Most existing transfer fault diagnosis methods have gained noteworthy domain adaptability and generalization capability for rotating machinery under the scenario where source and target have identical label space. Nevertheless, inconsistent fault modes that occur under different operating conditions in an actual industrial environment can make it challenging to construct a transfer fault diagnosis model across domains since domain shift and category inconsistencies. To address this issue, a dual weighted-class adversarial network (DWCAN) is proposed for rotary machine fault diagnosis using multisource domain with class-inconsistent data. First, the model consists of a common feature extractor, weighted intraclass and interclass adversarial discriminator, and classifier. Then, based on the intraclass and interclass adversarial strategy, the DWCAN method learns domain invariant fault representations across domains by narrowing domain and class inconsistencies and achieves intraclass compactness and interclass separability from the aspect of local class alignment. Lastly, two experimental datasets are adopted to demonstrate the promising performance of DWCAN compared with other methods.
19 September 2025 in Measurement Science and Technology
Remaining useful life (RUL) prediction, as a pivotal technology in prognostics and health management, plays a critical role in proactive fault prediction and maintenance planning. Although data-driven methods have achieved remarkable progress in RUL prediction, their practical industrial application is hindered by high model complexity and poor interpretability. While existing physics-informed neural network (PINN) approaches can address these limitations, their neglect of computational quantity and uncertainty quantification in prediction results undermines their effectiveness for predictive maintenance. In this paper, we propose a variational autoencoder (VAE) based lightweight PINN for RUL prediction. The self-attention mechanism is introduced into a VAE to effectively extract degradation features and approximate the nonlinear dynamical equations of degradation systems as prior physical knowledge. Subsequently, the self-attention assisted VAE-PINN captures the underlying physical mechanisms during degradation and predict RUL. Validated on a benchmark dataset, the proposed method demonstrates superior prediction performance and improved interpretability with significantly fewer parameters compared to state-of-the-art approaches.
19 April 2015 in The International Journal of Advanced Manufacturing Technology