断层(地质)
样品(材料)
人工智能
计算机科学
模式识别(心理学)
地质学
地震学
化学
色谱法
作者
Huaitao Shi,Xinyi Fu,Yating Hu,Xiaotian Bai
出处
期刊:CRC Press eBooks
[CRC Press]
日期:2025-02-12
卷期号:: 259-267
标识
DOI:10.1201/9781003470083-24
摘要
In small sample cases, maintaining high fidelity to rolling bearing damage can be challenging because bearing damage characteristics are not obvious. Aiming at these problems, this paper presents a rolling bearing fault diagnosis method based on a deep learning-assisted digital twin with few samples. In this method, the Trans-CycleGAN model is proposed to augment and fit the virtual and real data of the digital twin in a high-fidelity manner. In addition, the L-LEG Transformer model is introduced, which can improve the feature recognition accuracy and classification ability of time-frequency diagrams of different sizes. The proposed method is validated in two sets of rolling bearing test bed tasks, respectively. Results from experiments show that virtual data can be fitted with high fidelity to actual data. Fault classification is more accurate. The feasibility and effectiveness of the proposed deep learning-assisted digital twin method are demonstrated.
科研通智能强力驱动
Strongly Powered by AbleSci AI