基于AIC的RBF网络在旋转机械故障诊断中的应用  

The application of RBF neural network based on AIC for the fault diagnosis on rotary machine

作　　者：黄强[1] 丁志华[1] 张晓[1] 

机构地区：[1]九江学院,九江332005

出　　处：《机械设计与制造》2011年第10期69-71,共3页Machinery Design & Manufacture

基　　金：江西省自然科学基金项目(2008GQC0002);江西省教育厅科技资助项目(GJJ08448)

摘　　要：提出了一种适用于各类旋转机械的精度高、响应快的故障诊断方法。首先利用赤迟信息准则(AIC)确定RBF神经网络的隐含层节点数和中心,然后建立基于该神经网络的故障诊断模型,再利用旋转机械振动信号来分析并判定其运行状况。以柴油机连杆铜套磨损故障为例进行分析,实验及仿真结果表明:基于AIC的RBF网络技术的诊断方法在柴油机故障诊断中是有效可行的,对于单一故障的识别率为100%,对其他复杂旋转机械的振动诊断同样具有参考价值。A diagnosis method with high precision and quickly response for rotary machines is proposed.Firstly,the AIC is used to calculate the number and center of nodes of the RBF neural network.Then the fault diagnosis model is built.At last,the vibration signals is used to estimate the states of rotary machines.The bushing abrasion of diesel engine is analysed,according to experiment and simulation researches,it indicates that the diagnosis method with RBF neural network based on AIC is effective.The veracity of identification is 100% to the single fault.It is a valuable reference to the vibration diagnosis for other complex rotary machines.

关 键 词：旋转机械 柴油机 赤迟信息准则 故障诊断 

分 类 号：TH16[机械工程—机械制造及自动化] TK42[动力工程及工程热物理—动力机械及工程]