ANN-optimized NR-SHE method for single-sourced, multi-configurable switched capacitor-based multilevel inverter for sustainable power applications

Abstract

Multilevel inverters (MLIs) are widely preferred in sustainable power applications for their ability to deliver efficient output with low total harmonic distortion (THD) while minimizing the use of power semiconductor components. The single-sourced multilevel inverter based on switched capacitors has fewer components and better boosting. The proposed design eliminates the need for large DC-DC converters and transformers commonly used in traditional systems by incorporating switching capacitors. Furthermore, the self-balancing nature of these capacitors eliminates the necessity for additional sensors or control algorithms. In addition to supporting both symmetric and asymmetric configuration, the design achieves twice the voltage gain with a lower total blocking voltage (TBV). The ANN-based Newton-Raphson (NR) method, combined with selective harmonic elimination, can achieve a reduction in total harmonic distortion by up to 6%. A lower total blocking voltage, less losses, a smaller component count and an enhanced gain factor are some of the benefits of the system. In addition, the suggested system has an efficiency of up to 96% and lower overall voltage stress. Experimental validation under varying conditions has been conducted in addition to simulations using MATLAB/Simulink to assess the dynamic performance and operation of the suggested topology.