Rapid service restoration of the distribution network after a disaster is of great significance to the reliability of the power system.Traditional methods model this problem as a constrained optimization problem,which is very complex and computa-tionally expensive.The service restoration problem of the distribution network is modeled as a Markov decision process.An algo-rithm based on deep reinforcement learning is proposed,in which an intelligent agent is trained on simulated power system data to acquire the decision-making capability of service restoration and network reconfiguration.Experiments are performed on a 30-node distribution network testcase to analyze and verify the capability of the intelligent agent in simulated fault scenarios,and the results demonstrate the effectiveness of the algorithm.
维普
万方数据