摘要
介绍了一种利用二维谱图的形状统计特征进行放电模式识别的方法。利用发电机线棒工业仿真模型进行局部放电信号检测,放电信号来自四种不同的故障模式。分别使用基于距离的模式归类法和前馈网络进行模式识别,根据统计特征对放电模式的描述能力和两种识别方法的分类能力进行了分析比较。结果表明统计识别方法的分类效果是令人满意的。
This paper presents the results on the use of statistical operators for recognition of 2-dimensional discharge patterns. Discharge signal is measured from four kinds of the industrial models of stator winding bars. Discharge patterns with different types and intensities are obtained. The distance based methods and back-propagation network are used for recognition. The description capabilities of statistical operators and the discrimination abilities of above two classifying algorithms on discharge patterns are compared. The results indicate that statistical operators are effective in pattern description and back propagation network is satisfied in classification.
出处
《电工技术学报》
EI
CSCD
北大核心
2006年第4期41-45,共5页
Transactions of China Electrotechnical Society
关键词
局部放电
模式识别
模式归类法
人工神经网络
Partial discharge, pattern recognition, distance-based classifier, artificial neural network