基于K-L变换和支持向量机的滚动轴承故障模式的识别

提出了应用K-L变换和支持向量机相结合进行滚动轴承故障诊断的方法。K-L变换可以将高维相关变量压缩为低维独立的主特征向量,而支持向量机可以完成模式识别和非线性回归。利用上述原理根据轴承振动信号的变化特征,采用K-L变换对其提取状态主特征向量,然后利用建立的支持向量机多故障分类器完成滚动轴承故障模式的识别。试验结果表明,K-L变换分解后的主特征向量与支持向量机相结合可以有效地、准确地识别轴承的故障模式,为轴承故障诊断向智能化发展提供了新的途径。

Fault Pattern Recognition of Rolling Bearing Based on K-L Transform and Support Vector Machine

A fault diagnosis approach for the rolling bearing based on the K-L transform(KLT) and support vector machine(SVM) was proposed. The KLT can transform a multidimensional correlated variable into a less dimensional independent eigenvector, and SVM enables the pattern recognition and nonlinear regression. Based on these concepts, the KLT was employed to extract the state eigenvector from the bearing vibration signals, then the SVM classifier was established to recognize the fault pattern. Experimental results showed that the combination of the eigenvector decomposed by the KLT with the SVM is able to recognize the fault patterns of the rolling bearing effectively and accurately, providing a new approach to the development of fault diagnosis intelligentization.