基于小波包变换的径向基神经网络在故障诊断中的应用

提出了一种新的旋转机械故障诊断方法。基于小波包变换的频率划分特性，对旋转机械的振动信号进行小波包分解，建立旋转机械六种典型故障特征矢量，准确地提取了故障的特征信息，结合RBF神经网络训练速度快的优点，将RBF神经网络应用于故障特征的选择，最后，利用所确定特征及RBF分类器进行故障诊断。实验结果表明，该方法可实现典型故障的可靠诊断。而且由于利用小波包变换代替了传统的FFT，故本方法对于诊断频率分布范围较广而复杂且信号具有较强时变性的复杂故障有着良好的应用前景。

Application of Radial Basis Function Network based on Wavelet Packet Transform to Fault Diagnosis

A new diagnosis method of rotating machinery is presented in the paper. The procedures of the method are that the first the vibration signals of a rotating machine are decomposed by using a wavelet package so the characteristics of frequency partition are obtained; then characteristic vectors of six typical faults of rotating machine are build; and feature information of the faults is correctly extracted by using a RBF network; finally, the faults of the machine are diagnosed on the base of the established characteristic vectors and the RBF classifier. The experimental results demonstrate that the method can be applied into reliable diagnosis of typical faults of rotating machinery. Because wavelet package transform is used instead of traditional FFT, the method has fair prospects of application for the complex faults where frequency components are complex and wider and fault signals are varying with time.