Real-Time Fault Diagnosis Method for Multi-Source Information Fusion
Author Information
Author(s): Chen Danmin, Zhang Zhiqiang, Zhou Funa, Wang Chaoge
Primary Institution: Henan Finance University
Hypothesis
Can a two-level transfer learning method improve real-time fault diagnosis using multi-source heterogeneous information?
Conclusion
The proposed method effectively integrates features from one-dimensional signals and screenshots, achieving real-time fault diagnosis with low time complexity.
Supporting Evidence
- The method avoids convolution operations, which reduces time complexity.
- Experiments showed that TTDNN outperformed traditional CNN methods in fault diagnosis accuracy.
- TTDNN achieved similar online diagnosis times to DNN while providing higher accuracy.
Takeaway
This study created a new way to quickly find problems in machines by using different types of data together, making it faster and easier to diagnose faults.
Methodology
The study used a two-level transfer learning approach to fuse multi-source heterogeneous information for fault diagnosis.
Limitations
The study did not filter irrelevant source data before migration, which could negatively impact the target domain.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website