A Source-Load-Storage Cooperative Power Regulation Method for Active Distribution Network Based on Deep Reinforcement Learning

With the development of various new energy technologies, the penetration rate of various distributed power sources and controllable loads in the distribution network, including distributed photovoltaic (PV) power sources and energy storage systems, has continuously increased, making the operation of active distribution networks (ADNs) more complex and difficult to control. In this study, we propose using the deep reinforcement learning (DRL) method to solve the problem of source-load-storage cooperative power regulation problem for ADN. Firstly, we describe an ADN model with distribution PV systems, battery energy storage systems (BESS), and controllable loads, then the Markov decision process of source-load-storage cooperative power regulation in ADN is analyzed and we use the TD3 algorithm to achieve effective control. The proposed method is validated by using the IEEE 33 bus system as an example.