Comparative Analysis of GA Variants for Solving the SHE Problem in Multilevel Inverters

In recent years, the Cascaded H-Bridge Multilevel Inverter (CHB-MLI) topology, controlled by the Selective Harmonic Elimination (SHE) method, has been widely preferred in high-power applications —especially where power quality is critical— due to its superior output quality, low switching losses, and reduced voltage stress on the power switches. However, since the SHE method involves nonlinear transcendental equations, it becomes evident that analytical solution methods are insufficient as the inverter voltage levels increase. To overcome this challenge, researchers frequently resort to metaheuristic optimization algorithms. In this context, the present study compares the performance of different Genetic Algorithm (GA) variants in solving the SHE equations of a Cascaded H-Bridge Multilevel Inverter (CHB-MLI) system. A total of eight algorithms, including the Standard GA (SGA) and several improved variants from the literature, were evaluated. Each algorithm was tested through independent runs to determine the optimal switching angles which minimize a fitness function specifically created for the SHE equations. The obtained optimal switching angles were applied to a CHB-MLI system modelled in the MATLAB/Simulink environment, and the performance of each variant was analysed based on output voltage quality, fundamental component accuracy, and Total Harmonic Distortion (THD) criteria. The results reveal the strengths and weaknesses of different variants in the optimization process and indicate the potential of a well-designed GA as an effective alternative for addressing complex engineering problems, such as the solution of SHE equations.