电力系统及其自动化学报2025，Vol.37Issue(8)：38-48,11.DOI:10.19635/j.cnki.csu-epsa.001525
基于缺失数据插补和改进集成策略的电力系统暂态稳定评估
Transient Stability Assessment of Power System Based on Missing Data Imputation and Improved Ensemble Strategy
摘要
Abstract
A transient stability assessment method based on missing data imputation and an improved ensemble strategy is proposed to address the issue of decline in the performance of a power system transient stability assessment model due to the potential loss of electrical operation data during the measurement and transmission processes.First,the topo-logical structure of the power system is explicitly modeled by means of a graph recurrent imputation network,the poten-tial spatio-temporal connection between features is learned,and a missing data imputation model is constructed.In addi-tion,a pixel reliability matrix and a feature reliability matrix are generated based on the imputation performance.Sec-ond,an ensemble assessment model is constructed based on an improved stacking method and a closed-form continu-ous-time neural network.Meanwhile,the reliability matrixes are integrated into the training and application processes of this assessment model.Finally,a time adaptive assessment method is introduced to alleviate the computational pres-sure caused by the simultaneous application of the imputation and assessment models.The specific results of a New Eng-land 39-node power system show that the proposed method has strong robustness and maintains a high assessment level in the case of multiple types of data missing.关键词
缺失数据/暂态稳定评估/图循环插补网络/可信度矩阵Key words
missing data/transient stability assessment/graph recurrent imputation network/reliability matrix分类
信息技术与安全科学引用本文复制引用
李欣,吴凌霄,李新宇,赵伟杰,李阳,郭攀锋..基于缺失数据插补和改进集成策略的电力系统暂态稳定评估[J].电力系统及其自动化学报,2025,37(8):38-48,11.基金项目
国家自然科学基金资助项目(52107107). （52107107）
