The microgrid is susceptible to assaults as a cyber-physical system due to its open and varied surroundings. By creating and injecting a false attack vector to evade system detection, the false data injection attack poses a danger to grid security. The variety of assaults makes it impossible to use fixed approaches to identify false data injection attacks. This study introduced a method using deep reinforcement learning (DRL) to spot false data injection attacks and disconnect the microgrid from the main grid to reduce the impact of the attack. Initially, we examined an attack model assuming infinite attack resources and full topological information. Identification of false data injection attacks (FDIAs) is suggested using a deep reinforcement learning-based approach. Simulation on MATLAB Simulink of the microgrid islanding system demonstrated the effectiveness of the detection technique and the validity of the attack mitigation model.
