Open AccessBearing degradation trend prediction under different operational conditions based on CNN-LSTM
Author(s) -
Guozeng Liu,
Jingzhe Zhao,
Xin Zhang
Publication year - 2019
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/612/3/032042
Subject(s) - bearing (navigation) , convolution (computer science) , degradation (telecommunications) , computer science , dimension (graph theory) , convolutional neural network , artificial intelligence , deep learning , feature extraction , long short term memory , reduction (mathematics) , artificial neural network , feature (linguistics) , vibration , pattern recognition (psychology) , machine learning , recurrent neural network , mathematics , physics , quantum mechanics , telecommunications , linguistics , philosophy , geometry , pure mathematics
Bearing degradation research is an important part in condition-based maintenance. This paper tries to propose an end-to-end model to realize bearing degradation trend prediction by vibration signals. Convolution neural network is good at data dimension reduction and feature extraction, and long short-term memory is good at deal with time sequences. The experiment proves that the combination of the two deep learning methods has a good effect, which avoid some disadvantages of traditional methods. Results shows that the model has application value in industrial practice.