Dynamic Voltage Stability Enhancement in Electric Vehicle Battery Charger Using Particle Swarm Optimization

Electric vehicles (EVs) are poised to lead the transportation sector as the primary choice of automobile due to their efficiency and environmental benefits. EVs with enhanced autonomy and reduced pricing have become feasible in the market, enabling a gradual transition for higher EV penetration. However, electric vehicles require highly efficient and stabilized charging stations in urban areas to ensure the vehicle’s charging time is not compromised. In this regard, the Vienna rectifier with a voltage-oriented controller (VOC) plays a significant role in improving the power quality of the utility grid for EV battery charger applications. The low stability of the battery charger’s output voltage and current is due to the trial-and-error method used to select the PI controller gains. In order to improve the voltage and current stability, the particle swarm optimization (PSO) technique is used to optimize VOC’s PI controller gains. The code composer studio (CCS) platform integrates the PSO technique for EV battery chargers in the experimental setup. The Vienna rectifier with VOC for EV battery charger is implemented using TMS 320F28337 digital signal controller in the test board. Findings indicate that the PSO optimized VOC improves the output voltage and current stability by 12% compared to the existing trial-and-error technique. Furthermore, the proposed system is tested in an experimental setup that provides input current THD to less than 5% for different load variations (up to 1.5kW) to meet the IEEE-519 standards. Results from simulations and experimental setup verify that the proposed PSO-PI controller-based Vienna rectifier significantly improves EV battery chargers’ output voltage and current stability.


I. INTRODUCTION
Fossil fuels are widely used to power the existing transporta- 21 tion sectors in the modern world, which increases pollution, 22 The associate editor coordinating the review of this manuscript and approving it for publication was N. Prabaharan . noise, and global warming [1], [2]. Another key issue for 23 the existing transportation industry is the fast depletion of 24 underground petroleum resources due to the overuse of fossil 25 fuels and the rise in fossil fuel prices [3], [4], [5]. The rising 26 cost of fossil fuels, environmental pollution, and the finite 27 lifespan of fossil fuels have motivated automobile makers to 28 integral square error (ISE), integral absolute error (IAE), 85 rise time (t r ), starting time (t s ), and peak overshoot (M P ) 86 [23], [24]. The reduction in these system parameters helps 87 to improve the stability of the system. In this regard, the 88 researchers and designers always choose the new algorithm 89 that has less complexity, uses fewer parameters, and is more 90 efficient than the existing algorithms [25]. The existing trial-91 and-error method of PI controller tuning techniques is inflex-92 ible, unstable, and complex. As a result of the lack of knowl-93 edge of mathematical models and trial-and-error methods, the 94 robustness of the PI controller is reduced, resulting in poor 95 controller performance. In order to address the periodic errors 96 in the output voltage, sliding mode controllers (SMC) are 97 implemented in rectifier systems for different load variations. 98 As a result, the total harmonic distortion at the input current 99 is maintained at less than 5% to meet the IEEE-519 standards 100 with linear and non-linear loads [26]. 101 Evolutionary computing techniques, artificial neural net-102 works (ANN), and fuzzy logic are used to design the opti-103 mized PI controllers. Due to the fast development of computer 104 power, the PI controller based on a computer is designed 105 within a short period. The tuning strategies based on the 106 optimization technique are more efficient than the existing 107 trial-and-error method due to their independence from system 108 dynamics and PI control structure [27], [28], [29]. Heuris-109 tic algorithm-based optimization strategies used in control 110 engineering are one of the powerful ways of solving control 111 issues in a wide range of situations [30], [31], [32], [33], 112 [34], [35], [36]. These algorithms are particularly useful 113 in process control due to their simple structure, enhanced 114 optimization, and fast response. They are more effective at 115 solving complex optimization problems with many dimen-116 sions than conventional optimization approaches. Because of 117 their adaptability, these algorithms are well-suited to con-118 temporary classical design methodologies. Regardless of the 119 model order, these algorithms serve as a critical tool for 120 developing classical and modified structured controllers for 121 an unstable process model class. The genetic algorithm (GA) 122 [37] and particle swarm optimization (PSO) technique [38] 123 are the two key strategies commonly used in controller design 124 applications for optimization. Due to the intensive study 125 of various algorithms, the PSO technique has significantly 126 been improved for numerous industrial applications. As a 127 self-tuning algorithm, the PSO technique uses the Objective 128 Function (OF) provided to assist the algorithm in identifying 129 the optimum K p , K i , and K d values for the process. As a 130 typical criticism of nature-inspired design approaches and 131 bio-inspired metaheuristics, it is often argued that they both 132 need modifications or adjustments in parameters prior to 133 optimization. The classical PSO technique, on the other hand, 134 contains fewer heuristic variables than the GA technique, 135 making it more straightforward for optimization. Therefore, 136 the PSO technique is selected to optimize VOC's PI controller 137 in this study as a simpler technique.

138
This study mainly focuses on optimizing VOC's PI 139 controller-based Vienna rectifier for EV battery chargers. 140  0.31 seconds, respectively, which is better than the trial-and-148 error method. Also, the peak overshoot value is 1.21% for 149 the Vienna rectifier with VOC for the EV battery chargers.

150
The system parameters such as rise time, settling time, and  data centers, welding power sources, and electric aircraft 173 applications. It is often used as a front-end power converter 174 as it can provide input current with THD less than 5% and an 175 improved power factor at the grid side to satisfy the IEEE-519 176 standards. The Vienna rectifier also has high-power density 177 and high-power handling capability for conversion of AC/DC 178 applications. The block diagram of a three-phase Vienna 179 rectifier integrated with the C2000 microcontroller is shown 180 in Fig. 1. In this study, the Vienna rectifier is used as a 181 front-end converter with VOC for the EV battery charger. The 182 VOC is a highly efficient controller for EV battery chargers 183 compared to existing controllers with Vienna rectifier. The 184 Park's and Clark's transformation of VOC is shown in Fig. 2, 185 and the three PI controller in the voltage-oriented controller 186 is shown in Fig. 3. Park's transformation helps to transform 187 input three-phase quantities such as phase A, phase B, and 188 phase C into two-phase stationary quantities (α and β). Also, 189 Clark's transformation in the VOC helps to transform sta-190 tionary two-phase quantities into two-phase rotating quan-191 tities or reference frames (d-axis and q-axis). Similarly, the 192 inverse Park's transformation and Clark's transformation help 193 to convert the rotating two-phase reference frame (d axis and 194 q axis) into a stationary reference frame and the two-phase 195 VOLUME 10, 2022  reference frame into three-phase ABC systems, respectively.

196
The existing trial-and-error method-based Vienna rectifier 197 with a voltage-oriented controller helps to reduce the input 198 current THD to less than 5% and improve the power factor 199 at the utility grid side. In addition, the PSO optimization 200 of VOC's PI controller with Vienna rectifier for EV battery 201 charger helps to optimize the gain constants of PI controller 202 to improve the system's stability.

203
The synthesis of the PI controllers is mathematically 204 described by, where K p is the proportional gain constant, K i is the integral 207 gain constant, and e (t) is the difference between the set point 208 and the plant output.
209 VOLUME 10, 2022  using MATLAB, and the results are presented in Table 2.

236
The three main stages of PSO algorithm can be explained as 237 follow:

238
• Evaluating the fitness value of each particle.

239
• Updating local and global best fitness and positions.

240
• Updating the velocity and the position of each particle.

241
The following equations give the particle position and veloc-242 ity update for optimizing the PSO algorithm [47].  In this study, the VOC's PI controller has been optimized by VOLUME 10, 2022

IAE PI
The cumulative objective function for the proposed PSO 274 optimized VOC's PI controller is where, equations (4) and (5)      provides the minimum settling time, peak overshoot, and 299 rise time among the two conventional objective functions.

300
As stated before, any objective function could be used to 301 optimize the PI parameters. However, the challenging part is 302 reducing the rise time without increasing the peak overshoot 303 value. By decreasing the rise time, the system will attempt 304 to track the set point faster, resulting in higher inertia and 305 a higher risk of peak overshoot value. With the help of the 306 best convergence values of integral absolute error, the optimal 307 time response (rise time, settling time, and peak overshoot) 308 has been obtained in all cases.

309
The PSO optimized objective functions such as integrated 310 square error and integrated absolute error of voltage and 311 current controller in the VOC's PI controller for EV charging 312 VOLUME 10, 2022  Fig. 9. The test 344 board has experimented with various load conditions and 345 different periods for electric vehicle charging stations. The 346 load used in this study is a resistive load (R L ) for the exper-347 imental validation. The output performance parameters are 348 recorded using a power quality analyzer. The input current 349 and voltage for the Vienna rectifier with PSO technique for 350 the 650V DC with 1131W and output power are illustrated in 351 Fig. 10 and Fig. 11, respectively. Also, the input current and 352 input voltage for the Vienna rectifier with the PSO technique 353 for the 650V DC output voltage with 1176.5W output power 354 is illustrated in Fig. 12 and Fig. 13, respectively. According 355 to the experimental test analysis, it has been shown that 356 the input current THD is 2.47% which is less than 5% to 357 meet the IEEE-519 standards. The input current THD for 358 different load conditions is illustrated in Fig. 14, Fig. 15, 359 and Fig. 16 Table 3. In addition, the overall control circuit 367 with Vienna rectifier with VOC controller for EV battery 368 charger in order to reduce the input current THD less than 369 5% and to improve the power factor at the utility grid side is 370 shown in Fig. 17.