电工技术学报  2019, Vol. 34 Issue (11): 2405-2416    DOI: 10.19595/j.cnki.1000-6753.tces.180669
电力系统 |
基于离群点检测的高压并联电抗器本体电流互感器测量异常故障在线诊断
滕予非1, 吴杰1, 张真源2, 姜振超1, 黄琦2
1. 国网四川省电力公司电力科学研究院 成都 610041;
2. 电子科技大学机械与电气工程学院 成都 611731
Online Identification of Measurement Abnormality Fault Based on Outlier Detection for Current Transformer in High Voltage Shunt Reactor
Teng Yufei1, Wu Jie1, Zhang Zhenyuan2, Jiang Zhenchao1, Huang Qi2
1. State Grid Sichuan Electric Power Research Institute Chengdu 610041 China;
2. School of Mechanical and Electrical Engineering University of Electronic Science and Technology of China Chengdu 611731 China
全文: PDF (1187 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 为提升高压并联电抗器过电流在线监测系统的准确度,提出了基于离群点检测的高压并联电抗器本体电流互感器测量异常故障在线诊断方法。首先,利用Sigmoid函数对母线特定电压条件下电抗器过电流报警信号条件概率与母线电压有效值间的函数关系进行拟合,并选取函数中心点及不确定域作为特征指标,构成二维关键特征数据点(KCDP),用以进行故障诊断。其次,将正常工况下获得的多个关键特征数据点以及检测日关键特征数据点构成诊断数据集,利用孤立森林算法对其进行离群点检测,计算集合中每个关键特征数据点的异常分数。最后,假设正常工况下关键特征数据点的异常分数满足威布尔分布,基于分布的置信区间给出了高压并联电抗器本体电流互感器测量异常故障的诊断判据。某实际500 kV高压并联电抗器算例,验证了方法的有效性。
服务
把本文推荐给朋友
加入我的书架
加入引用管理器
E-mail Alert
RSS
作者相关文章
滕予非
吴杰
张真源
姜振超
黄琦
关键词 高压并联电抗器离群点检测孤立森林算法条件概率威布尔分布    
Abstract:In order to increase the accuracy of the overcurrent online monitor of shunt reactor, an online identification method of Measurement Abnormality Fault for Current Transformer (CT) of high voltage shunt reactor based on outlier detection is proposed. Firstly, sigmoid function is utilized to fit the relationship between conditional probability of overcurrent alarms and the RMS of the voltage in the bus. And the center point and uncertain area of the function are chosen as the indexes to generate the key characteristic data point (KCDP). Secondly, isolation forest algorithm is utilized to calculation anomaly score of diagnosis dataset, which includes KCDP under normal condition and that in detection day, in order to detect the outlier point. Finally, supposed the anomaly scores under normal condition meet the Weibull distribution, the judgment of the identification of measurement abnormality fault for CT is proposed. The effectiveness of the method is verified by case study of a practical 500kV high voltage shunt reactor.
Key wordsHigh voltage shunt reactor    outlier detection    isolation forest algorithm    conditional probability    Weibull distribution   
收稿日期: 2018-04-23      出版日期: 2019-06-14
PACS: TM45  
基金资助:国家电网总部项目(变电站自动化设备广域运维技术研究及应用)和四川省科技厅重点研发计划项目(2017GZ0163)资助
通讯作者: 吴杰 男,1986年生,硕士,高级工程师,研究方向为电力系统大数据应用等。E-mail:497564636@qq.com   
作者简介: 滕予非 男,1984年生,博士,高级工程师,研究方向为特高压交直流混联电网安全分析与控制等。E-mail:yfteng2011@163.com
引用本文:   
滕予非, 吴杰, 张真源, 姜振超, 黄琦. 基于离群点检测的高压并联电抗器本体电流互感器测量异常故障在线诊断[J]. 电工技术学报, 2019, 34(11): 2405-2416. Teng Yufei, Wu Jie, Zhang Zhenyuan, Jiang Zhenchao, Huang Qi. Online Identification of Measurement Abnormality Fault Based on Outlier Detection for Current Transformer in High Voltage Shunt Reactor. Transactions of China Electrotechnical Society, 2019, 34(11): 2405-2416.
链接本文:  
https://dgjsxb.ces-transaction.com/CN/10.19595/j.cnki.1000-6753.tces.180669          https://dgjsxb.ces-transaction.com/CN/Y2019/V34/I11/2405