Fault diagnosis of rolling bearing is essential to guarantee production efficiency and avoid catastrophic accidents. Domain adaptation is emerging as a critical technology for the intelligent fault diagnosis of rolling bearing. Most existing solutions learn domain-invariant features by statistical moment matching, adversarial training, or fusing two algorithms. However, these domain adaptation methodologies overemphasized learning domain-invariant features and ignored the generalization of classification performance on the target domain, which leads to inevitable misclassification. To address this issue, we propose a supervised contrastive learning-based domain adaptation network (SCLDAN) for cross-domain fault diagnosis of the rolling bearing in this paper. The SCLDAN develops a 1-D convolutional residual network to learn the raw signal features and employs the maximum mean discrepancy loss to achieve global domain alignment. In addition, a novel supervised contrastive learning approach is proposed, where a supervised contrastive loss and a mutual information loss are established to learn the class-specific information and improve the reliability of target prediction labels. Thus, the ambiguous data samples residing near the class boundaries of the target domain can be accurately identified, and the diagnosis accuracy is significantly improved. Extensive experiments on two experimental scenarios demonstrate the effectiveness of the proposed method.

Time-frequency analysis (TFA) technique is an effective approach to capture the changing dynamic in a nonstationary signal. However, the commonly adopted TFA techniques are inadequate in dealing with signals having a strong non-stationary characteristic or multi-component signals having close frequency components. To overcome this shortcoming, a new TFA technique applying a polynomial chirplet transform (PCT) in association with a synchroextracting transform (SET) is proposed in this study. It is shown that the energy concentration of the timefrequency representation (TFR) of a strong frequency modulated signal from a PCT transform can be further enhanced by a SET transform. The technique can also be employed to accurately extract the signal components of a multi-component nonstationary signal with close frequency components by adopting an iterative process. It is found that the TFR calculated from the proposed technique matches well with the ideal TFR which demonstrates the superiority of the current technique in dealing with non-stationary signals having rapidly changing dynamics. Results from the analysis of the experimental data under varying speed conditions confirm the validity of the proposed technique in dealing with non-stationary signals from practical sources.

Remaining useful life (RUL) prediction of rolling bearings is of paramount importance to various industrial applications. Recently, intelligent data-driven RUL prediction methods have achieved fruitful results. However, the existing methods heavily rely on the quality and quantity of the available data. For some critical bearings in industrial scenarios, the real run-to-failure data are insufficient, which impair the applicability of data-based methods for industrial practices. To address these issues, this article proposes a novel dynamic model-assisted RUL prediction approach for rolling bearing, in which sufficient simulation data are applied as the training data to solve the problem caused by insufficient real data. More specifically, a dynamic rolling bearing model is introduced for simulating the degradation process of physical structures. Then, a multilayer cross-domain transformer network is developed to implement RUL prediction and adapt the learned prediction knowledge from simulation to the actual measurements. Furthermore, a mutual information loss is utilized to preserve the generalized prediction knowledge of the measured data. The proposed approach can achieve a high RUL prediction accuracy with only limited measured data, which tackles the drawbacks of the existing data-driven methods. The experimental results of the rolling bearing degradation datasets demonstrate the effectiveness and superiority of the proposed RUL prediction approach.

Intelligent fault diagnosis methods based on deep learning have attracted significant attention in recent years. However, it still faces many challenges, including complex and variable working conditions, noise interference, and insufficient valid data samples. Therefore, a novel deep transfer learning bearing fault diagnosis model is designed in this work by fusing time-frequency analysis, residual network (ResNet) and self-attention mechanism (SAM). A multiscale time-frequency feature map (MTFFM) and global statistical feature matrix (GSFM) of vibration signals are first constructed using wavelet packet transform (WPT). A deep feature extraction network combining ResNet and SAM networks is then designed to realize the fused extraction of local and global time-frequency features. Finally, we construct a joint loss function by the combination of multi-kernel maximum mean discrepancy (MK-MMD) and the domain adversarial neural network (DANN) to optimize the depth feature extraction network, which improves the cross-domain invariance and fault state discrimination of depth features. The proposed optimization method fully exploits the advantages of high-dimensional spatial distribution difference evaluation and gradient inversion adversarial strategy. Its effectiveness is demonstrated through variable working condition transfer fault diagnosis tasks, showing superior performance compared with other intelligent fault diagnosis methods.