A Bearing Fault Diagnosis Method Based on L1 Regularization Transfer Learning and LSTM Deep Learning
Daiie Zhu, Xudong Song, Jie Yang, Yuyang Cong, Lijuan Wang
Abstract
In the practical application of rail transit, it is difficult to obtain bearing fault data and the training data of fault diagnosis model is insufficient, which leads to the low accuracy and generalization ability of fault diagnosis model. In this paper, a new bearing fault diagnosis method based on transfer learning is proposed. Based on transfer learning, we introduce L1 regularization, then adds it to the Long Short-Term Memory (LSTM) classification model, and uses a small amount of target domain data to fine tune the parameters of the model, and finally constructs a bearing fault diagnosis model. In this paper, the bearing data set of Case Western Reserve University is used to test the bearing fault diagnosis model. Compared with the conventional LSTM, Gated Recurrent Unit (GRU) and Bi-LSTM, the model proposed in this paper has higher accuracy in fault diagnosis as well as certain reliability and generalization ability.