面向转子故障诊断的核局部边界Fisher判别方法

Method of kernel local-margin fisher discriminant to rotor fault diagnosis

提出一种基于核的局部边界Fisher判别(KLMFD)算法用核函数将故障特征数据映射到高维核空间,以每点与所有局部邻域点中最远同类数据点和最近异类数据点构成的点对来计算类内散度和类间散度,构建边界局部核Fisher判别函数,求出最优故障识别向量,然后利用该向量对测试特征数据进行故障诊断。转子故障诊断实验表明,对于多传感器振动特征融合信号,KMLFD算法的故障诊断效果最好,当选取合适参数时能完全识别故障类型。

In order to better identify the fault of rotor system, one new method based on kernel local-margin fisher discriminant （KLMFD） is proposed.In this methd, fualt feature data are mapped to high-dimensional space by kernel function, computed with-class scatter and between-class scatte based on the farthest congeneric data point and the recent heterogeneous data point of each point in all the local neighborhood, constructed kernel local magin fisher discriminant function, found optimal fault diagnosis vector. Then fault of new testing data are identified by this vector.The experiment showed, KLMFD algorithm had best effect in comparison to other manifold learning algorithm to the rotor fault diagnosis, and can fully identify fault type when selecting the appropriate parameters.