Enactment of Deep Reinforcement Learning Control for Power Management and Enhancement of Voltage Regulation in a DC Micro-Grid System

Abstract

DC micro-grids are gaining more attention and becoming more predominant due to its robustness and ease of control. To confront the stability issues of the DC Micro-grid system, a model free intellectual Control approach is proposed and recognized with a deep Q-learning controller to influence the functions of DC–DC boost converter. This self-reliant model of deep Q-learning controller establish a suitable communication link among the agent-environment by applying the reward/penalty algorithm for attaining the stable output voltage with its training statistics. Upon learning the state of agent, an optimal proposal is defined by the controller to attain the stabilized voltage at the output for all possible operating scenarios. The simulation and experimental studies validate the performance of the controller with higher accuracy, reduced overshoot, and faster settling time upon the transient states.