电力系统自动化2025，Vol.49Issue(17)：101-111,11.DOI:10.7500/AEPS20241113003
基于电气坐标的电力系统动态安全分区评估及模型更新
Dynamic Security Partitioning Assessment and Model Updating of Power System Based on Electrical Coordinates
摘要
Abstract
The dynamic security assessment(DSA)of power systems considering the operation mode and fault location requires a large number of training samples,and the structure of assessment model is complex,which has the problems of high training costs and difficulty in online updates.To solve these problems,a method for dynamic security partitioning assessment and model increment updating of power system is proposed,which reduces the complexity of the assessment model and enables online model increment updating without catastrophic forgetting.Firstly,electrical coordinates are constructed as node position features,and the Chebyshev distance is introduced to measure the difference between node position features,and the adaptive partitioning of the power grid is carried out based on the space constrained mean shift(SCMS)clustering algorithm.Secondly,the operation modes and fault location features are extracted,and the Chebyshev distance is embedded in the double-hidden-layer radial basis function neural network(RBFNN).Partitioning assessment models for various fault locations in various regions are constructed.Finally,an online model increment updating method is proposed to avoid the problem of catastrophic forgetting during the model updating process.The effectiveness of the proposed method is verified using a transient power angle stability assessment of a provincial power grid in China as an application scenario.关键词
动态安全评估/分区/电气坐标/切比雪夫距离/神经网络Key words
dynamic security assessment/partitioning/electrical coordinate/Chebyshev distance/neural network引用本文复制引用
齐航,李常刚,刘玉田,闫炯程..基于电气坐标的电力系统动态安全分区评估及模型更新[J].电力系统自动化,2025,49(17):101-111,11.基金项目
国家自然科学基金资助项目(52177096). This work is supported by National Natural Science Foundation of China(No.52177096). （52177096）
