ISA Transactions, Год журнала: 2025, Номер unknown
Опубликована: Май 1, 2025
Язык: Английский
ISA Transactions, Год журнала: 2025, Номер unknown
Опубликована: Май 1, 2025
Язык: Английский
Reliability Engineering & System Safety, Год журнала: 2025, Номер unknown, С. 110847 - 110847
Опубликована: Янв. 1, 2025
Язык: Английский
Процитировано
9Measurement Science and Technology, Год журнала: 2025, Номер 36(3), С. 036123 - 036123
Опубликована: Фев. 18, 2025
Abstract To enhance the cross-domain diagnostic ability of model, domain adaptation method is adopted. When using traditional adaption methods to extract invariant characteristics axial flow fan faults, source and target domains will be close each other, thereby distribution trained changed. fault gather at classification boundary, model incorrectly classify some samples. In addition, single can lead poor generalization ability. resolve above issues, a multi-source intelligent diagnosis based on asymmetric adversarial training proposed. this method, used realize unidirectional movement from domain; triplet-center loss expand inter-class distance shorten intra-class in are extracted different domains, they inputted their respective classifiers, then aligning outputs classifier cosine similarity. improve strategy weights The industrial actual data verification results indicate that effective solving relevant practical problems.
Язык: Английский
Процитировано
0Proceedings of the Institution of Mechanical Engineers Part E Journal of Process Mechanical Engineering, Год журнала: 2025, Номер unknown
Опубликована: Апрель 17, 2025
Bearing fault diagnosis is crucial for mechanical system reliability. Numerous techniques have been developed to identify faults in bearings. However, the signals under time-varying speed condition are nonstationary, and most methods suffer from nonstationary property caused by problem. The change of changes pulse frequency, but structure remains same. Shift-invariant dictionary learning (SIDL) can learn repetitive signal without limitation structure's size. Thus, pulses condition. Union circulants (UCDL) a kind SIDL, where algorithm takes advantage explicit circulant structures has powerful ability into atoms. In this work, we use UCDL extract features condition, hidden Markov model (HMM) used diagnose faults. We further found that with global sparse coding named basis pursuit, which maintain stability coefficient solution, better performance diagnosis. To validate proposed method improved both simulation experimental processed, results prove high efficiency SIDL achieved an average diagnostic accuracy 100% simulations 98.32% experiments, superior traditional methods.
Язык: Английский
Процитировано
0ISA Transactions, Год журнала: 2025, Номер unknown
Опубликована: Май 1, 2025
Язык: Английский
Процитировано
0