山东电力技术2026，Vol.53Issue(6)：1-12,12.DOI:10.20097/j.cnki.issn1007-9904.250666
考虑台风时空演变的配电网灾前-灾中应急调度方法
Emergency Scheduling Method for Distribution Networks Considering Spatiotemporal Evolution of Typhoons in Pre-disaster and During-disaster Stages
摘要
Abstract
Typhoon disasters severely impact the operation of distribution networks,presenting challenges such as strong randomness in prediction,difficulties in pre-disaster and in-disaster coordinated scheduling,and high computational complexity.This paper proposes an emergency scheduling method for distribution networks that considers the spatiotemporal evolution of typhoons,covering both pre-disaster and in-disaster phases.First,calculate the line failure probability based on typhoon weather forecast information from the previous day,and generate a probabilistic failure scenario set under typhoon conditions.Then,a"pre-disaster and in-disaster coordinated scheduling"model for distribution networks is established.The pre-disaster preventive scheduling model aims to maximize the expected load supply under various failure scenarios,optimizing the pre-deployment of emergency power vehicles and the operation mode of the distribution network.The in-disaster real-time response model aims to maximize load supply under real-time failures during the typhoon,adjusting the pre-deployment of emergency power vehicles.For model solving,considering the need for rapid intraday solutions,the soft actor-critic(SAC)reinforcement learning algorithm is employed to solve the in-disaster real-time response model.To enhance convergence,the Bayesian optimization algorithm is introduced to optimize the hyperparameters of the SAC algorithm,and a meta-learning mechanism is embedded to quickly address new sudden failures during typhoons.Finally,case study results demonstrate that the proposed method offers significant advantages in load supply and scheduling efficiency,effectively improving the emergency response capability of distribution networks under typhoon disasters.关键词
配电网应急调度/台风灾害/SAC强化学习/BO优化/元学习机制Key words
distribution network emergency scheduling/typhoon disasters/SAC reinforcement learning/Bayesian optimization/meta-learning mechanism分类
信息技术与安全科学引用本文复制引用
田江,许会广,徐秀之,张思源,林恒先..考虑台风时空演变的配电网灾前-灾中应急调度方法[J].山东电力技术,2026,53(6):1-12,12.基金项目
国网江苏省电力有限公司科技项目(J2024177).Science and Technology Foundation of State Grid Jiangsu Electric Power Company(J2024177). （J2024177）
