ResSE-MSCNet Module based Generalized Zero-Sample Fault Diagnosis

A general challenge for intelligent fault diagnosis is the lack of fault data and its associated labels. Therefore, the problem of zero-sample fault diagnosis is of great importance. In this work, we proposed a ResSE-MSCNet module based generalized zero-sample fault diagnosis (GZSFD) method. The method integrates an optimized feature extraction module for deep feature extraction and a KL divergence-based classifier to effectively differentiate between seen and unseen faults. We introduced the ResSE and MSCNet modules to further refine the classification accuracy. The proposed method is evaluated using the open-source CWRU industrial bearing dataset and subjected to multiple tests, demonstrating well performance compared to existing approaches.