基于SVM的旋转机械故障诊断方法

提取时域与频域共20个特征参数作为数据样本，选择适合旋转机械振动信号的径向基函数及相关参数，基于一对多法构造支持向量机(SVM)多类分类器，实现旋转机械滚动轴承的故障诊断。通过对振动信号特征进行训练与测试，并与BP神经网络进行对比结果表明，该SVM多类分类器可较好地解决小样本问题，在训练时间和识别正确率上均优于BP神经网络。

Rotating Machinery Fault Diagnosis Method Based on SVM

This paper extracts 20 characteristic parameters of time domain and frequency domain as data sample,chooses Radial Basis function(RBF) and related parameters which are suitable for rotating machinery vibration signal,and constructs a one-against-all Support Vector Machine(SVM) multi-class classifier to identify health status of rolling bearing.Compared with Back-propagation(BP) neural network,the SVM classifier with the vibration features of rolling bearing.Experimental results indicate that the SVM classifier can better solve the problem of small sample,is superior to the BP neural network in the training time and recognition accuracy.