Performance investigation of the shunt active power filter using neural network

This paper present a neural network based shunt active power filter for power quality improvement. It is used to improve power quality by compensating harmonics and reactive power requirement of non-linear loads. The compensation process proposed is simple, and is based on sensing source currents and regulating the dc link voltage. Artificial neural network (ANN) control is used to control the dc link voltage in place of the traditional proportional plus integral controller. The ANN controller trained offline using data from PI controller. The performance of the ANN controller is compared with the PI controller by MATLAB/Simulink simulation study.