Power Quality Improvement of a Reduced Switch MLI Using a Modified ANN-Based SHE Technique

A cross-connected source (CCS) based multilevel inverter (MLI) is proposed for the generation of “9” and 13-levels using asymmetrical switching. The reduced switch CCS-MLI is controlled with a low frequency selective-harmonic-elimination pulse-width-modulation (SHE-PWM) strategy. The switching angles are synthesized by solving SHE equations utilizing the proposed modified artificial-neural-network (ANN) like architecture. The ANN can solve multiple transcendental equations starting from random initial guess points and without any initial data. This adds a significant improvement to the existing techniques that suffer severely with an increased number of SHE equations to be solved. To verify these unique features of the modified ANN, a CCS-MLI is realized with a higher number of levels. The enhanced performance and reasonable convergence rate of the ANN validate the suitability of the proposed control. Furthermore, CCS-MLI is simulated for resistive-inductive load considering THD as a performance parameter. The power quality improvement of the proposed inverter is shown for different modulation index (MI) and significantly low THD is reported for both “9” and 13-level case. Experimental results of a hardware prototype are also illustrated. Furthermore, the proposed system is implemented as a central inverter of standalone PV system to test its practical feasibility.