Prediction of Bearing Degradation Trend based on LSTM
- Resource Type
- Conference
- Authors
- Yuhang, CHEN; Bing, HAN
- Source
- 2019 IEEE Symposium Series on Computational Intelligence (SSCI) Computational Intelligence (SSCI), 2019 IEEE Symposium Series on. :1035-1040 Dec, 2019
- Subject
- Bioengineering
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Robotics and Control Systems
Signal Processing and Analysis
Degradation
Market research
Feature extraction
Vibrations
Logic gates
Time-domain analysis
Computer architecture
bearing degradation
trend prediction
LSTM network
- Language
The prediction of bearing degradation trend is important for remaining useful life (RUL) estimation. However, there are no clear indicators due to the presence of noise, and long step prediction is not effective enough. This paper proposed a predict method using LSTM in which the time domain and spectral kurtosis related features was selected by monotonicity. Then PCA was employed to fuse features. Health indicator was generated which was increasing as the bearing approaches to failure. A sequence-to-one regression LSTM network was employed to predict the trend of health indicator. Intelligent Maintenance Systems (IMS) bearing run to failure dataset was employed to verify the validity of the method. The result shows that proposed method can effectively predict the degradation trend. This method can be used to real-time estimate the RUL.