基于动态贝叶斯网络的轴承故障在线诊断方法  

Bearing Fault Online Diagnosis Method Based on Dynamic Bayesian Network

作　　者：王泽渊 宋仁旺[1] 石慧[1] WANG Ze-yuan;SONG Ren-wang;SHI Hui(School of Electronic Information Engineering,Taiyuan University of Science and Technology,Taiyuan 030024,China)

机构地区：[1]太原科技大学电子信息工程学院,太原030024

出　　处：《太原科技大学学报》2023年第4期303-308,共6页Journal of Taiyuan University of Science and Technology

基　　金：国家自然科学基金青年科学基金(61703297);山西省自然科学基金(201801D121166,201901D111259);太原科技大学校博士启动基金(20162029)。

摘　　要：为了实现滚动轴承的实时在线监测与诊断,充分利用监测数据时间序列时间片之间的相关性,提出基于动态贝叶斯网络(DBN)的轴承在线故障诊断方法。首先进行轴承故障分析和特征参数的计算,使用皮尔逊相关系数法去除冗余特征参数,并将保留的参数作为DBN的观测节点;然后采用基于依赖分析的算法建立网络的拓扑结构,使用最大期望(EM)算法学习网络参数,建立DBN在线监测与诊断模型;最后采用所提方法对西安交通大学XJTU SY滚动轴承故障数据进行仿真分析,故障检测率为99.58%,诊断准确率为98.75%,验证了所提方法对滚动轴承在线故障诊断的有效性。In order to realize real-time online monitoring and diagnosis of rolling bearings,and make full use of the correlation between the time series of monitoring data,an online fault diagnosis method based on dynamic Bayesian network(DBN)is proposed.First,bearing fault analysis and calculation of characteristic parameters are carried out,Pearson correlation coefficient method is used to remove redundant characteristic parameters,and the retained parameters are used as observation nodes of DBN;then the algorithm based on dependency analysis is used to establish the network topology and the maximum expectation(EM)algorithm is used to learn network parameters and DBN online monitoring and diagnosis model is established;finally,the proposed method is used to simulate and analyze the fault data of XJTU SY rolling bearing of Xi’an Jiaotong University.The fault detection rate is 99.58%,and the diagnosis accuracy rate is 98.75%,which is verified the effectiveness of the proposed method for online fault diagnosis of rolling bearings.

关 键 词：动态贝叶斯网络 EM算法 在线监测 故障诊断 

分 类 号：TP133[自动化与计算机技术—控制理论与控制工程]