基于近邻传播算法的负荷不良数据辨识  

Identification of Bad Load Data Based on Affinity Propagation Algorithm

作　　者：李山 杨冬 蒋哲 周宁 房俏 李常刚[2] LI Shan;YANG Dong;JIANG Zhe;ZHOU Ning;FANG Qiao;LI Changgang(State Grid Shandong Electric Power Research Institute,Jinan 250003,China;Electrical Engineering School,Shandong University,Jinan 250061,China)

机构地区：[1]国网山东省电力公司电力科学研究院,山东济南250003 [2]山东大学电气工程学院,山东济南250061

出　　处：《山东电力技术》2021年第6期1-5,共5页Shandong Electric Power

基　　金：国家自然科学基金项目(51407107)。

摘　　要：状态估计是电力调度系统中一项重要的基础功能,正确的测量数据是保障状态估计结果准确的关键,不良数据的存在会大大降低估计结果的可信度。为提高状态估计的准确度,提出了一种基于近邻传播(Affinity Propagation,AP)算法的负荷不良数据辨识方法。首先介绍了AP算法的相关原理,给出了基于AP算法进行聚类的一般步骤。然后基于相似性和平滑性两个特征,定义了乘积特征值和最小特征值作为分类依据。最后,以某地区的实际负荷采样数据为算例,验证了所提不良数据辨识方法的有效性。State estimation is an important basic function in the power system dispatching system.Correct measurement data is the key to ensure the accuracy of the state estimation results.The existence of bad data will greatly reduce the reliability of the estimation results.In order to improve the accuracy of state estimation,a bad load data identification method based on affinity propagation(AP)algorithm was proposed in this paper.Firstly,the principle of AP algorithm was introduced,and the general steps of clustering based on AP algorithm were given.Then,the product eigenvalue and the minimum eigenvalue were defined as the classification basis based on two features of the similarity and smoothness.Finally,the effectiveness of the proposed identification method is verified by the actual load sampling data in a certain area.

关 键 词：不良数据辨识 AP算法 聚类 

分 类 号：TM714[电气工程—电力系统及自动化]