基于SAE和BiGRU的滚动轴承剩余寿命预测
Remaining Life Prediction of Rolling Bearing Based on SAE and BiGRU
魏熙朋 1林建辉 2易彩2
作者信息
- 1. 西南交通大学机械工程学院 成都 610031
- 2. 西南交通大学牵引动力国家重点实验室 成都 610031
- 折叠
摘要
为提高对滚动轴承剩余使用寿命的预测准确性,提出了一种基于稀疏自编码器(SAE)和双向门控循环单元(BiGRU)的剩余寿命预测方法.首先利用四个评价指标对从滚动轴承振动信号中提取出来的时域、频域以及时频域特征进行筛选,构建敏感退化特征集.然后为解决各个特征之间存在的信息冗余问题,利用SAE网络对敏感退化特征进行融合降维.最后将融合敏感退化特征输入BiGRU模型中完成对滚动轴承剩余寿命的预测.采用公开的滚动轴承全寿命数据集进行验证,结果表明,与长短期记忆网络(LSTM)以及门控循环单元(GRU)相比,该方法具有更高的剩余寿命预测准确性.
Abstract
In order to improve the prediction accuracy of rolling bearing remaining life,a remaining life prediction method based on sparse auto encoder(SAE)and bi-directional gated recurrent unit(BiGRU)is proposed.Firstly,the time domain,fre-quency domain and time-frequency domain features extracted from rolling bearing vibration signals are screened by using four evalu-ation indexes,and the sensitive degradation feature set is constructed.Then,in order to solve the problem of information redundan-cy between features,SAE network is used to reduce the dimension of sensitive degraded feature set.Finally,the fusion sensitive degradation features are input into BiGRU model to complete the prediction of rolling bearing remaining life.The results show that,compared with LSTM and GRU,this method has higher accuracy of remaining life prediction.
关键词
滚动轴承/剩余寿命预测/稀疏自编码器/双向门控循环单元Key words
rolling bearing/remaining life prediction/sparse auto encoder/bi-directional gated recurrent unit引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据