ICIM-DBN多测点融合故障诊断及在电机上应用  

ICIM-DBN Multi-Point Fusion Fault Diagnosis and its Application on Motor

作　　者：杨涛[1] 袁荷伟[1] 宋丹丹[2] 程辉 YANG Tao;YUAN He-wei;SONG Dan-dan;CHENG Hui(School of Automobile,He'nan Transportation Vocational and Technical College,He*nan Zhengzhou 450005,China;School of Automotive and Mechanical Engineering,Changsha University of Science and Technology,Hu'nan Changsha 410076,China;Zhengzhou Yutong Bus Co.,Ltd.,He'nan Zhengzhou 450061,China)

机构地区：[1]河南交通职业技术学院汽车学院,河南郑州450005 [2]长沙理工大学汽车与机械工程学院,湖南长沙410076 [3]郑州宇通客车股份有限公司,河南郑州450061

出　　处：《机械设计与制造》2023年第4期101-104,109,共5页Machinery Design & Manufacture

基　　金：河南省高等职业学校青年骨干教师培养计划资助项目(2019GZGG034)。

摘　　要：在深度信念网络(DBN)故障诊断模型中进行参数训练时容易出现局部搜索结果,导致DBN故障诊断模型训练效率降低并引起错误诊断的结果。为更加准确诊断复杂机电设备的故障,综合运用DS证据理论和优化深度信念网络故障诊断模型,构建了以优化深度信念网络为基础的多测点故障诊断方法。选择异步电动作为测试对象,完成电机的故障诊断,研究结果表明:所有测点ICIM-DBN故障诊断模型都在30代训练时发生收敛,说明本文设计的混沌免疫算法可以获得全局最优结果。故障诊断准确率都提高到10%,并且轴承故障诊断准确率也可以达到98.6%。利用此方法能够准确分辨故障信号,可以实现对各个测点故障数据的综合判断,从而防止受故障信号衰减影响以及测试误差而降低故障诊断准确率情况。Local search results are easy to appear when parameters are trained in DBN fault diagnosis model,which leads to the decrease of training efficiency of DBN fault diagnosis model and the result of wrong diagnosis.In order to diagnose the faults of complex mechanical and electrical equipment more accurately,a multi-point fault diagnosis method based on optimized deep belief network was established by using DS evidence theory and optimized deep belief network fault diagnosis model.The induction motor was selected as the test object to complete the fault diagnosis of the motor.The research results show that all the ICIM-DBN fault diagnosis models at the test points converge during the 30th generation of training,which indicates that the unproved chaotic immune algorithm in this paper can quickly calculate the global optimal parameters of the model.The accuracy rate of fault diagnosis is improved to 10%,and the accuracy rate of bearing fault diagnosis can also reach 98.6%.This method is used to diagnose multi-point fault signals,and the fault irvfbrmatioTi cf all points can be fully utilized to avoid the decrease of fault diagnosis accuracy due to the attenuation offault signals and sensor test errors.

关 键 词：故障诊断 多测点 深度信念网络 DS证据理论 电机故障 

分 类 号：TH16[机械工程—机械制造及自动化]