基于双树复小波和深度信念网络的轴承故障诊断  

Bearing Fault Diagnosis Based on DTCWT and DBN

作　　者：张淑清[1] 胡永涛[1] 姜安琦[2] 李军锋[1] 宿新爽 姜万录[3] ZHANG Shuqing1 HU Yongtao1 JIANG Anqi2 LI Junfeng1 SU Xinshuang1 JIANLi Wanlu3. 1.Institute of Electrical Engineering Yanshan University Qinhuangdao Hebei 066004 2.School of Information Science and Engineering Central South University Changsha 410006 3.Institute of Mechanical Engineering

机构地区：[1]燕山大学电气工程学院,秦皇岛066004 [2]中南大学信息科学与工程学院,长沙410006 [3]燕山大学机械工程学院,秦皇岛066004

出　　处：《中国机械工程》2017年第5期532-536,543,共6页China Mechanical Engineering

基　　金：国家自然科学基金资助项目(51475405;61077071);河北省自然科学基金资助项目(F2015203413;F2016203496;F2015203392)

摘　　要：提出了一种基于双树复小波(DTCWT)和深度信念网络(DBN)的轴承故障诊断新方法。采用DTCWT对轴承振动信号进行分解实验,结果表明DTCWT能够很好地将信号分解到不同频带。进而提取能量熵作为故障特征,采用DBN小样本分类模型对轴承故障进行分类,并与传统分类器进行比较,结果表明该方法能准确识别不同故障类型,扩展了DBN在机械故障诊断中的应用。Based on DTCWT and DBN, a new method of bearing fault diagnosis was proposed. Ex- periments on bearing vibration signals decomposition show that the signals may be well decomposed into different frequency bands by DTCWT. Then, power entropy of different frequency bands were taken as the fault features and input to the model for classification and the traditional classifiers were taken as the comparison. Results show that the method may identify different fault types accurately, which expands the applications of DBN.

关 键 词：双树复小波 深度信念网络 受限波尔兹曼机 故障诊断 

分 类 号：TN911.6[电子电信—通信与信息系统]