18 July 2022 in Journal of Materials Science
19 September 2023 in IEEE/ASME Transactions on Mechatronics
The success of fault diagnosis based on deep learning benefits from a large amount of labeled fault samples. However, the scarcity of labeled fault samples in fault diagnosis of wind turbine gearbox (WTG) makes it difficult to train a satisfactory diagnostic model. To address this issue, this article proposed a semisupervised weighted centroid prototype network (SSWCPN) for WTG fault diagnosis. Specifically, SSWCPN is a few-shot semisupervised learning framework, which alleviates the matter of overfitting caused by the lack of supervision information. First, to capture abundant semisupervised information to guide network training, a sample selection model based on the evolution trend of posterior probability is proposed, which could efficiently cherry-pick out the unlabeled samples of high confidence to refine prototypes. Second, a new prototype updating strategy based on a weighted centroid prototype is designed, which controls the prototype drifting issue caused by incorrect pseudolabels and the introduction of new data distribution. Finally, experiments performed on test-bench data and successful application on WTG data show that the proposed SSWCPN-based WTG fault diagnosis achieves the best fault diagnosis performance among the comparison methods.
26 August 2022 in Journal of Cleaner Production
18 April 2021 in Journal of Mechanical Science and Technology
28 July 2023 in Frontiers in Materials
The stability of the forming layer shape is a critical factor that impacts the final quality of sample morphology in wire and arc additive manufacturing (WAAM). This paper investigates the forming process and control methods to identify ways to optimize the process and improve the quality of the final product. The study aims to enhance the quality and precision of samples produced by the WAAM process by establishing a morphology control method based on image feedback. The focus is on real-time image acquisition using a CCD and simultaneous extraction of forming process parameter characteristics. Using a central composite experimental design, a prediction model is developed to estimate key process parameters and feature sizes of deposits, including deposition height and width. To further analyze the dynamic characteristics of feature sizes of deposited layers, step response identification is conducted using three process parameters as the system input: forming speed, arc current, and wire feeding speed. The experiments are designed to determine the weight of each process parameter and achieve a high level of response speed and precision. The findings indicate that the feature size of the deposited layer is most sensitive to changes in forming speed, followed by wire feeding speed, while arc current has the least impact on feature size. After verification, the monitoring of the deposition height and width was found to be in good agreement with the prediction model, with an accuracy of over 90%. The results of this study can be used for size measurement and optimization of large aviation aluminum alloy components using WAAM technology and to improve the quality of products produced using these processes.
1 November 2021 in Transactions of the Indian Institute of Metals