Selective Harmonic Elimination of The Multilevel Inverter Using Artificial Neural Network

Abstract

The multilevel inverter is a powerful electronic device widely used for high power utility applications. The main purpose of the multilevel inverter is to provide sinusoidal waveforms with low level harmonic content to reduce distortion. Improving the inverter performance means improving the quality of the output voltage. Here we made an attempt of eliminating the desired harmonics in the output voltage waveform by solving the system of non-linear transcendental equations for getting the solution vector for different modulation index, here the solution of this system of equations was obtained using the method of least-squares (numerical method). After getting the solutions we have chosen the feed-foreword structure of ANN (artificial neural network). The ANN was made to learn by using the error-backpropagation algorithm. And finally we have obtained the weights to the corresponding neural network. The multilevel inverter chosen was three level cascaded inverter, which was simulated in the Matlab for different modulation indexes