Soft Computing DL-based Dynamic Voltage Restorer (DVR) for Improvement of Power Quality in Grid-Connected Systems

This paper presents an intelligent control strategy for a Dynamic Voltage Restorer (DVR) based on Soft Computing and Deep Learning (DL) techniques to enhance power quality in grid-connected systems. The proposed hybrid controller combines fuzzy logic and a deep neural network (DNN) to achieve adaptive voltage compensation during grid disturbances such as voltage sag, swell, and harmonics. The soft computing layer ensures real-time decision-making, while the DL model enhances prediction accuracy and dynamic response. The DVR system is modeled and simulated in MATLAB/Simulink under various fault and load conditions. Simulation results show that the proposed controller effectively restores load voltage, minimizes total harmonic distortion (THD), and offers faster transient recovery compared to conventional PI and fuzzy-based DVR controllers. The approach demonstrates improved adaptability and robustness, making it a promising solution for power quality enhancement in smart grid environments.