ARTIFICIAL NEURAL NETWORK CONTROLLER FOR PERFORMANCE OPTIMIZATION OF SINGLE PHASE INVERTER

Thyristorised Power control provides high efficiency. However, generated harmonics cause a nuisance inpower system operation. The work presented here, deals with reduction of harmonics (3rd-11th) by usingMultiple Pulse Modulation technique. Traditional numerical methods do not yield accurate pulse-positionsas non-linearity is involved in computation. In this paper, a continuous Hopfiled Neural Network isdesigned for Harmonic minimization in a 1 inverter output voltage. Results show considerableimprovement in voltage spectrum if trigger pulses are generated at the ANN positions as harmonic contentsare reduced with significant improvement in fundamental voltage resulting in reduction in device ratings.The ANN controlled voltage is used to drive a 1 Induction motor in MATLAB simulation and is comparedwith SPWM controlled VSI driving the same motor load. The quantitative analysis is given in tabular form.This shows feasibility of design of a controller for optimized performance of a single phase VSI.