基于容积卡尔曼滤波残差驱动的动态电力系统态势感知方法研究  

A Situational Awareness Method for Dynamic Power System Based on CKF Residual Driven

作　　者：冒波波 施昱青 彭飞 MAO Bobo;SHI Yuqing;PENG Fei(Shibei Power Supply Company of State Grid Shanghai Electric Power Company,Shanghai 200072,China;College of Electrical Engineering,Qingdao University,Qingdao 266071,China)

机构地区：[1]国网上海市电力公司市北供电公司,上海200072 [2]青岛大学电气工程学院,山东青岛266071

出　　处：《供用电》2020年第8期39-46,共8页Distribution & Utilization

基　　金：国家电网有限公司科技项目(52091419003H)。

摘　　要：提出了一种基于容积卡尔曼滤波(cubature Kalman filter,CKF)残差驱动的动态电力系统异常检测框架。首先,基于集中式自适应迭代CKF实现动态电力系统的多节点电压状态的迭代估计,通过充分利用相量测量单元的量测信息,实现对时变量测噪声的迭代估计,从而有效消除正常条件下系统节点电压的时空相关性,在随机系统高斯噪声假设下能够显著增强多节点电压残差的随机分布特性。在此基础上,系统非线性扰动/异常的影响可以通过系统电压残差矩阵特征值的变化合理表征。然后,提出了一种基于残差协方差变换的系统运行态势评估指标来量化上述系统非线性扰动/异常的影响。最后,基于典型光伏电站日出力曲线的IEEE 39节点测试系统算例,通过节点电压异常扰动注入实验分析对比,表明了该方法对于动态电力系统运行态势异常感知的有效性。A framework for dynamic power system anomaly detection based on cubature Kalman filter(CKF)driven highdimensional residual matrix analysis is proposed in this paper.Firstly,coordinated multi-node voltage state estimation of the dynamic power system can be achieved by the centralized adaptive CKF in which time-varying noise can be iteratedly evaluated based on the streaming PMUs datasets.Therefore,the spatial-temporal relationship among grid nodes under system normal conditions can be eliminated and the associated random characteristics can be enhanced under the noise assumption of Gaussian distribution.On this basis,the effects of system nonlinear disturbances/abnormalities can be reasonably characterized by the change of eigenvalues of streaming voltage residual matrix.Furthermore,a situational awareness indicator based on residual covariance transformation is proposed to quantify the effect of the abnormal condition,and its awareness performance is derived.Finally,the proposed method is validated based on the IEEE-39 node test system integrated with typical daily power curve of photovoltaic power plant and abnormal voltage disturbance injection.The comparisons of the derived indicator and other indicators show that the proposed architecture and the derived indicator is more appropriate for the situation awareness of dynamic power system.

关 键 词：容积卡尔曼滤波 高维随机矩阵分析 动态电力系统 态势感知 状态估计 

分 类 号：TM71[电气工程—电力系统及自动化]