基于IGA-BP神经网络滚动轴承故障的诊断  

Fault Diagnosis of Rolling Bearings Based on An IGA-BP Neural Network

作　　者：赵永满[1] 梅卫江[1] 王春林[1] 

机构地区：[1]石河子大学机械电气工程学院,石河子832003

出　　处：《石河子大学学报（自然科学版）》2010年第3期379-382,共4页Journal of Shihezi University(Natural Science)

基　　金：新疆生产建设兵团农业机械重点实验室基金项目

摘　　要：为提高滚动轴承故障模式识别技术的研究,基于IGA-BP神经网络的故障诊断原理,运用IGA对BP神经网络的权值与阈值进行调整和优化,利用小波包分解获得轴承振动信号的特征向量,进行了滚动轴承故障的诊断实验研究,对故障模式进行识别。结果表明,IGA-BP神经网络方法具有很强的故障识别能力,说明利用IGA-BP神经网络方法进行轴承故障诊断是可行的。A method and principle of fault diagnosis based on an improved genetic algorithm（IGA） and back-propagation neural network is studied.A fault diagnosis experiment is carried out through tuning the weight and threshold of BP NN through the IGA and extracting the feature vectors of the roller bearing signals on the basis of wavelet analysis,and the fault models are recognized.The conclusion indicats that the combination method of the IGA and BP NN possess good capability of fault diagnosis identification.The results show that it is feasible to implement fault diagnosis of rolling bearings with the IGA and BP NN.

关 键 词：滚动轴承 故障 诊断 BP神经网络 改进遗传算法 

分 类 号：TH17[机械工程—机械制造及自动化]