HOU Zhaoguo, WANG Huawei, XIONG Minglan, WANG Junzhou
JOURNAL OF VIBRATION AND SHOCK. 2023, 42(9): 236-246.
A gearbox fault diagnosis method based on weighted fusion of multi-channel data and deep transfer model was proposed to solve the problems of large fluctuation of fault identification accuracy of single sensor, low data utilization, low reliability and insufficient generalization ability of fault diagnosis model under multiple working conditions. Firstly, in order to fully mine the information of multi-channel data of gearbox, a multi-channel fusion method based on information entropy weighting is proposed. The fusion weight of data of each channel is calculated by using information entropy method, and the sampled data of each channel is weighted fused. Secondly, the deep transfer model is pre-trained by using the fusion data of source domain, and the model parameters obtained by the pre-training are used as the initialization parameters of the target domain model. Meanwhile, the parameters of the feature extractor of target domain model are frozen, and the parameters of the classifier of target domain model are fine-tuned by using the fusion data of target domain. In order to adapt to the new target sample recognition task, the deep transfer model is transferred from source domain to target domain. Finally, the multi-condition transfer diagnosis test results of gearbox show that the proposed method can be effectively used for gearbox fault diagnosis. Compared with the traditional transfer learning methods BDA, TCA, JDA, JGSA, GFK and the deep transfer learning methods AdaBN, MK-MMD, DCTLN, It has higher average transfer diagnosis accuracy and better generalization performance under variable working conditions.