Reference #: 01571
The University of South Carolina is offering licensing opportunities for DRCNN for Multi-task Bearing Fault Diagnosis with Information Fusion
Background:
First, most industrial systems are working in variable operating conditions and environments. The information of operating conditions, such as load profile, rotating speed, and environmental factors, etc., and domain knowledge, such as fault mode, fault mechanisms, and fault characteristic frequencies, etc., have significant influence on the accuracy and performance of diagnosis. However, this important information is not utilized in most of the existing DL based approaches. Second, DL based approaches provide automatic and effective feature learning solutions. For bearing monitoring data, some features are not informative or irrelevant to faults and they will result in low training efficiency or large diagnosis errors. Most of the existing works lack explicit discriminate feature learning mechanisms so that they give equal attention to all features.
Invention Description:
This invention developed an enhanced discriminate feature learning-based DR-CNN for multi-task bearing diagnosis with information fusion. The proposed approach integrates domain knowledge, operating conditions, and data from multiple sensors in a dynamic training process to enhance the performance of multi-task diagnosis
Potential Applications:
The verification results show that the method can achieve high training accuracy, fast convergence speed, and high diagnosis accuracy in bearing fault diagnosis. The method can also be extended to other mechanical systems fault diagnosis tasks, such as wind turbines, helicopters, and shipboard, etc.
Advantages and Benefits:
Different from other works, the proposed approach integrates domain knowledge, operating conditions, and data from multiple sensors in a dynamic training process to enhance the performance of multi-task diagnosis. The method can achieve high training accuracy, fast convergence speed, and high diagnostic accuracy for multitask bearing fault diagnosis with a single network.