We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
LSTM Neural Network Fault Diagnosis Method for Rolling Bearings Based on Information Fusion.
- Authors
Cheng Zhong; Yu Liu; Jie-Sheng Wang; Zhong-Feng Li
- Abstract
The early fault features of rolling bearings are weak and the noise in the working environment is large. If a single feature extraction method is used, it is easy to cause feature loss. A rolling bearing fault diagnosis method based on multiple feature extraction and information fusion was proposed, which effectively avoids the problem of fault feature loss caused by single feature extraction method being vulnerable to external environment and signal characteristics. Firstly, wavelet accumulation energy parameters are calculated by wavelet decomposition, and empirical mode decomposition (EMD) and ensemble empirical mode decomposition (EEMD) are adopted. EEMD algorithm decomposed the signal and then reduced the dimension to get the characteristic parameters of signals, calculated the time domain performance index to get the time domain characteristic parameters of signals, and then used low variance and Pearson coefficient filtering to reduce the dimension of feature vector so as to facilitate the setting of the neural network classifier. Finally, LSTM neural network is used as rolling bearing fault diagnosis classifier to verify the validity of the proposed method.
- Subjects
ROLLER bearings; FAULT diagnosis; HILBERT-Huang transform; NOISE (Work environment); DIAGNOSIS methods
- Publication
IAENG International Journal of Computer Science, 2022, Vol 49, Issue 4, p1088
- ISSN
1819-656X
- Publication type
Article