A Time-Series Segmentation and Contrastive Learning Method for Fault Diagnosis of Rotating Machinery

The reliability and stability of rotating machinery are critical to industrial productivity and safety. In this study, a novel multi-fault diagnosis method for rotating machinery is proposed, combining time series segmentation and contrast learning techniques. The method effectively improves the accuracy of fault classification by segmenting raw sensor signals and extracting robust feature representations using contrast learning. We evaluate the performance of the method on the publicly available dataset and show that it outperforms existing methods in terms of both fault classification accuracy and generalization ability. This research provides an efficient and scalable solution for predictive maintenance strategies in industrial environments.