|
|
Publications
Gao, D., Zhu, Y.S., Yan, K., Fu, H., Ren, Z.J., Kang, W. and Guedes Soares, C. (2023), Joint learning system based on semi-pseudo-label reliability assessment for weak fault diagnosis with few labels, Mechanical Systems and Signal Processing, Vol. 189, 110089
The deep neural network has an excellent performance in fault feature extraction, given that a large amount of labelled data are available for network training. However, data labelling is a difficult task in practical engineering, which creates difficulties for fault diagnosis, especially when the faults are weak. To tackle this problem, a semi-pseudo-labelling diagnosis system is proposed in this paper by considering the confidence and reliability of samples to suit the situation where the labels are insufficient and the faults are weak. By adding pseudo-labels, the unlabelled data whose fault information is swamped by a large amount of noise can achieve low-density separation and entropy regularization in the sample space, thus supporting the training of deep learning models for weak fault diagnosis. Regarding the problems of traditional pseudo-labelling in weak-fault-related feature extraction, a series of solutions are proposed to instantiate it in the field of fault diagnosis. The designed model reduces the label noise of the pseudo-label and enhances the weak-fault-related feature extraction capability. The effectiveness of this method is validated on the datasets collected by simulating fault bearings and real failure bearings.
If you did not manage to obtain a copy of this paper: Request a copy of this article
For information about all CENTEC publications you can download: Download the Complete List of CENTEC Publications
|