极端事件中基于深度强化学习的电力系统主动控制策略

打开文本图片集

【Abstract】 In industrial power grids，extreme events such as wildfires require massive line and load switching operations.Human operators often struggle to make reliable decisions under such high-pressre conditions to minimize system impact and losses.To address this issue， the paper proposes a Deep Reinforcement Learming （DRL）-assted proactive control strategy for power systems during extremeevents.First，the proactive control problem is modeled as a Markov Decision Process （MDP），along with the development of a wildfire spread model and a power system operation model.Then，basedontheoperationalcharacteristicsofthepowergridduringwildfireevents，statespacetransformation， optimized action execution，andcustomized reward functions are conducted to improve the efciencyofsubsequentDRL training.FinalytheDeepDeterministicPolicyGradient （DDPG）algorithminDRLisemployedtotraintheagenttoassist proactive control in minimizing load interruptions.Simulation experiments on the IEEE24-bus test system demonstrate that the proposed method can effectively reduce load interruptions during extreme events.

【Key words】 power systems; intelligent control; proactive control; Deep Reinforcement Learning; resilience [中图分类号】TM74 [文献标志码]A [文章编号]1674-3229（2025）03-0091-06

0 引言

随着全球气候变暖，极端自然灾害对电力系统的威胁日益加剧[1-2]。（剩余8257字）