Selective Harmonic Elimination in Single Phase Inverter using Artificial Neural Network

Abstract

Selective harmonic elimination (SHE) is a widespread PWM method applied to voltage source inverters (VSI) to control fundamental voltage and eliminate selected harmonics. In order to eliminate the harmonics, the switching angles are obtained by solving the non-linear transcendental equations using Newton-Raphson method. However, in the proposed online Artificial Neural Network (ANN) method, the firing angles are calculated by Bayesian Regularization Back- propagation learning algorithm where look up table is not necessary to store the firing angles. Further, with the help of the obtained switching instants, PWM signals are generated and applied to the single phase inverter in MATLAB Simulink. The output waveforms of inverter with R load for various values of modulation index are analyzed. Furthermore, the performances of different ANNs were compared with the gradient descent algorithm.