Neural network control technique based sensorless DSTATCOM for the power conditioning

Abstract

Distributed static compensator (DSTATCOM) is an important power device for mitigation of harmonics, power factor correction and voltage regulation in a distributed network. This paper describes the conception and simulation of a three-phase VSC based DSTATCOM for a three phase ac mains feeding to linear and nonlinear loads. The operation of DSTATCOM is governed by a novel control technique known as neural network (NN) or artificial neural network (ANN). ANN based DSTATCOM has been used, in order to maintain the direct current link voltage. Hysteresis current controller is employed to generate switching signals for the voltage source converter (VSC). To obtain the switching signals for the operation of DSTATCOM, the source reference current is compared with the sensed source current and it is fed to the hysteresis controller. Simulation results show the better results using ANN controller over the conventional PI controller. The DSTATCOM is verified using the simulations in Matlab/Simulink.