Wang Renzheng, Zhang Yuncheng, Chen Ying, Xin Zhenyao, Fan Di
College of Electronic and Information Engineering, Shandong University of Science and Technology, Qingdao, 266590, China.
Sci Rep. 2024 Nov 30;14(1):29773. doi: 10.1038/s41598-024-80923-z.
Selective Harmonic Elimination Pulse Width Modulation (SHEPWM) has excellent harmonic characteristics, but its nonlinear transcendental system of equations is difficult to be solved, and the practical application encounters a bottleneck. In this paper, a modulation optimization method for seven-level SHEPWM inverter based on the Evolutionary Particle Swarm Optimization (EPSO) algorithm is proposed to address this problem, so that the algorithm quickly converges to the global optimum solution. The EPSO algorithm incorporates a population optimization strategy in two phases to improve the population diversity in real time. In the initialization phase, the initialized population is optimized using Opposition-Based Learning (OBL) to improve the quality of the initial population. In the iterative stage, we combine the adaptive Particle Swarm Optimization (PSO) algorithm, Tunicate Swarm Algorithm (TSA), Adaptive Gaussian Variation, Quasi-Opposition-Based Learning (QOBL) and other optimization methods to solve the problem of insufficient population diversity in the process of searching for the optimal solution, to break through the local optimum, and to improve the convergence speed and accuracy of the algorithm. Experiments of the algorithm in 19 benchmark functions and seven-level SHEPWM inverter optimization modulation show that the optimization ability of the EPSO algorithm is ahead of TSA, INFO, MA (Mayfly Algorithm), EO (Equilibrium Optimizer) and other optimization algorithms. The solution speed is about three times that of PSO, which achieves fast and highly accurate convergence, with a small error in the output of inverter, and better harmonic distortion rate than the standard requirement.
选择性谐波消除脉宽调制（SHEPWM）具有优异的谐波特性，但其非线性超越方程组难以求解，实际应用中遇到瓶颈。本文提出一种基于进化粒子群优化（EPSO）算法的七电平SHEPWM逆变器调制优化方法来解决这一问题，使算法快速收敛到全局最优解。EPSO算法在两个阶段融入群体优化策略，以实时提高群体多样性。在初始化阶段，利用基于反向学习（OBL）对初始化群体进行优化，提高初始群体质量。在迭代阶段，结合自适应粒子群优化（PSO）算法、樽海鞘群算法（TSA）、自适应高斯变异、基于准反向学习（QOBL）等优化方法，解决在寻找最优解过程中群体多样性不足的问题，突破局部最优，提高算法的收敛速度和精度。该算法在19个基准函数和七电平SHEPWM逆变器优化调制方面的实验表明，EPSO算法的优化能力优于TSA、INFO、MA（蜉蝣算法）、EO（平衡优化器）等优化算法。求解速度约为PSO的三倍，实现了快速且高精度的收敛，逆变器输出误差小，谐波畸变率优于标准要求。
