A Low Resource Fault Diagnosis Model Based on CDAN Domain Adaptation

Facing the problem of insufficient labeled data, which is common in the field of fault diagnosis, this study proposes an approach based on the CDAN domain adaptation model. The performance of the fault diagnosis model is improved by fusing data from different fault types. Experiments conducted on the Case Western Reserve University sliding bearing dataset validate the effectiveness of this method compared to traditional model training and direct data augmentation strategies, which achieves higher data utilization and fault diagnosis accuracy. Feature space visualization analysis further confirms the effectiveness of feature alignment between source and target domains.