arxiveess.SP2026-07-15
Fine-Grained Open-Set Fault Diagnosis via Metric-Guided Time-Frequency Configuration Selection and Class-Specific Autoencoders
Reliable fault diagnosis of rotating machinery is essential for the safe and stable operation of industrial systems. Although deep learning methods perform well under closed-set conditions, real machinery may encounter previously unseen fault states. Existing open-set fault diagn…