Fault Diagnosis Method for Rolling Bearings Under Noisy Environments Based on EEMD and CNN-BiLSTM
A fault diagnosis method based on ensemble empirical mode decomposition(EEMD)and convolutional neural network bi-directional long short-term memory network(CNN-BiLSTM)is proposed to address the problems of inadequate feature extraction,overfitting and insufficient generalization ability of traditional deep neural networks when rolling bearings fail under noisy environments.During signal preprocessing stage,EEMD is used to decompose the vibration signals under noisy environments into a series of intrinsic mode functions,reducing the impact of noise.In the first layer of CNN part,large convolutional kernel and multi-branch structure are used to obtain different receptive fields,and some data are randomly discarded in each branch to enhance the anti-interference ability of the model,so as to extract diversified feature information with more generalization ability.In subsequent parts,residual structures are used to avoid the phenomenon of gradient disappearance when the network is deep,solving the problem of deep network degradation.In BiLSTM part,two parallel branch structures are used to enhance the utilization of temporal information,thereby improving the accuracy of the model under different operating conditions and noise environments.Using bearing datasets from Case Western Reserve University and Xi'an Jiaotong University,the proposed method is verified and compared with other deep learning methods and traditional machine learning methods.The results show that the proposed method achieves excellent fault diagnosis performance under various operating conditions and noise environments.
rolling bearingfault diagnosisensemble empirical mode decompositionconvolutional neural networkbi-directional long short-term memory neural network
NETL
NSTL
万方数据
