基于多重分形去趋势波动分析算法的表面磨损形式特征研究  

Study of Surface Wear Modes Characteristics Based on Multifractal Detrended Fluctuation Analysis Algorithm

作　　者：张一兵[1] 陈聪 刘立鹏 胡瑞 ZHANG Yibing;CHEN Cong;LIU Lipeng;HU Rui(School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan Hubei 430070,China;CRRC Zhuzhou Electric Co.,Ltd.,Zhuzhou Hunan 412000,China;School of Mechanical and Electrical Engineering,Nanchang Institute of Technology,Nanchang Jiangxi 330000,China)

机构地区：[1]武汉理工大学机电工程学院,湖北武汉430070 [2]中车株洲电机有限公司,湖南株洲412000 [3]南昌工程学院机械与电气工程学院,江西南昌330000

出　　处：《润滑与密封》2022年第2期15-21,共7页Lubrication Engineering

基　　金：国家自然科学基金项目(51765044)。

摘　　要：为了识别表面的磨损形式以研究零件表面的摩擦学特性,以不同载荷下的磨料磨损和黏着磨损2种磨损形式的表面为研究对象,应用LSTM-1型磨损表面形貌测量仪和稳健高斯滤波方法对磨损表面形貌进行数据采集和滤波处理后,使用多重分形去趋势波动分析算法(MF-DFA)计算磨损表面高频信息的广义赫斯特指数,并通过分析该指数与表面形貌磨损纹理特征之间的关系,使用主成分分析法提取用于识别2种磨损形式的特征,然后采用K-means聚类、支持向量机(SVM)和BP神经网络方法,分别对所提取的特征参数进行分类,比较不同分类器识别结果的准确率。研究结果表明:广义赫斯特指数可用于区分磨损表面犁沟类和凹坑类纹理特征的指标,作为机器学习特征对表面磨损形式识别分类。In order to identify the wear modes of the surface and study the tribological characteristics of the part surface,the surface of the two wear modes of abrasive wear and adhesive wear under different loads was studied.The wear surface topography data collection and filtering process were carried out by applying the LSTM-1 wear surface topography measuring instrument and robust Gaussian filtering method,the multifractal detrended fluctuation analysis algorithm(MF-DFA)was used to calculate the wear surface generalized Hurst exponents of the high frequency information.By analyzing the relationship between the exponents and surface topography wear texture features,the features used to identify the two wear modes were extracted by using principal component analysis.K-means clustering and support vector machine(SVM)and BP neural network methods were used to classify the extracted feature parameters,and the accuracy of the recognition results of different classifiers were compared.The research results show that the generalized Hurst exponents can be used to distinguish the texture features of furrows and pits on worn surfaces,which can be used as a machine learning feature to identify and classify surface wear modes.

关 键 词：表面形貌 MF-DFA 磨损形式特征 广义赫斯特指数 机器学习 

分 类 号：TH117[机械工程—机械设计及理论]