The article discusses the importance of fault diagnosis in rotating machinery for the safety and stability of modern industrial systems. It highlights the challenge of distribution discrepancy between training data and real-world operation scenarios, which can reduce the performance of existing systems. To address this issue, the authors propose a transfer learning-based method using acoustic and vibration signals. This method involves a feature fusion technique to provide more reliable information about faults. The system uses a DNN-based classifier for more effective diagnosis, and the model is pre-trained and fine-tuned for better performance. The method has shown improved results compared to existing techniques.
Publication date: 25 Oct 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2310.14796
