Contrastive Learning and Convolution Transformer Network-based Reducer Fault Diagnosis Method With Few Labeled Samples

Objectives To address the issue of low fault diagnosis accuracy of traditional neural networks with few labeled samples, a method based contrastive learning and convolution transformer network is proposed.

Methods First, raw monitoring data are transformed into similar sample pairs by data augmentation. At the same time, the similar sample pairs are mapped to the deep feature space by a feature extractor. Then, the transformer network is utilized to design cross-prediction tasks for local comparison and global comparison to achieve clustering of data with the same fault type by comparing the intrinsic similarity between the same batches of data. Finally, the downstream classification network is trained with few labeled samples to improve the diagnostic performance of the proposed model.

Results The effectiveness of the proposed method is verified by a self-built reducer test rig. The results show that accuracy of the proposed method reaches 98.38% with few labeled samples, which has obvious advantages compared with the existing methods.

Conclusions The research results can provide the key technology for the fault diagnosis of industrial equipment with few labeled samples, helping the development of intelligent manufacturing.