ANN-Based Reference Voltage Generation Scheme for Control of Dynamic Voltage Restorer

Abstract

Dynamic voltage restorer (DVR) is usually employed to mitigate sag/swell in supply voltages so that load voltage is regulated at nominal value. This paper proposes artificial neural network (ANN) based reference voltage generation (RVG) scheme for the control of 3-phase DVR. ANN replaces the traditional control of DVR, which involves abc-dq0 and dq0-abc transformations, estimation of d-q axes voltage errors and proportional-integral (PI) controllers along with their tuning. In proposed control scheme, the feedforward ANN utilizes present and previous samples of supply voltage and peak magnitude of load voltage for RVG, which when impressed across the injection transformer results in sag/swell mitigation. It is important to note that the proposed scheme is free from transformations and controller tuning. The performance of 3-phase DVR with the proposed ANN based RVG scheme results in standard IEEE-519 compliant operation with load voltage regulated at nominal value even under sag/swell in supply voltage. This is verified through MATLAB/SIMULINK based simulaiton studies.