Promotion Strategy and Assessment Method of Power System Resilience Based on Proximal Policy Optimization Algorithm

Typhoon disasters can easily lead to disconnection of distribution network lines and split into multiple isolated island operations. How to efficiently realize distribution network reconfiguration and distributed generation scheduling to promote post-disaster load recovery is an urgent problem to be solved. In this paper, a strategy and evaluation method for improving the resilience of distribution network based on deep reinforcement learning proximal policy optimization (PPO) algorithm is proposed. Firstly, the failure probability model of distribution network under typhoon weather is analyzed, and the resilience evaluation index considering load importance and system performance is proposed. Secondly, the inner optimization model is constructed with the goal of minimizing the node voltage deviation, load loss and power generation cost of the post-disaster distribution network. Then, the resilience improvement strategy of post-disaster distribution network is transformed into a multistage Markov decision process, and an outer reinforcement learning optimization model is constructed. The switching instruction of tie-line switch is taken as the control variable, and the maximization of load recovery is taken as the optimization objective, and the PPO algorithm is used to solve the problem. Finally, the improved IEEE 33-bus distribution network is used to verify the effectiveness of the proposed method in improving the resilience of the post-disaster distribution network and the convergence of the PPO algorithm.