摘要
针对载流故障的时域多样性,提出基于变尺度主成分分析(PCA)的载流故障早期预警方法。首先构造即时温度序列和多种时间尺度的平均温度序列,然后对各温度序列分别进行主成分分析以提取故障的早期特征,并采用K-means算法对异常温度点进行聚类分析以实现故障定位。实验结果表明,该方法能有效地进行载流故障诊断,并使故障的预警时间比常规的温度阈值法显著提前。
A variable-scale PCA（Principal Component Analysis） based current-carrying fault early warning approach is proposed with respect to its variability in time domain.The real-time temperature series and the moving average temperature series in various time scales are constructed,PCA is applied to each series to detect the early features,and K-means algorithm is then employed in clustering analysis for the abnormal temperature sites to locate the faults.Experiment results show that the proposed method can effectively diagnose the current-carrying faults much earlier than conventional temperature-threshold method.
出处
《电力自动化设备》
EI
CSCD
北大核心
2012年第5期147-151,共5页
Electric Power Automation Equipment
关键词
电力设备
主成分分析
K-MEANS
尺度
载流故障
早期预警
故障检测
监测
故障定位
electric equipments
principal component analysis
K-means
scale
current-carrying fault
early warning
fault detection
monitoring
electric fault location
