一种基于数据驱动和负时间序列PCA的滚动轴承健康指标构建方法  

Constructing Health Indicators of Rolling Bearings Based on Data-driven Method and Negative Time Series of PCA Theory

作　　者：王栩沂 丁泽良[1] 米承继[1] 王思睿 Wang Xuyi;Ding Zeliang;Mi Chengji;Wang Sirui(College of Mechanical Engineering,Hunan University of Technology,Zhuzhou,Hunan 412007,China)

机构地区：[1]湖南工业大学机械工程学院,湖南株洲412007

出　　处：《机电工程技术》2025年第5期103-107,196,共6页Mechanical & Electrical Engineering Technology

基　　金：湖南省自然科学基金项目(2023JJ50186);湖南省自然科学基金项目(2022JJ50056)。

摘　　要：滚动轴承在制造和装配等过程中存在不确定性因素,导致滚动轴承健康指标初始退化程度不一致,对早期故障不敏感。针对此问题提出了一种基于数据驱动方法和负时间序列主成分分析法(PCA)的表征滚动轴承退化性能的健康指标构建方法。首先,基于数据驱动方法从轴承的原始振动数据中提取16个表征轴承各退化阶段所有故障信息的时域特征;然后,使用均值化方法对特征数据进行降噪,基于标准化方法统一量纲;最后,利用负时间序列的主成分分析法构建轴承的健康指标。通过在实验数据中的应用,并与有效值(RMS)指标进行对比,结果表明负时序指标均先于RMS指标发现早期故障,验证了此方法的有效性。Uncertainties in the manufacturing and assembly processes of rolling bearings result in inconsistent initial degradation levels of their health indicators and insensitivity to early failures.In response to this issue,a method is proposed to construct health indicators characterizing the degradation performance of rolling bearings based on a data-driven method and negative time series of principal component analysis(PCA)theory.First,16 time-domain features,which represent all fault information at various stages of bearing degradation,are extracted from the original vibration data using a data-driven method.Then,feature data are denoised using a mean normalization method,and dimensions are unified using a standardization method.The final step is to construct the health indicators of the bearings.This step is mainly achieved by using PCA based on negative time series.Through application to experimental data and comparison with root mean square(RMS)indicators,the results show that the negative time series indicators all find the early faults before the RMS indicators,the effectiveness of this method can be verified.

关 键 词：滚动轴承 健康指标 数据驱动 主成分分析(PCA) 

分 类 号：TH133.33[机械工程—机械制造及自动化] TP391[自动化与计算机技术—计算机应用技术]