基于互信息和SVR的滚动轴承剩余寿命预测  

Prediction of the Remaining Useful Life of Rolling Bearings Based on Mutual Information and SVR

作　　者：孟建军[1] 胡文涛 MENG Jianjun;HU Wentao(Lanzhou Jiaotong University,Mechatronics T&R Institute,Lanzhou 730070,China)

机构地区：[1]兰州交通大学机电技术研究所,兰州730070

出　　处：《机械设计与研究》2020年第6期92-95,共4页Machine Design And Research

基　　金：甘肃省高等学校科研项目:(2017D-09,2018C-10)。

摘　　要：针对数据驱动下如何准确预测滚动轴承剩余有效寿命的问题,提出了一种基于互信息和支持向量回归(SVR)的滚动轴承剩余寿命预测的新方法。该方法使用互信息对轴承的时域和频域里的多个退化特征进行约简,只保留符合要求的特征;通过实验平台采集测试轴承的退化数据,利用机器学习建立轴承的SVR剩余寿命预测模型,以约简后的退化特征作为预测模型的输入,以剩余寿命作为输出,对轴承的剩余寿命进行预测;同时,使用相关系数法对滚动轴承进行剩余寿命预测,并对两种方法的预测结果进行对比分析,结果表明,利用互信息对特征进行约简,使用SVR对轴承进行剩余寿命预测具有良好的效果。Aiming at the problem that how to precisely calculate the residual life of rolling bearings under datadriven,a new method based on mutual information and support vector regression(SVR)is used to predict the remaining useful life of rolling bearings.This method usesthe mutual information to reduce multiple degradation characteristics in the time-domain and frequency-domain of the bearings,and only keeps the characteristicsthat meets the requirements;collectsthe degradation data of the testing bearings through an experimental platform,and then builds a SVR prediction model for the remaining useful life of the bearings by machine learning,thus predicting the remaining useful life of the bearings,withthe reduced degradation characteristics as input of the prediction model and the remaining useful life as output;And also,the correlation coefficient method is used to predict the remaining useful life of rolling bearings,and a contrastive analysis of the prediction results of the two methods is made.The results show that using mutual information to reduce the characteristics and SVR to predict the remaining useful life of bearings have good effects.

关 键 词：滚动轴承 支持向量回归 特征约简 剩余寿命 互信息 

分 类 号：U279.323.1[机械工程—车辆工程]