Performance Evaluation Of PSO Variants For Selective Harmonic Elimination In Multi-Level Inverters

In this study, the performance of different Particle Swarm Optimization (PSO) variants in solving the Selective Harmonic Elimination (SHE) equations of a Cascaded H-Bridge Multilevel Inverter (CHB-MLI) was compared. The SHE method consists of nonlinear transcendental equations, which are particularly difficult to solve analytically in high-level systems. The aim of the study is to examine the overall improvements provided by the PSO variants and to specifically evaluate their effectiveness in solving such complex engineering problems. In addition to the standard PSO, several improved versions from the literature have been considered. For each variant, the optimal switching angles that solve the SHE equations were determined and applied to a CHB-MLI model in the MATLAB/Simulink environment. The convergence behaviors of the algorithms, the total harmonic distortions (THD) and the amplitudes of the fundamental components of the output voltages were analyzed statistically. As a result of these analyses, the strengths and weaknesses of the PSO variants in the optimization processes were revealed. Based on the findings, a hybrid model was also proposed, which integrates the strong features of the successful variants. It was observed that the proposed hybrid model stands out from the other variants by exhibiting a stable and competitive performance even in the worst-case scenarios. These findings indicate that effectively developed PSO variants can be a powerful alternative for solving real-world optimization problems.