电网技术2024，Vol.48Issue(4)：1519-1531,中插41,14.DOI:10.13335/j.1000-3673.pst.2022.2450
基于强化学习的混合元启发式暂态电压稳定特征选择方法及可解释性研究
Reinforcement Learning-based Hybrid Element Heuristic Transient Voltage Stability Feature Selection and Its Interpretability
摘要
Abstract
Under the development of new power systems,it is of great significance to extract the key response features strongly related to the stability of transient voltage with an effective feature selection for the studies of the mechanism of transient voltage instability and the potential security risks of the system.Therefore,a new feature selection method is proposed based on the composite framework of the improved filtering method and the hybrid element heuristic packaging method.The improved Max-Relevance and Min-Redundancy criterion of symmetric uncertainty value is firstly used to have a coarse screen of the features.Then the Q-learning reinforcement learning is integrated into the meta-heuristic optimization algorithm,and the exploitation and exploration compromise strategy is used to enhance the feature fine selection ability to obtain the optimal critical response feature subset.On this basis,the Shapley additive explanation is applied to comprehensively analyze the influences of each of the screening features on the transient voltage stability and the weak links of the system.The effectiveness of the proposed method is verified by an example of a new power system.关键词
暂态电压稳定/特征选择/强化学习/混合元启发式/沙普利值加性解释Key words
transient voltage stability/feature selection/reinforcement learning/hybrid element heuristics/Shapley additive explanation分类
信息技术与安全科学引用本文复制引用
甄永赞,阮程..基于强化学习的混合元启发式暂态电压稳定特征选择方法及可解释性研究[J].电网技术,2024,48(4):1519-1531,中插41,14.基金项目
国家重点研发计划项目(响应驱动的大电网稳定性智能增强分析与控制技术)(2021YFB2400800).Project Supported by the National Key R&D Program of China(Response-driven Intelligent Enhanced Analysis and Control for Bulk Power System Stability)(2021YFB2400800). （响应驱动的大电网稳定性智能增强分析与控制技术）
