基于压缩动态模式分解的滚动轴承健康监测指标构建  

Construction of indicators for monitoring rolling bearing health based on compressed dynamic mode decomposition

作　　者：蔡志鑫 党章 文明[1,2] 吕勇 余震[1,2] Cai Zhixin;Dang Zhang;Wen Ming;Lv Yong;Yu Zhen(Key Laboratory of Metallurgical Equipment and Control of Ministry of Education,Wuhan University of Science and Technology,Wuhan 430081,China;Hubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering,Wuhan University of Science and Technology,Wuhan 430081,China;National Demonstration Center for Experimental Mechanical Education,Wuhan University of Science and Technology,Wuhan 430081,China)

机构地区：[1]武汉科技大学冶金装备及其控制教育部重点实验室,湖北武汉430081 [2]武汉科技大学机械传动与制造工程湖北省重点实验室,湖北武汉430081 [3]武汉科技大学机械国家级实验教学示范中心,湖北武汉430081

出　　处：《武汉科技大学学报》2022年第5期388-393,共6页Journal of Wuhan University of Science and Technology

基　　金：国家自然科学基金资助项目(51575408);湖北省自然科学基金创新群体项目(2020CFA033);武汉科技大学机械传动与制造工程湖北省重点实验室开放基金资助项目(MTMEOF2019B05).

摘　　要：目前轴承健康监测指标主要通过统计模型和机器学习等方法建立,其过程比较繁琐且需要人工干预。动态模式分解(DMD)在振动信号分析中已得到有效应用,但DMD在高采样频率下计算耗时严重且占用的计算空间较大。本文结合压缩感知算法和DMD,提出了基于压缩动态模式分解(CDMD)的滚动轴承健康指标构建方法。首先利用一个低维随机观测矩阵生成压缩的Hankel矩阵,以提高轴承振动信号的奇异值分解速度;然后以模式分解得到的特征值为基础,计算出特征值实部的峭度和模式幅值的均方根值(RMS),从而构成更加简单而有效的轴承健康指标。通过仿真信号和实验信号对本文方法进行验证,结果表明,CDMD可以代替DMD实现信号的加速分解并获得精确的特征值,所构建的指标能有效区分轴承生命周期中的不同运行阶段。At present,indicators for monitoring bearing health are mainly established by using statistical models and machine learning methods which are cumbersome and require manual intervention.Dynamic mode decomposition(DMD)has been effectively applied in vibration signal analysis,but it’s time-consuming and takes a lot of computing space at high sampling frequency.Therefore,a method for building rolling bearing health indicators based on compressed dynamic mode decomposition(CDMD)was proposed by combining compressed sensing algorithm with DMD.Firstly,a low-dimensional random measurement matrix was used to generate a compressed Hankel matrix in order to improve the singular value decomposition speed of bearing vibration signals.Then,on the basis of eigenvalues obtained by DMD,the kurtosis of the real part of the eigenvalues and the root mean square(RMS)value of the mode amplitude were calculated to form simpler and more effective bearing health indicators.Simulation and experimental signals were used to verify the proposed method.The results show that CDMD can replace DMD to accelerate signal decomposition and obtain accurate eigenvalues,and the constructed indicators can effectively distinguish different running stages in the life cycle of bearings.

关 键 词：滚动轴承 健康指标 压缩动态模式分解 故障监测 特征值 峭度 模式幅值 

分 类 号：TH133.33[机械工程—机械制造及自动化]