Deep Reinforcement Learning-based Data-Driven Active Power Dispatching for Smart City Grid

In order to enhance the frequency regulation performance of islanded microgrids, a data-driven active power dispatching (DDB-APD) method is proposed to realize the multi-objective comprehensive In addition, a proximal policy optimization (PPO) algorithm is introduced to achieve load frequency control in microgrids through data-driven policy gradient optimization. In this method, the original active power controller is replaced with an agent to achieve fast active power control by setting a reasonable reward function. Simulations of an isolated microgrid in the Southern Power Grid illustrate that the proposed method can effectively reduce microgrid frequency deviations and generation costs.