桂林理工大学学报2016，Vol.36Issue(4)：823-829,7.DOI:10.3969/j.issn.1674-9057.2016.04.030
基于聚类算法的脉冲电压绝缘老化实验漏电流检测方法
Leakage current detection method in insulation aging experiment under pulse power based on cluster algorithm
摘要
Abstract
The leakage current is taken as discrimination standard for insulation failure in insulation accelerated experiment.In order to obtain the leakage current by removing the capacitor transient current from the loop current detection signal,a cluster analysis algorithm that makes decision by use of the distance between neighbor sampling points is designed.Capacitor transient current is mainly concentrated in the time of square pulse edge'0 ～5τ in the high-frequency bipolar pulse power accelerated aging experiment.The transient process is finished basically within 10τ and the loop current of sample may be regarded as the leakage current after that.Based on the current variety process and the corresponding numerical relations,cluster analysis can classify the leakage current data from the sampling period data and calculate the mean value of the leakage current data.The simulation results prove that on under this conditin the sampling periods are more than 50,the biggest relative error of the acquired leakage current by the cluster analysis algorithm is 2％.Thus it can eliminate the effect of capacitor transient process current and random noise on detection of leakage current effectively,meet the requirement of accelerated aging experiment under pulse electrical stress well and improve the reliability of the experiment result.关键词
漏电流检测/聚类分析/绝缘寿命Key words
leakage current detection/cluster analysis/insulation aging分类
信息技术与安全科学引用本文复制引用
蒋存波,王西乐,焦阳,杨家志,王怀红,孔祥丽..基于聚类算法的脉冲电压绝缘老化实验漏电流检测方法[J].桂林理工大学学报,2016,36(4):823-829,7.基金项目
国家自然科学基金项目(51167004) （51167004）
广西自然科学基金项目(2011GXNSFA018022 （）
2013GXNSFBA019250) （）
专利支持项目(KY2015ZL100) （KY2015ZL100）
