基于证据融合理论的多不良数据辨识  

Multi Bad Data Identification Based on Evidence Fusion Theory

作　　者：卢志刚[1] 程慧琳[1] 冯磊[1] 杨丽君[1] 

机构地区：[1]电力电子节能与传动控制河北省重点实验室(燕山大学),河北省秦皇岛市066004

出　　处：《电网技术》2012年第1期123-128,共6页Power System Technology

基　　金：国家自然科学基金项目(61071201);河北省自然科学基金项目(F2010001319)~~

摘　　要：为更好地克服残差污染和残差淹没,实现多不良数据的辨识,引入电气距离、节点相关系数和灵敏度作为证据,运用证据融合理论确定测量关联度,测量关联度可反映测量数据出现残差污染和残差淹没的可能性。对于残差较大且测量关联度较小的数据,发生残差污染的可能性较小,可直接辨识为不良数据;对于残差较大且测量关联度较大的数据,采用模糊聚类方法,隔离不良数据,对系统进行分区,并逐步修正不良数据,选择可靠测量数据重新进行状态估计。算例结果验证了该方法的有效性。To overcome the occurrence of residual pollution and residual submerge in multi bad data identification effectively, Electric distance, correlation coefficient and sensitivity as evidences are introduced. The measurement correlative degree is determined by evidence fusion theory which can reflect the possibilty of residual pollution and submerge occurrence in measured data. There is smaller possibility of residual pollution occurrence in the data with larger residual and lower measurement correlative degree, so the data can be directly identified as bad data; for the data with larger residual and higher measurement correlative degree the bad data should be isolated by fuzzy clustering, then the system should be partitioned and bad data is modified step by step, and then reliable measured data can be chosen for the state reestimation. The effectiveness of the proposed method is verified by simulation results of IEEE 14-bus system.

关 键 词：残差污染 残差淹没 数据辨识:证据融合 测量关联度 

分 类 号：TM744[电气工程—电力系统及自动化]