Reactive Voltage Optimal Control of Distribution Networks Based on Deep Reinforcement Learning with Deep Q Network

In recent years, reactive voltage optimal control is an advanced control methodology in modern power distribution networks. In distributed networks increased renewable energy resources became a significant challenge for unpredictable localized voltage fluctuations leads to voltage sags, swells and losses also compromise power quality and system reliability. The traditional methods which Assisted for increased renewable energy resources in distributed networks. Therefore, this paper proposes Reactive Voltage Optimal Control-Deep Q Network (RVOC-DQN) for voltage power compensation to enhance fast response time and instantaneous voltage correction. The proposed method employed on GridLAB-D dataset to integrate and control renewable energy source generation using Moving Average Filter for system faults. The preprocessed data features are extracted using Static Volt Ampere Reactive Compensators-Closed-Loop Voltage Control (SVC-CLVC) for focused control in voltage regulation to dynamic grid conditions with load and generation patterns. Finally, RVOC-DQN is employed for rapid voltage regulation and continuous operation with improved data quality in distributed networks. The proposed RVOC-DQN method attained high stability of the system in terms of Root Mean Square Error (RMSE), Correlation and values are (0.26) of RMSE and the Correlation value is (0.835) when compared with existing Active Distribution Networks.