多尺度熵方法在机械故障诊断中的应用研究进展  

Research Progress on the Application of Multi-scale Entropy Method in the Mechanical Fault Diagnosis

作　　者：郑近德[1] 姚殷柔 潘海洋[1] 童靳于[1] 刘庆运[1] ZHENG Jinde;YAO Yinrou;PAN Haiyang;TONG Jinyu;LIU Qingyun(School of Mechanical Engineering,Anhui University of Technology,Maanshan 243032)

机构地区：[1]安徽工业大学机械工程学院,安徽马鞍山243032

出　　处：《安徽工业大学学报（自然科学版）》2024年第1期46-57,97,共13页Journal of Anhui University of Technology（Natural Science）

基　　金：国家自然科学基金项目(51975004);安徽省教育厅杰出青年基金项目(2022AH020032)。

摘　　要：机械设备状态监测与故障诊断的关键是故障特征的表征与提取,采用基于熵及相关方法建立的非线性动力学指标能够提取蕴藏在振动信号中的非线性故障特征信息。自熵方法引入以来,通过不断修改和改进来提高熵估计的准确性,多尺度熵进一步拓展了时间序列其他尺度上包含的复杂度信息,其在设备状态监测与故障诊断中得到广泛应用。本文对单一尺度熵及多尺度样本熵、多尺度模糊熵、多尺度排列熵和多尺度散布熵等多尺度熵方法在机械智能故障诊断中的应用进行综述,总结不同方法的特点优势与不足;针对多变量数据处理问题,综述由单变量推广到多变量的多元多尺度熵的应用发展过程。最后结合多尺度熵相关方法在机械智能故障诊断中面临的问题与挑战,对未来发展方向进行展望,即在工业大数据应用、故障机理、可解释性角度构建基于熵的深度学习模型。The key to condition monitoring and fault diagnosis of mechanical equipment is the characterization and extraction of fault features.The multi-scale entropy and its related methods-based nonlinear dynamic indexes can be used to effectively extract the nonlinear fault characteristic information contained in vibration signals.Since the introduction of the entropy method,the accuracy of entropy estimation has been improved and the application areas of entropy have been expanded through modifications and improvements.Multiscale entropy further explores the complexity information contained in the other scales of the time series,which has been widely used in mechanical condition monitoring and fault diagnosis.In this paper,firstly the application of multi-scale entropy methods in mechanical intelligent fault diagnosis including single-scale entropy,multi-scale entropy,multi-scale fuzzy entropy,multi-scale permutation entropy and multi-scale scatter entropy was reviewed,meanwhile the advantages and disadvantages of the related methods were summaried.Aiming at the problem of multivariate data processing,the multivariate multiscale entropy extended from univariate to multivariate was reviewed.Finally,combining the problems and challenges faced by relevant methods for focused multiscale entropy in mechanical intelligent fault diagnosis,the three future development directions were outlooked,i.e.,the application in industrial big data,the combination with the study of fault mechanism,and constructing entropy-based deep learning models from the interpretability perspective.

关 键 词：多尺度熵 多元多尺度熵 智能故障诊断 滚动轴承 机械设备 

分 类 号：TN911.7[电子电信—通信与信息系统] TH165.3[电子电信—信息与通信工程]