基于ICA-EMD与SVM的滚珠丝杠故障诊断

Study on ball screw fault diagnosis based on ICA-EMD and SVM

针对滚珠丝杠振动信号的非平稳性,采用经验模态分解(EMD)进行信号处理。而传感器采集到的振动信号存在混叠现象,EMD分解能力受限,因此采用独立成分分析方法,重新分离三个振动源信号,对得到的新振动信号进行EMD处理,最后对各IMF成分进行能量分析,结合IMF能量及其他特征值,利用支持向量机(SVM)实现智能故障诊断。

To non-stationary vibration signal of the ball screw,this paper used empirical mode decomposition(EMD) for signal processing. Because vibration signal collected by sensors has aliasing phenomenon,the EMD decomposition ability is limited,thus using independent component analysis method to separate three vibration source signals. Finally,getting energy analysis of each IMF component,in combination with the IMF and other energy eigenvalue,this paper realized the intelligent fault diagnosis through support vector machine(SVM).