基于Boosting RBF神经网络的滚动轴承故障诊断  

Fault Testing on Rolling Bearing Based on Boosting RBF Neural Network

作　　者：龙仙爱[1] 杨顺[1] 夏利民[2] 

机构地区：[1]湖南涉外经济学院计算机系,长沙410205 [2]中南大学信息科学与工程学院,长沙410075

出　　处：《计算机与数字工程》2006年第9期15-18,37,共5页Computer & Digital Engineering

基　　金：国家自然科学基金重大项目(编号79816101);中国科学院知识创新工程领域前沿项目

摘　　要：提出了一种新颖的基于RBF神经网络滚动轴承故障诊断方法。以滚动轴承动态信号的能量信息作为特征,RBF神经网络作为分类器进行滚动轴承故障自动分类与诊断。为了进一步提高神经网络的泛化能力和故障诊断的准确性,采用Boosting方法,进行网络集成。对七类滚动轴承进行了实验,结果表明该方法具有很好的故障诊断效果。In this paper, we present a novel method for fault testing on rolling bearing based on boosting RBF Neural network. First, frequency band is extraetecl from dynamic vibration signals of rolling bearing gained from acceleration meter using wavelet paekets transform. Then, RBF Neural network is used to elass the faults of rolling bearing. In order to improve the precision of the RBF neural network for fault testing, we use Boosting algorithm to build an integration - neural network to test rolling beating. A set of experiments of fault testing on rolling bearing are presented. Experiment results have shown good detective performance of our newly presented method.

关 键 词：滚动轴承 故障诊断 小波包分解 BOOSTING方法 RBF神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]