Maintaining machinery can be time-consuming and error-prone, but advancements in deep transfer learning are enabling automated fault diagnosis. Researchers have developed a Label Recovery and Trajectory Designable Network (LRTDN) to improve fault diagnosis by correcting incorrectly labeled data and aligning data distributions between source and target domains. This approach enhances the generalization of diagnosis models to different machine conditions. The LRTDN consists of a residual network with dual classifiers, an annotation check module, and adaptation trajectories to address challenges in deep transfer learning. By distinguishing unique features between source and target domains, the model can accurately diagnose faults in different machines. Testing the LRTDN on bearing fault diagnosis showed significantly higher accuracy rates compared to other methods, with an average accuracy of 79.9% for one test and 95% for another. This innovative approach reduces the need for extensive data collection and training, making fault diagnosis more efficient and accurate in various machine conditions.
Source link
Source link: https://www.bisinfotech.com/new-study-improves-fault-diagnosis-in-machines-with-deep-learning/
New study enhances machine fault diagnosis with deep learning. #Technology
