基于FCM-LSTM的滚动轴承多阶段寿命预测  

Multi-stage life prediction of rolling bearings based on FCM-LSTM

作　　者：刘宇航 石宇强[1] 王俊佳[1] LIU Yuhang;SHI Yuqiang;WANG Junjia(School of Manufacturing Science and Engineering,Southwest University of Science and Technology,Mianyang 621010)

机构地区：[1]西南科技大学制造科学与工程学院,四川绵阳621010

出　　处：《机械设计》2023年第5期43-50,共8页Journal of Machine Design

摘　　要：针对滚动轴承逐渐呈现多阶段退化的退化特性,文中提出基于模糊C均值聚类(Fuzzy C-Means Clustering, FCM)和长短时记忆神经网络(Long Short-Term Memory, LSTM)的滚动轴承多阶段寿命预测方法。该方法的步骤是:使用小波包分解提取时频域特征,构建滚动轴承的健康指标;采用FCM将滚动轴承的退化过程分为多个阶段;使用LSTM对其在不同阶段的使用寿命进行预测,其预测结果可用于维修决策的制订与执行;利用开源试验数据集验证了该方法的合理性,表明了分阶段的寿命预测能有效提高预测精度。In view of gradual degradation of rolling bearings in multiple stages,in this article the method of multi-stage life prediction is proposed based on Fuzzy C-Means Clustering(FCM)and Long Short-Term Memory(LSTM).The steps are as follows.Firstly,wavelet packet decomposition is used to extract the time-frequency characteristics and work out the health indicators of rolling bearings.Secondly,the degradation process of rolling bearing is divided into several stages by means of FCM.Then,LSTM is used to predict the service life of rolling bearings at different stages,and the prediction results are used for formulation and implementation of maintenance decisions.Finally,the open-source experimental data set is used to verify that this method is rational,which shows that multi-stage life prediction enjoys a higher standard of accuracy.

关 键 词：滚动轴承 模糊C均值聚类(FCM) 多阶段退化 寿命预测 长短时记忆神经网络(LSTM) 

分 类 号：TH133.33[机械工程—机械制造及自动化]