Neural network control for three-leg VSC based DVR in distribution system

This paper deals with neural network control for three-leg Voltage Source Converter (VSC) based DVR in distribution system to mitigate voltage sag, swell and harmonics etc. The proposed neural network control is based on the least mean-square algorithm which is known as adaptive linear element to extract the fundamental component of load voltages. The reference signals for three-leg VSC based DVR are extracted from reference load voltages. The neural network control for DVR is able to self-support its dc bus through the control under varying loads. The main advantage of neural network control is to eliminate filter and which improves performance of DVR. The proposed DVR injects voltages in series with source voltage to regulate voltage at rated voltage. The proposed neural network control based DVR is validated through computer simulation studies using MATLAB/SIMULINK.