We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Resnet-based deep learning multilayer fault detection model-based fault diagnosis.
- Authors
Jaber, Mustafa Musa; Ali, Mohammed Hasan; Abd, Sura Khalil; Jassim, Mustafa Mohammed; Alkhayyat, Ahmed; Majid, Mohammed Sh.; Alkhuwaylidee, Ahmed Rashid; Alyousif, Shahad
- Abstract
Fault detection has taken on critical relevance in today's automated manufacturing processes. Defect tolerance, dependability, and safety are some of the fundamental design attributes of complex engineering systems provided by this method. Fault Diagnosis is made more difficult by a lack of performance; data-driven design and the capacity to transfer learning are also essential considerations. This paper proposes the ResNet-based deep learning multilayer fault detection model (ResNet-DLMFDM) to enrich high performance, design, and transmission-learning skills. Wavelet pyramid packet decomposition and each sub drive coefficient utilize the input of each deep research network channel for multi-kernel domain analysis. Pseudo-label networks have been developed conceptually to investigate different interval lengths of sequential functionality and to gather local database flow sequence functions to improve existing error detection processes. Experiment findings reveal that the proposed approach outperforms current algorithms regarding data correctness, storage space utilization, computational complexity, noiselessness, and transfer performance. The results are obtained by analyzing the multi-kernel and showing the domain ratio of 87.6%, increased storage space ratio of 88.6%, wavelet decomposition performance ratio of 84.5%, and the high accuracy of the data transmission ratio of 83.5%, and the noiseless diagnosis ratio of 93.8%.
- Subjects
FAULT diagnosis; DEEP learning; ENGINEERING systems; MANUFACTURING processes; DATABASES; COMPUTATIONAL complexity
- Publication
Multimedia Tools & Applications, 2024, Vol 83, Issue 7, p19277
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-16233-9