摘要
本文在建立了电机多元时序模型的基础上,通过逐步分析电机故障机理,确定将电机故障分为两大类(机械故障、电气故障).利用模型参数φ_i作为初始模式向量并对其进行K-L变换,剔除两类故障分类信息的相同点,提取其异同点,将较多维的模式向量压缩为几维的模式向量,并利用Fisher分类器,建立分类函数,从而能正确识别电机故障类型.
<ABSTRACT> Based on the established multi-variate time series models of motor,this paper classifies motor faults into two types mechanic and electric through primary analysis of motor fault origination. By using the model parameter as the initial pattern vector and doing K-L transformation of it,whereupon the identical points of the classfication information of the faults are deleted and the u-nidentical extracted,contacting the multi-dimensional pattern vector into sever al-dimensional one, and establish classification function by using Fisher classificator,hence it can recognize correctly fault type of motor.
出处
《长沙铁道学院学报》
CAS
CSCD
1998年第3期67-71,共5页
Journal of Changsha Railway University
关键词
模式向量
模式识别
故障诊断
电机
pattern vector
pattern recognition
fault testing
motor
