Harmonic Optimization in a Nine-Level Cascaded H-Bridge Inverter: Leveraging Support Vector Machine Algorithm with LSPWM Technique

This paper presents a novel approach that integrates a support vector machine (SVM) regression model with a nine-level cascaded H-bridge multilevel voltage source inverter triggered by LSPWM. The objective is to optimize low-frequency odd harmonics in the output voltage while achieving a uniform output current, thus improving the power quality of the inverter. Specifically, the focus is on predicting the fifth, seventh, and eleventh order harmonics from the phase voltage and optimizing their amplitudes. The innovative SVM-based LSPWM technique predicts the best-fitted amplitude range of odd harmonics under variable modulation indices and switching angles using simplified low-order equations, offering an alternative to complex nonlinear equations. Simulation and experimental results closely align with theoretical analysis, affirming the adaptability of the proposed work for industrial applications.