Power Quality Enhancement Using Deep Learning Technique Based Dynamic Voltage Restorer

This research is about improving the electrical power quality of the network that feeds a nonlinear load by reducing the total harmonic distortion and the neutral conductor current of the network and thus improving the overall performance of the system. To implement adaptive harmonic compensation for dynamic voltage restorer under changing load conditions, a deep learning, based control framework is proposed. The proposed approach employs recurrent learning architectures, including gated-recurrent units and LSTM networks, to optimize the operation of the DVR. The recital of the deep learning models is compared with the conventional pq0 theory for harmonic identification and compensation. Simulation results demonstrate that the proposed method satisfies international harmonic distortion limits while effectively minimizing neutral current. In addition, the controller maintains stable DC-link voltage regulation in the presence of small voltage disturbances. The effectiveness of the proposed deep learning framework is validated using performance metrics such as accuracy, computational complexity, and loss function. The results indicate high prediction accuracy with reduced computational burden. The developed strategy is applicable to three-phase, four-wire distribution networks and is suitable for both residential and commercial power quality enhancement applications.