摩擦学系统润滑磨损故障诊断特征提取研究综述  

Review on Feature Extraction of Lubrication and Wear Fault Diagnosis in Tribology System

作　　者：石新发 贺石中 谢小鹏 孙宇航[1,3] SHI Xinfa;HE Shizhong;XIE Xiaopeng;SUN Yuhang(Guangzhou Mechanical Engineering Research Institute Co,Ltd,Guangdong Guangzhou 510530,China;School of Mechanical and Automotive Engineering,South China University of Technology,Guangdong Guangzhou 510640,China;National and Local Joint Engineering Center of Industrial Friction and Lubrication Technology,Guangdong Guangzhou 510530,China)

机构地区：[1]广州机械科学研究院有限公司,广东广州510530 [2]华南理工大学汽车与机械工程学院,广东广州510640 [3]工业摩擦润滑技术国家地方联合工程研究中心,广东广州510530

出　　处：《摩擦学学报》2023年第3期241-255,共15页Tribology

基　　金：国家重点研发计划资助项目(2018YFB2001604);广州机械科学研究院有限公司博士后专项(17300065)资助.

摘　　要：润滑磨损故障是机械装备安全、健康、可靠运行的严重威胁,在对其诊断中存在的数据源多,造成数据维度高、形式多样化、结构与关系复杂以及数据与故障之间的映射关系不明确等问题,严重影响了诊断的效率、准确性和针对性.随着装备智能化、集成化和大型化发展,润滑磨损故障诊断也将进入大数据和智能化时代,对诊断数据的应用与分析水平要求更高.特征提取能实现原始数据降维、数据关系建立和故障敏感性信息获取,是润滑磨损故障诊断的基础性工作,也是实现数据高效应用的前提.通过对润滑磨损故障诊断流程与技术分析,从诊断实验室检测、工业现场监测和在线实时监测等3个方面,研究装备润滑磨损故障诊断所需获取信息的组成,明确了其特征提取研究的内容与方向;在对磨损颗粒图像、磨损定量数据、润滑油性能劣化和润滑油污染等4个方面特征提取研究现状进行综述的基础上,提出了当前装备润滑磨损故障诊断特征提取所面临的挑战性问题;最后根据以上挑战性问题,结合装备发展趋势,指出了今后润滑磨损故障特征提取的研究方向.As the main sources of equipment failure,lubrication and wear faults are the serious threats to the safe,healthy,and reliable operation of industrial equipment.Lubrication and wear fault diagnosis,which have the history of more than sixty years,are the important aspect of tribology research and industrial application.As the reason of huge data sources involved in the diagnosis work,the data used for lubrication and wear fault diagnosis has the characteristic of high data dimension,diversified forms,complex structure and relationship,and unclear mapping relationship between data and fault,which seriously affects the efficiency,accuracy and pertinence of the diagnosis.On the other hand,with the development of intelligent,integrated,and large-scale equipment,lubrication wear fault diagnosis will enter the era of big data and intelligence,which will have a higher requirement for the application and analysis level of diagnostic data.As the basic work of lubrication wear fault diagnosis and the premise of data efficient application,feature extraction can reduce the dimension of original data,establish the data relationship,and obtain fault sensitivity information.Hence,a comprehensive overview on lubrication and wear fault diagnosis feature extraction is necessary.Through the analysis of the process and technology of lubrication and wear fault diagnosis,the composition of the information collected from the equipment lubrication and wear fault diagnosis was studied from three aspects of diagnostic laboratory testing,industrial field monitoring,and online real-time monitoring,and the research direction and content of feature extraction were clarified.Research achievements of lubrication and wear fault diagnosis feature extraction in 40 years were summarized from four aspects,which were the feature of wear particle image identification,wear quantitative data of wear element and particle,lubricating oil performance degradation,and lubricating oil pollution of external medium and particles,and the technology and algorithm o

关 键 词：机械装备 摩擦学系统 润滑磨损 故障诊断 特征提取 

分 类 号：TH117.3[机械工程—机械设计及理论]