A Neural Network-Based Control Strategy for Three-Phase Four-Leg Dynamic Voltage Restorer for Both Voltage Sag/Swell and Harmonic Compensation

摘要

Conventional dynamic voltage restorers are based on three-phase inverters, which are unable to compensate unbalanced voltage; therefore, in this paper, a four-leg three-phase inverter topology is employed to be used in DVR structure to compensate both balanced and unbalanced voltage sag/swells. Various control strategies have been proposed for controlling DVR. However, they suffer from slow dynamic response or need massive mathematical calculation. In the proposed method, an indirect neural network recognition method is used to train a neural network, then, the trained network is used to compensate voltage sag and swell. Furthermore, the offered control strategy is able to compensate voltage harmonics. The suggested strategy is simulated in MATLAB/Simulink environment under various conditions. The results show the validity of the proposed control strategy.