficient but also capable of hovering and adapting to various scenarios while maintaining reliability
comparable to commercial airliners. This ambitious task demands numerous iterations, necessitating the
collaboration and contributions of experts from diverse fields to bring forth innovative solutions [3]. Numerous
studies reveal that drones have some negative limitation related to estimation of disposable energy. Kirschstein
et al. [4] introduces an energy usage model tailored for drones, aiming to delineate the energy needs for drone
deliveries based on environmental conditions and flight patterns. The model is subsequently employed to gauge
the energy consumption of a stationary package delivery system, which operates from a designated depot and
serves a specific number of clients. A comparative analysis is conducted between drone energy consumption
and the energy requirements of diesel and electric trucks that cater to the same clients from a similar depot. To
bolster the accuracy and dependability of SoC estimation, the dynamic attributes of a lithium-ion battery (LiB)
are approximated through an auto-regressive and moving average model. This model effectively compensates
for potential discrepancies stemming from voltage and discharge current measurements, thereby elevating the
precision and reliability of state of charge (SoC) estimation [5].
Gaining profound insights into LiB performance and its underlying instrument offers valuable
knowledge. Such understanding facilitates battery performance testing, enabling the identification of several
factors that impact performance and the governing laws behind their influence. To develop practical battery
system models for battery management systems (BMS) [6], modeling methods can be employed. These models
provide satisfactory accuracy while minimizing complex computations. During operation, adaptive control
technology is utilized to identify battery system parameters, estimate critical battery states such as SoC, state
of health (SoH), and state of function (SoF), and detect faults. This information is then communicated to the
vehicle manager over network, ensuring the vehicles' safe and dependable operation [7]. In their research,
Xia et al. [8] conducted a study focused on the routing problem of drones featuring load-dependent
characteristics. The researchers introduced docking hubs as collaborative facilities for trucks and drones,
effectively expanding the service coverage. To handle the complexities arising from nonlinear load-dependent
energy consumption, they proposed a mixed-integer model. While predicting the operational range for drones
has not been a significant challenge due to fast and readily available refueling options, accurately estimating
the driving or flying range for battery-operated vehicles has become crucial. This increased importance arises
from the widespread adoption of battery-operated vehicles in countless areas [9]. Addressing energy
consumption becomes a fundamental constraint in realm of drone operations, unlocking full possible of
achieving maximum range, cost reduction, and accurate SoC estimation. Optimizing energy usage directly
influences drones' operational range, determining the distance they can cover and the payload they can carry.
To fully harness the benefits of drones, it is imperative to develop efficient energy management strategies that
maximize range while minimizing power consumption. These strategies necessitate considering numerous
factors, including flight dynamics, payload weight, wind conditions, and operational requirements. Developing
an accurate energy estimation algorithm and integrating it into operational planning enable drones to extend
their operational range, facilitating rapid and environmentally friendly operations. Ultimately, this contributes
to advancing sustainable and efficient last-mile logistics solutions.
The optimization of energy consumption plays a critical role in determining the operational range and
payload capacity of drones. To fully harness the potential benefits of drones, efficient energy management
strategies must be developed to maximize range while minimizing power usage. This challenge also extends
to aerial drones, where precise flight range planning is essential to ensure continuous service and prevent
battery depletion during flights. Accurate estimation of the SoC through an effective BMS is crucial for reliable
power usage modeling. However, existing comprehensive drone models may not precisely align with the actual
energy strained from LiB due to its non-idealities. To address these issues, this research presents a novel
contribution focused on predicting and optimizing drone range. The proposed approach offers flexibility in
accommodating varying levels of accuracy and complexity in both drone and battery models, resulting in
improved range estimation and planning capabilities. Additionally, drones are often hailed as energy-efficient
transportation options, given their battery-powered operation. However, to ensure an accurate representation
of energy requirements for specific drone applications, we have developed a MATLAB program based on
extended Kalman filter (EKF) to precisely estimate battery charge.
The paper follows a structured organization as follows: In the introduction, the problem statement is
highlighted, emphasizing the significance of energy consumption in drone operations. Moving to the system
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 757
modeling section, the focus shifts to the modeling of drones' energy consumption, offering insights into the
approach used for estimating energy usage in drones. Next, the paper delves into battery modeling, presenting
a mathematical model for characterizing battery behavior. Additionally, the utilization of the hybrid pulse
power characterization (HPPC) test and the recursive least square with forgetting factor (FFRLS) for parameter
identification in battery modeling is introduced. In nonlinear filter section, the implementation of EKF is
discussed, serving as a powerful tool for battery state estimation, and overcoming the limitations of linear
filters. Subsequently, an overview of BMS employed in the study is provided, highlighting its role in effectively
managing and optimizing battery performance. Finally, the paper concludes by summarizing the key points,
emphasizing the contributions made by the research, and discussing the broader implications of the study in
the field of drone energy consumption and battery management.
2. METHOD
The proposed approach for estimating the operational range of drones involves the gathering of both
drone and battery data. The drone data encompasses mechanical and electrical characteristics that influence
power consumption, such as motor power, weight, and aerodynamic drag. This data provides valuable insights
into the energy requirements of different drone types. On the other hand, battery data focuses on the electrical
properties of the battery cell and its scaling for the battery pack, enabling a comprehensive understanding of
the energy storage capabilities.
To estimate the drone's range, a battery model is formulated, taking into consideration power
conversion efficiency. By analyzing voltage and current waveforms, the model calculates the battery's SoC
and, consequently, the drone's range. The methodology is designed to be adaptable to various drone types and
also takes into account variables like weather conditions. It incorporates efficiency trade-offs and adjustments,
including speed control mechanisms, to optimize power consumption and maximize the drone's operational
range. This approach ensures a comprehensive estimation of the drone's range and provides insights for
enhancing flight performance and efficiency.
Thevenin's model offers a promising solution by incorporating a Rint model with a parallel RC circuit,
effectively addressing the limitations of the Rint model, which fails to capture the dynamic properties of Li-
Maximizing energy efficiency in drones through accurate state of charge … (Kamal Anoune)
758 ISSN: 2252-8792
ion batteries. As depicted in Figure 1, this model visualizes the terminal voltage as U L, the ohmic voltage as
UR, and the ohmic internal resistance as R0. The RC circuit includes a polarization resistor Rp and a polarization
capacitor Cp, effectively characterizing the polarization effect in Li-ion batteries, with the voltage across the
polarization element denoted as Up. By employing Kirchhoff's law, in (1) establishes the voltage and current
representations within the equivalent circuit, enabling an improved and more accurate estimation of the
battery's SoC. These advanced modelling technics own a big possibility for enhancing battery management
systems and optimizing battery performance in numerous applications, including electric vehicles and portable
electronic devices [12].
𝑈𝐿 = 𝑈𝑜𝑐 − 𝐼𝑅0 − 𝑈𝑝
{ ̇ 1 1 (1)
𝑈𝑝 = − 𝑈𝑝 + 𝐼
𝐶𝑝 𝑅𝑝 𝐶𝑝
2.1.2. Hybrid pulse power characterization (HPPC) method for parameter identification
The hybrid pulse power characterization (HPPC) method stands as a pivotal technique in
comprehensively characterizing LiB. It involves subjecting the battery to a series of hybrid pulse power
profiles, essentially a set of distinct current pulses with varying magnitudes and durations, while concurrently
measuring the battery's response in terms of voltage and current. This method operates under varying
conditions, encompassing different load levels, charging, and discharging cycles, to gather an expansive
dataset. HPPC conditions encompass a spectrum of scenarios, simulating real-world usage patterns by
integrating abrupt changes in load conditions, rapid charging, and discharging. Through this controlled yet
diverse dataset, HPPC facilitates the extraction of critical parameters that significantly influence LiB behavior.
Parameters obtained through HPPC encapsulate vital attributes such as internal resistance, capacity, and voltage
response at varying SoC. Method enables the derivation of critical battery model parameters like open circuit
voltage (OCV) versus SoC relationship, which forms the fundamental basis for predicting the battery's behavior
under different operating conditions. This method not only aids in enhancing battery model accuracy but also
plays a crucial role in predicting battery behavior and performance across a range of real-world scenarios.
The battery model is developed using parameters derived from characterization. Extracting these
parameters requires a series of characterization tests, following established standards and research
methodologies [13]. Initially, the capacity test determines the battery's effective nominal capacity through a
standard charge and discharge test. The second OCV test, is conducted to extract the OCV-SoC
relationship and battery model parameters. This approach deviates from employing the hybrid pulse power
characterization (HPPC) test, as the study exclusively focuses on the standard C-rate. Baccouche et al.'s
research [14] emphasizes accurate modeling of the nonlinear OCV-SoC relationship crucial for adaptive Li-ion
battery operation. This model, employing five parameters within double exponential and quadratic functions,
closely aligns with experimental curves, boasting a mere 1 mV fitting error. It covers wide temperature ranges,
accounts for OCV voltage hysteresis, and when integrated into EKF for SoC estimation, significantly
minimizes execution time and reduces estimation error to 3%, outperforming other models at 5%. Rigorous
experiments validate the model's reliability and precision across diverse loads and temperatures.
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 759
Figure 2 illustrates the experimental OCV discharge test [14]. The test involves initially fully charging
the battery, followed by discharging current pulses equivalent to 5% of the SoC step. This is interspersed with
a 30-minute rest time to ensure complete cell discharge and stabilize the OCV at a consistent value. Extracting
OCV points at different SoC levels is essential to construct the analytical model representing the mapping
between OCV and SoC
HPPC test involves several key steps. Firstly, it is employed to establish the (OCV-SoC) connection,
and it also serves to identify parameter values from ECM derived through the offline parameter identification
method [15]. The HPPC test commences by placing the battery cell in at 25 °C for four hours. Subsequently,
constant current of 1C is applied to the cell until it reaches a voltage of 4.2 V, which is then maintained until
the current decreases to ≤0.05C. After 1 hour break, cell is discharged with a 1C current until it reaches a SoC
of 90%. Following another 1-hour break, the cell is further discharged with a 3C current for 10 seconds,
followed by a 30-second rest period. Lastly, the cell is loaded with a 2.25C current for 10 seconds. This
sequence is repeated for various SoC values, ranging from 80% to 10%. Subsequently, a sextic polynomial (2)
is employed to accurately fit the relationship where 𝑘0 ~𝑘6 are the constants.
By utilizing the HPPC test in conjunction with advanced fitting algorithms, the battery's parameters can be
accurately determined in real-time, providing valuable insights for battery management and performance
optimization in practical applications.
𝑅𝑝
𝑈𝑡 (𝑠) − 𝑈𝑜𝑐 (𝑠) = 𝐼𝑡 (𝑠) (𝑅0 + ) (3)
1+𝑅𝑝 𝐶𝑝 𝑠
where s denotes the frequency operator. By introducing 𝐸𝑡 (𝑠) = 𝑈𝑡 (𝑠) − 𝑈𝑜𝑐 (𝑠) transfer function can be
represented as in (4):
𝑈(𝑠)−𝑈𝑜𝑐 (𝑠) 𝑅𝑝
𝐺(𝑠) = = (𝑅0 + ) (4)
𝐼(𝑠) 1+𝑅𝑝 𝐶𝑝 𝑠
to convert the transfer function into a discrete form, this paper employs the commonly used bilinear
transformation, with the formulation provided in (5).
2 1−𝑧 −1
𝑠= (1)
𝑇𝑠 1+𝑧 −1
Where z is the discretization operator and Ts is set to 1 in this paper. Moreover, the discrete form of equation
(G(s)) can be expressed as in (6).
𝑎2 −𝑎3 𝑧 −1
𝐺(𝑧 −1 ) = (6)
1+𝑎1 𝑧 −1
(mp +mv )v
PCons = + Pelec (9)
370ηr
The power consumption in kW can be calculated using the following approximation, as shown in (9).
Furthermore, a study by Anderea provides parameter values, where the cruising velocity has a direct impact on
the power consumption of the drones (see Table. 1). Figure 3 illustrates the power consumption (P_Cons) for
various combinations of payload mass and cruising velocity using the given input parameters. The 3D bar plot
visually represents power consumption trends across different payload masses and cruising velocities. The
overview of the data indicates the size of the matrix and a minimum, maximum, mean, and standard deviation
of power consumption. The optimal operating point is identified as the payload mass and cruising velocity
combination with the lowest power consumption, while the worst operating point is the combination with the
highest power consumption. The 3D bar plot helps visualize how power consumption changes with varying
payload masses and cruising velocities, providing valuable insights into power consumption behavior.
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 761
The maximum range, represented by "d" and measured in kilometers, is influenced by the head wind
factor (HWF), which signifies the ratio of headwind to airspeed. To illustrate with numerical examples: Let's
assume a maximum range varying from 2 to 25 km, an airspeed range from 0 to 45 km/h, and a headwind of
30 km/h. The graph in Figure 4 illustrates multiple lines representing the worst-case power consumption for an
aerial vehicle system, considering different combinations of cruising velocity and maximum range. The results
demonstrate that power consumption increases with higher cruising velocities and maximum ranges,
attributed to the heightened energy requirements for propulsion and electronics. For example, at a cruising
velocity of 30 km/h and a maximum range of 10 km, the worst-case power consumption is approximately 8.5
kW. As the cruising velocity increases to 45 km/h and the maximum range extends to 15 km, the power
consumption peaks at around 12.3 kW. This analysis emphasizes the significance of optimizing flight
parameters to minimize power demands and progress efficiency and endurance of the aerial vehicle system.
Additionally, the power consumption trend reveals that higher cruising velocities generally lead to increased
power consumption due to greater air resistance. The sensitivity to the head wind factor (HWF), which is
assumed as 2/3 of the headwind-to-airspeed ratio, is not explicitly depicted in the graph; however, it is
considered a constant factor in the power consumption equation. Next, let's consider the economic aspect:
Energy cost per kilometer compared to cruising velocity.
An estimation of the average energy cost per kilometer can be represented as 𝐶𝑒𝑙𝑒𝑐 is the cost of electricity à
0.12 $/kW h and 𝐶ℎ𝑎𝑟𝑒𝑓𝑓 is charging efficiency (about 0.8).
Figure 5 illustrates the energy cost per kilometer (E_Cost) for an aerial vehicle system, considering
various cruising velocities within the range of 0 to 45 km/h. The results show that as the cruising velocity
increases, the energy cost per kilometer generally decreases. For instance, at a low cruising velocity of 5 km/h,
the energy cost is approximately $0.022 per kilometer. However, as the cruising velocity rises to 30 km/h, the
energy cost reduces to around $0.008 per kilometer. This inverse relationship is due to higher cruising velocities
resulting in better energy efficiency and lower energy consumption per kilometer. The analysis emphasizes the
economic benefits of optimizing cruising velocities for cost-effective operations, wherein higher speeds can
lead to significant savings in energy expenses per kilometer traveled [8], [20].
Maximizing energy efficiency in drones through accurate state of charge … (Kamal Anoune)
762 ISSN: 2252-8792
The expression of discrete state-space model equation, which incorporates system's dynamics, takes
the following form:
𝑥𝑘+1 = 𝑓(𝑥𝑘 , 𝑢𝑘 ) + 𝑑𝑘
(13)
𝑦𝑘 = 𝑔(𝑥𝑘 , 𝑢𝑘 ) + 𝑠𝑘
Incorporating the nonlinear state transition function 𝑓(𝑥𝑘 , 𝑢𝑘 ) and measurement function 𝑔(𝑥𝑘 , 𝑢𝑘 ), where x_k
represents state variables and u_k denotes input variables, the state equation can be formulated. Additionally,
considering measurement noises, we can further express in (14).
Upon considering the discussed linearization process, it becomes evident that EKF effectively overcomes
limitations of linear Kalman filter (LKF) by integrating the nonlinear system model into both the state
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 763
prediction and correction steps. EKF effectively utilizes nonlinear LiB modeling to predict both the system
state and output. At each time step 𝑘, the nonlinear LiB modelling is linearized based on the predicted state
𝑥̂_𝑥̂𝑘− to derive the matrices 𝐴̂𝑘 , 𝐵̂𝑘 , and 𝐶̂𝑘 . These matrices are crucial in calculating and updating the
covariance matrix of the state estimation errors, as well as in determining the Kalman gain.
In essence, the EKF overcomes the challenges of linearization encountered in LKF by skillfully
incorporating the nonlinear battery model. By doing so, it ensures accurate state estimation and improved
overall performance. This integration of nonlinear battery models in the EKF enables the filter to handle the
complexities and variations present in real-world battery systems, providing reliable and precise state
estimations even in dynamic and challenging conditions. Thus, the EKF stands as a powerful tool for state
estimation in BMS, enhancing performance and reliability of LiBs applications [26], [27]. Even though EKF
estimation is utilized for SoC estimation, initial values of Kalman parameters are strongminded in following
manner:
−1
𝑃0 = [1𝑒 0 ], 𝑄 = [2𝑒 −8 0 ] and 𝑅 = 2𝑒 −1 (15)
0 1𝑒 −1 0 5𝑒 −3
Figure 6. Evaluation of present load profile and Ampere-hours SoC for examined drones [28]
Maximizing energy efficiency in drones through accurate state of charge … (Kamal Anoune)
764 ISSN: 2252-8792
The proposed SoC estimation model comprises three essential components, each component is
necessary for estimation process. First part is data input step, where the input data is initialized. To achieve
this, the model utilizes the FFRLS algorithm and conducts the HPPC test. Through this process, the initial
values of system variables 𝑘0 ~𝑘6 and parameters (𝑅0 , 𝑅𝑝 , 𝐶𝑝 ) are carefully determined. These values form the
fundamental basis for predicting the load state, thereby ensuring accurate and reliable SoC estimation. Next,
the HPPC algorithm is employed to establish the correlation between OCV and SoC of LiBs. The HPPC test
involves subjecting the battery to a series of hybrid pulse power profiles, which generates valuable data about
the battery's behavior. This resulting data is then fitted using a polynomial function of order 6. By fitting the
data with a polynomial equation, the model accurately models the rapport between LiBs OCV and its SoC.
This information is crucial to predict battery's behavior and facilitates the prediction of its performance under
different load conditions. The coefficients 𝑘0~6 , as shown in Table 2, are derived from this fitting process [29].
The system parameter identification is used to obtain model parameter according to the OCV-SoC
curve. The result of parameters based on the FFRLS algorithm is given in Table 3. The graph labeled as
Figure 7 illustrates three distinct curves portraying the SoC of the studied system. The first curve represents
the observed SoC, showcasing actual measured SoC values and serving as the reference for comparison. The
second curve, labeled as SoC extended Kalman filter (SoC_EKF), closely follows the observed SoC curve,
indicating a highly accurate SoC approximation. SoC_EKF employs advanced estimation techniques to
improve the precision. On the other hand, the third curve represents the SoC Amper hours (SoC_AH). While
the SoC_AH curve provides an acceptable estimation of the SoC, it falls slightly short of achieving the same
level of accuracy as the SoC_EKF. The SoC_AH relies on Amper hours techniques to estimate the SoC, which
may introduce some degree of error during the estimation process.
As time progresses, it becomes apparent that the differences between the observed SoC and the
estimated SoCs increase significantly. This trend indicates a deterioration in the accuracy of both the SoC_EKF
and the SoC_AH over time. Several factors could contribute to these increasing errors, including battery aging,
measurement uncertainties, and limitations in the estimation algorithms employed.
According to Figure 8, the EKF method provides an accurate estimation of the SoC, with a lower root
mean squared error (RMSE) value 0.78% making it superior for precise drone battery SoC estimation. These
findings contribute to enhanced drone performance, reliability, and overall safety. In conclusion, SoC_AH
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 765
offers an acceptable estimation of the SoC, but it is slightly less accurate compared to SoC_EKF. This is
because SoC_AH utilizes ampere-hours techniques for estimation, which introduces some degree of error
during the process. To measure the amount value of differences among the estimated SoC versus actual values,
the RMSE is employed, resulting in an RMSE value (0.78e-04) for SoC_EKF. The overall results underscore
a deal between accuracy and computational complexity in selecting an estimation technique for real-world
applications. The SoC_EKF method offers highly accurate estimations but may require more computational
resources, while the SoC_AH method provides a more practical and computationally efficient alternative with
an acceptable level of accuracy.
Figure 8. Errors between the observed SoC and the estimated SoCs
4. CONCLUSION
Addressing energy consumption in drone operations is paramount for maximizing range, reducing
costs, and accurately estimating the SoC. Optimizing energy usage is crucial in determining drones' operational
range and payload capacity, enabling them to cover greater distances and carry heavier loads. By integrating
precise energy estimation algorithms into operational planning, drones can extend their range and facilitate
rapid, eco-friendly missions. Recently, LIBs have gained significant attention due to their sustainable
development. Nevertheless, accurately measuring SoC of LIBs remains a challenge for ensuring their
safe operation.
In this paper, we delve into drones' energy consumption modeling and present a mathematical model
for their energy consumption and battery behavior. We introduce the HPPC test and FFRLS for parameter
identification. In the nonlinear filter section, we focus on implementing an accurate EKF for battery state
estimation, surpassing the limitations of linear filters. The EKF algorithm provides precise SoC data, enhancing
the BMS by enabling more accurate monitoring and optimization of battery performance. In our study, we
present our findings, along with a comprehensive analysis and results interpretation. We investigate the
accuracy of SoC estimation techniques: ampere-hours and extended Kalman filter. The study result
demonstrates that while ampere-hours estimation provides an acceptable SoC estimation, it is slightly less
accurate compared to extended Kalman filter estimation. The ampere-hours technique introduces some degree
of error during the estimation process. Finlay, we employed the RMSE, a amount magnitude of differences
between estimated SoC values and actual values. resulting in an RMSE value of 0.78% for extended Kalman
filter estimation. These results highlight the superiority of the SoC_EKF technique in providing more precise
estimates of the SoC for drone batteries.
Maximizing energy efficiency in drones through accurate state of charge … (Kamal Anoune)
766 ISSN: 2252-8792
ACKNOWLEDGEMENTS
The author would like to express their sincere gratitude to the dedicated staff of the Physics
Department, their unwavering commitment to academic excellence and their tireless efforts have played a
pivotal role in the success of our academic and research endeavors.
REFERENCES
[1] J.-P. Yaacoub, H. Noura, O. Salman, and A. Chehab, “Security analysis of drones systems: Attacks, limitations, and
recommendations,” Internet of Things, vol. 11, p. 100218, Sep. 2020, doi: 10.1016/j.iot.2020.100218.
[2] J. Zhang, J. F. Campbell, D. C. Sweeney II, and A. C. Hupman, “Energy consumption models for delivery drones: A comparison
and assessment,” Transportation Research Part D: Transport and Environment, vol. 90, p. 102668, Jan. 2021, doi:
10.1016/j.trd.2020.102668.
[3] R. D’Andrea, “Guest Editorial Can Drones Deliver?,” IEEE Transactions on Automation Science and Engineering, vol. 11, no. 3,
pp. 647–648, Jul. 2014, doi: 10.1109/TASE.2014.2326952.
[4] T. Kirschstein, “Comparison of energy demands of drone-based and ground-based parcel delivery services,” Transportation
Research Part D: Transport and Environment, vol. 78, p. 102209, Jan. 2020, doi: 10.1016/j.trd.2019.102209.
[5] Z. Liu, X. Dang, B. Jing, and J. Ji, “A novel model-based state of charge estimation for lithium-ion battery using adaptive robust
iterative cubature Kalman filter,” Electric Power Systems Research, vol. 177, p. 105951, Dec. 2019, doi:
10.1016/j.epsr.2019.105951.
[6] G. S. Reddy, M. Premkumar, S. Ravi, and L. Abualigah, “An intelligent converter and controller for electric vehicle drives utilizing
grid and stand-alone solar photovoltaic power generation systems,” International Journal of Applied Power Engineering (IJAPE),
vol. 12, no. 3, p. 255, Jul. 2023, doi: 10.11591/ijape.v12.i3.pp255-276.
[7] L. Lu, X. Han, J. Li, J. Hua, and M. Ouyang, “A review on the key issues for lithium-ion battery management in electric vehicles,”
Journal of Power Sources, vol. 226, pp. 272–288, 2013, doi: 10.1016/j.jpowsour.2012.10.060.
[8] Y. Xia, W. Zeng, C. Zhang, and H. Yang, “A branch-and-price-and-cut algorithm for the vehicle routing problem with load-
dependent drones,” Transportation Research Part B: Methodological, vol. 171, pp. 80–110, May 2023, doi:
10.1016/j.trb.2023.03.003.
[9] D. Baek et al., “Battery-Aware Operation Range Estimation for Terrestrial and Aerial Electric Vehicles,” IEEE Transactions on
Vehicular Technology, vol. 68, no. 6, pp. 5471–5482, Jun. 2019, doi: 10.1109/TVT.2019.2910452.
[10] P. Shrivastava, T. K. Soon, M. Y. I. Bin Idris, and S. Mekhilef, “Overview of model-based online state-of-charge estimation using
Kalman filter family for lithium-ion batteries,” Renewable and Sustainable Energy Reviews, vol. 113, p. 109233, Oct. 2019, doi:
10.1016/j.rser.2019.06.040.
[11] K. Anoune, M. Bouya, A. Astito, and A. Ben Abdellah, “Design and sizing of a Hybrid PV-Wind-Grid System for Electric Vehicle
Charging Platform,” MATEC Web of Conferences, vol. 200, p. 00008, Sep. 2018, doi: 10.1051/matecconf/201820000008.
[12] W. Xu et al., “A novel adaptive dual extended Kalman filtering algorithm for the Li‐ion battery state of charge and state of health
co‐estimation,” International Journal of Energy Research, vol. 45, no. 10, pp. 14592–14602, Aug. 2021, doi: 10.1002/er.6719.
[13] A. Nikolian et al., “Lithium Ion Batteries—Development of Advanced Electrical Equivalent Circuit Models for Nickel Manganese
Cobalt Lithium-Ion,” Energies, vol. 9, no. 5, p. 360, May 2016, doi: 10.3390/en9050360.
[14] I. Baccouche, S. Jemmali, B. Manai, N. Omar, and N. Amara, “Improved OCV Model of a Li-Ion NMC Battery for Online SOC
Estimation Using the Extended Kalman Filter,” Energies, vol. 10, no. 6, p. 764, May 2017, doi: 10.3390/en10060764.
[15] M. Souaihia, B. Belmadani, R. Taleb, and K. Tounsi, “State of charge estimation by using extended Kalman filter based on improved
open circuit voltage model,” International Journal of Applied Power Engineering (IJAPE), vol. 10, no. 1, p. 1, Mar. 2021, doi:
10.11591/ijape.v10.i1.pp1-10.
[16] N. Shi, Z. Chen, M. Niu, Z. He, Y. Wang, and J. Cui, “State-of-charge estimation for the lithium-ion battery based on adaptive
extended Kalman filter using improved parameter identification,” Journal of Energy Storage, vol. 45, p. 103518, Jan. 2022, doi:
10.1016/j.est.2021.103518.
[17] M. Wu, L. Qin, G. Wu, and C. Shi, “State of Charge Estimation of Lithium ion Battery Based on a New Linear Online Model,” in
2020 39th Chinese Control Conference (CCC), Jul. 2020, pp. 5282–5286, doi: 10.23919/CCC50068.2020.9189457.
[18] X. Lai et al., “A joint state-of-health and state-of-energy estimation method for lithium-ion batteries through combining the
forgetting factor recursive least squares and unscented Kalman filter,” Measurement, vol. 205, p. 112187, Dec. 2022, doi:
10.1016/j.measurement.2022.112187.
[19] S. Qin, D. Qin, H. Wu, T. Wang, J. Chen, and P. Wang, “State of Charge estimation of lithium-ion power battery based on online
parameter identification method and BP neural network,” International Journal of Electrochemical Science, vol. 17, no. 1, p.
220138, Jan. 2022, doi: 10.20964/2022.01.05.
[20] K. Anoune, M. Ghazouani, M. Ghazi, and A. Laknizi, “Energy-Reducing Opportunities by Improving Power Quality: A Case Study
of Industrial Plants,” 2023, pp. 622–631.
[21] S. Jung, Y. Jo, and Y.-J. Kim, “Flight Time Estimation for Continuous Surveillance Missions Using a Multirotor UAV,” Energies,
vol. 12, no. 5, p. 867, Mar. 2019, doi: 10.3390/en12050867.
[22] A. Rahighi, S. M. H. Seyed Kashani, and B. Sakhaee, “Developing a battery monitoring system software in matlab simulink
environment using Kalman filter,” International Journal of Applied Power Engineering (IJAPE), vol. 8, no. 1, p. 1, Apr. 2019, doi:
10.11591/ijape.v8.i1.pp1-10.
[23] M. Lagraoui, A. Nejmi, H. Rayhane, and A. Taouni, “Estimation of lithium-ion battery state-of-charge using an extended Kalman
filter,” Bulletin of Electrical Engineering and Informatics, vol. 10, no. 4, pp. 1759–1768, 2021, doi: 10.11591/eei.v10i4.3082.
[24] M. Hossain, M. E. Haque, and M. T. Arif, “Kalman filtering techniques for the online model parameters and state of charge
estimation of the Li-ion batteries: A comparative analysis,” Journal of Energy Storage, vol. 51, p. 104174, Jul. 2022, doi:
10.1016/j.est.2022.104174.
[25] P. Takyi-Aninakwa, S. Wang, H. Zhang, H. Li, W. Xu, and C. Fernandez, “An optimized relevant long short-term memory-squared
gain extended Kalman filter for the state of charge estimation of lithium-ion batteries,” Energy, vol. 260, p. 125093, Dec. 2022, doi:
10.1016/j.energy.2022.125093.
[26] S. Wang, C. Fernandez, C. Yu, Y. Fan, W. Cao, and D.-I. Stroe, “A novel charged state prediction method of the lithium ion battery
packs based on the composite equivalent modeling and improved splice Kalman filtering algorithm,” Journal of Power Sources,
vol. 471, p. 228450, Sep. 2020, doi: 10.1016/j.jpowsour.2020.228450.
Int J Appl Power Eng, Vol. 13, No. 3, September 2024: 755-767
Int J Appl Power Eng ISSN: 2252-8792 767
[27] C. Wu, W. Hu, J. Meng, X. Xu, X. Huang, and L. Cai, “State-of-charge estimation of lithium-ion batteries based on MCC-AEKF
in non-Gaussian noise environment,” Energy, vol. 274, p. 127316, Jul. 2023, doi: 10.1016/j.energy.2023.127316.
[28] Y. Chen, D. Baek, A. Bocca, A. Macii, E. Macii, and M. Poncino, “A Case for a Battery-Aware Model of Drone Energy
Consumption,” in 2018 IEEE International Telecommunications Energy Conference (INTELEC), Oct. 2018, pp. 1–8, doi:
10.1109/INTLEC.2018.8612333.
[29] Z. He et al., “State-of-charge estimation of lithium ion batteries based on adaptive iterative extended Kalman filter,” Journal of
Energy Storage, vol. 39, p. 102593, Jul. 2021, doi: 10.1016/j.est.2021.102593.
BIOGRAPHIES OF AUTHORS
Maximizing energy efficiency in drones through accurate state of charge … (Kamal Anoune)