Neural network based shunt active filter for harmonic and reactive power compensation under non-ideal mains voltage

This paper presents a new method for harmonic and reactive power compensation with power factor improvement using an artificial neural network (ANN) and a new control algorithm for active power filter (APF) for power quality conditioning for variable load under non-ideal mains voltage conditions. The neural network controller comprises two similar adaptive linear neurons (ADALINE), and it has been designed to extract fundamental frequency components from non-sinusoidal and unbalanced currents instead of conventional low pass filter. Reactive power compensation is done without sensing load currents, which gives simplicity in control with less number of current sensors. The performance of the APF with the proposed neural network compensation algorithm is found to be considerably effective and adequate to compensate harmonics and reactive power. The results show excellent behaviors and performances.