装备故障诊断方法研究  

Research on the Equipment Fault Diagnosis Method

作　　者：崔学忠[1] 李冬[2] 桑亮[3] 李馥彤[2] 

机构地区：[1]92941部队 [2]海军航空工程学院研究生管理大队 [3]91245部队

出　　处：《计算机仿真》2010年第9期181-185,共5页Computer Simulation

摘　　要：研究装备故障预报,便于及时维修,选取诊断方法对于高效的故障诊断具有重要意义,因此概括了故障诊断的方法。为了提高诊断率,保证运行安全,采用小波分析和回归型支持向量机两种先进的故障诊断方法,仿真结果表明两种方法都能准确定位故障点,用时大致相同,但支持向量机同时实现原始信号的重构,而且残差比小波滤波的信号噪声还要小,可以优先采用,对故障诊断方法的选取具有指导意义。采用人工神经网络方法较好地解决了齿轮故障问题,并且诊断结果是准确的。Nowadays,selecting diagnosis method has a great meaning in effective fault diagnosis. The paper summarizes the fault diagnosis method,introduces wavelet analysis and SVR fault diagnosis method and compares the two methods in detail. The results indicate that both methods can locate the fault point exactly,and the diagnosis time is basically the same. But SVR realizes the reconstruction of original signal,and residual is smaller than the noise of wavelet filter,therefore,it should be adopted preferably,which has directive meaning for the selection of fault diagnosis method. For the unsettled diagnosis problem of traditional experience method,neural network settles the problem preferably,and the diagnosis is done with higher accuracy.

关 键 词：故障诊断 回归型支持向量机 小波分析 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]