Planning for Network Expansion Based on Prim Algorithm and Reinforcement Learning

Existing methods for expanding distribution network grids lack flexibility in handling multiple scenarios, particularly in the expansion planning of photovoltaic(PV)-enabled distribution networks where the problem of adding load points is mainly addressed based on empirical knowledge without considering the economic and reliability aspects throughout the entire planning process. To address this issue, this paper proposes a planning method that combines Prim algorithm and reinforcement learning (RL). Firstly, historical data of PV and load characteristics are extracted, and multiple typical PV and corresponding load scenarios are generated using clustering algorithms as planning scenarios. Secondly, the Prim algorithm is introduced to improve the action rules and environment of RL, and the mathematical model of planning is transformed into a corresponding reward function. By using RL, the optimal planning path can be found, which enhances the overall operational efficiency of the network and ensures the power supply of the new load points, providing a reference for future operation planning scenarios. Finally, calculations and data analysis based on IEEE standard cases verify the scientific and efficient nature of the proposed method.