旋转机械故障诊断的模糊神经网络方法研究.pdfVIP

第八届全国振动理论及应用学术会议论文集，上海，2003 年 11 月 旋转机械故障诊断的模糊神经网络方法研究 1 2 1 魏春荣 ，孙明 ，刘树林 （1.大庆石油学院，大庆 163318；2.哈尔滨铁路局齐齐哈尔工程总公司，齐齐哈尔 161000） 摘 要： 针对模糊分类和神经网络在旋转机械故障诊断中单独应用所存在的弊端，尤其是模糊分类应用 对象受限制的问题，提出了一种应用小波包和模糊神经网络 （ANFIS）进行故障诊断的方法。首先利用小波 包对旋转机械故障的振动信号进行特征提取，特征提取后构造的特征向量能突出反映旋转机械的故障特征， 再通过模糊神经网络(ANFIS)进行故障诊断。在本文中分别运用 BP 网络、模糊神经网络和模糊分类系统对 旋转机械中几种典型的故障进行了诊断和分析，并将它们的诊断结果进行了对比。对比结果表明：ANFIS 模糊神经网络系统具有更精确的诊断准确率，能够弥补模糊和神经网络单独应用所存在的不足，在旋转机械 故障诊断中具有较好的应用前景。 关键词： 故障诊断；小波分解；模糊分类；神经网络；模糊神经网络 The Research of Fuzzy Neural Network Method for Rotating Machinery Fault Diagnosis 1 2 1 WEI Chun-rong ，SUN Ming ，LIU Shu-lin (1.Daqing Petroleum Institute, Daqing 163318; 2.Qiqiha’er Engineering Corporation, Ha’erbin Railway Bureau, Qiqiha’er 161000) Abstract: In this paper, a method of fault diagnosis using wavelet package and fuzzy neural network is proposed to solve the shortcoming of the single-handed application of fuzzy classification or neural network in rotating machinery fault diagnosis, especially the problem that application object of fuzzy classification is restricted. Firstly, the characters of the vibration signals in rotating machinery fault will be extracted by using wavelet package, the characteristic vector constructed after character extraction can reflect the failure symptoms of rotating machinery, then the fault diagnosis can be done through fuzzy neural network. BP network, fuzzy neural network and fuzzy classification are separately used for the diagnosis and analysis of several typical faults in rotating machinery in this paper, and finally their diagnosis resu