Rolling bearings are the core components of rotating machinery. Their health directly affects the performance, stability and life of rotating machinery. To prevent possible damage, it is necessary to detect the condition of rolling bearings for fault diagnosis. With the rapid development of intelligent fault diagnosis technology, various deep learning methods have been applied in fault diagnosis in recent years. Convolution neural networks (CNN) have shown high performance in feature extraction. However, the pooling operation of CNN can lead to the loss of much valuable information and the relationship between the whole and the part may be ignored. In this study, we proposed CNNEPDNN, a novel bearing fault diagnosis model based on ensemble deep neural network (DNN) and CNN. We firstly trained CNNEPDNN model. Each of its local networks was trained with different training datasets. The CNN used vibration sensor signals as the input, whereas the DNN used nine time-domain statistical features from bearing vibration sensor signals as the input. Each local network of CNNEPDNN extracted different features from its own trained dataset, thus we fused features with different discrimination for fault recognition. CNNEPDNN was tested under 10 fault conditions based on the bearing data from Bearing Data Center of Case Western Reserve University (CWRU). To evaluate the proposed model, four aspects were analyzed: convergence speed of training loss function, test accuracy, F-Score and the feature clustering result by t-distributed stochastic neighbor embedding (t-SNE) visualization. The training loss function of the proposed model converged more quickly than the local models under different loads. The test accuracy of the proposed model is better than that of CNN, DNN and BPNN. The F-Score value of the model is higher than that of CNN model, and the feature clustering effect of the proposed model was better than that of CNN.

In view of the limitations of existing rotating machine fault diagnosis methods in single-scale signal analysis, a fault diagnosis method based on multi-scale permutation entropy (MPE) and multi-channel fusion convolutional neural networks (MCFCNN) is proposed. First, MPE quantitatively analyzes the vibration signals of rotating machine at different scales, and obtains permutation entropy (PE) to construct feature vector sets. Then, considering the structure and spatial information between different sensor measurement points, MCFCNN constructs multiple channels in the input layer according to the number of sensors, and each channel corresponds to the MPE feature sets of different monitored points. MCFCNN uses convolutional kernels to learn the features of each channel in an unsupervised way, and fuses the features of each channel into a new feature map. At last, multi-layer perceptron is applied to fuse multi-channel features and identify faults. Through the health monitoring experiment of planetary gearbox and rolling bearing, and compared with single channel convolutional neural networks (CNN) and existing CNN based fusion methods, the proposed method based on MPE and MCFCNN model can diagnose faults with high accuracy, stability, and speed.

Multi-view information fusion can provide more accurate, complete and reliable data descriptions for monitoring objects, effectively improve the limitations and unreliability of single-view data. Existing multi-view information fusion based on deep learning mostly focuses on the feature level and decision level, with large information loss, and does not distinguish the view weight in the fusion process. To this end, a multi-view data level information fusion model CAM_MCFCNN with view weight was proposed based on a channel attention mechanism and convolutional neural network. The model used the channel characteristics to implement multi-view information fusion at the data level stage, which made the fusion position and mode more natural and reduced the loss of information. A multi-channel fusion convolutional neural network was used for feature learning. In addition, the channel attention mechanism was used to learn the view weight, so that the algorithm could pay more attention to the views that contribute more to the fault identification task during the training process, and more reasonably integrate the information of different views. The proposed method was verified by the data of the planetary gearbox experimental platform. The multi-view data and single-view data were used as the input of the CAM_MCFCNN model and single-channel CNN model respectively for comparison. The average accuracy of CAM_MCFCNN on three constant-speed datasets reached 99.95%, 99.87% and 99.92%, which was an improvement of 0.95%, 2.25%, and 0.04%, compared with the single view with the highest diagnostic accuracy, respectively. When facing limited samples, CAM_MCFCNN had similar performance. Finally, compared with different multi-view information fusion algorithms, CAM_MCFCNN showed better stability and higher accuracy. The experimental results showed that the proposed method had better performance, higher diagnostic accuracy and was more reliable, compared with other methods.

Deep neural network (DNN), with the capacity for feature inference and nonlinear mapping, has demonstrated its effectiveness in end-to-end fault diagnosis. However, the intermediate learning process of the DNN architecture is invisible, making it an uninterpretable black-box model. In this paper, a stacked residual multi-attention network (SRMANet) is proposed as a means of feature extraction of vibration signals, and visualizing the model training process, designing Squeeze-excitation residual (SE-Res) blocks to obtain additive features with minimal redundancy and sparsity. This study recommends the use of the attention fusion unit to ensure the interpretability of the model and ultimately to obtain representative features. By feeding the output gradient of the attention layer back to the original signal, the key feature components in the time domain signal can be effectively captured. Finally, the interpretability, identification accuracy and adaptability of the model under different operating conditions are verified on 12 different fault tasks in the planetary gearbox.

Generative models have been applied in many fields and can be evaluated with many methods. In the evaluation of generative models, the proper evaluation metric varies with the application field. Therefore, the evaluation of generative adversarial networks is inherently challenging. In this study, conditional deep convolutional generative adversarial networks were applied in mechanical fault diagnosis and then evaluated. We proposed three evaluation metrics of conditional deep convolutional generative adversarial networks: Jensen-Shannon divergence, kernel maximum mean discrepancy, and the 1-nearest neighbor classifier which were used to distinguish generated samples from real samples, test mode collapsing and detect overfitting based on the dataset of Electronic Engineering Laboratory of Case Western Reserve University and the planetary gearbox dataset measured in the laboratory. The Jensen-Shannon divergence could not well distinguish generated samples from real samples. However, the two metrics (maximum mean discrepancy and 1-nearest neighbor classifier) well-distinguished generated samples from real samples, thus verifying the applicability of conditional deep convolutional generative adversarial networks in the field of mechanical diagnosis.