Current deep-learning methods are often based on significantly large quantities of labeled fault data for supervised training. In practice, it is difficult to obtain samples of rolling bearing failures. In this paper, a transfer learning-based feature fusion convolutional neural network approach for bearing fault diagnosis is proposed. Specifically, the raw vibration signal features and the corresponding time-frequency image features of the input data are extracted by a one-dimensional convolutional neural network and a pre-trained ConvNeXt, respectively, and connected by a feature fusion strategy. Then, the fine-tuning method based on transfer learning can effectively reduce the reliance on labeled samples in the target domain. A wide convolution kernel is introduced in the time-domain signal feature extraction to increase the receptive field, which is combined with the channel attention mechanism to further optimize the feature quality. Finally, two common bearing datasets are utilized for fault diagnosis experiments. The experimental results show that the proposed model achieves an average accuracy of more than 98.63% in both cross-working conditions and cross-device diagnosis tasks. Meanwhile, anti-noise experiments and ablation experiments further validate the accuracy and robustness of the proposed method.

Bearings are an essential component of rotating mechanical equipment. Traditional signal processing-based fault diagnosis methods usually require a massive labeled data for training, but bearings generally operate in the equipment under normal fault-free conditions. This paper proposes an improved adaptive batch normalization (AdaBN) transfer learning bearing fault diagnosis method for batch normalization (BN) in traditional deep learning architecture. The AdaBN network preprocesses the raw vibration signals, and then the preprocessed features are input to a depthwise sparable convolution neural model for training. Features are extracted by depthwise convolution and point convolution in the network. AdaBN can freeze all the parameters in the network except the BN layer. Finally, a small amount of labeled data is classified using transfer learning methods. A laboratory data set was used for validation, and the experimental validation showed that the accuracy of the bearing fault diagnosis method using AdaBN reached 85%.

Deep groove ball bearings are widely used in rotary machinery. Accurate for bearing faults diagnosis is essential for equipment maintenance. For common depth learning methods, the feature extraction of inverse time domain signal direction and the attention to key features are usually ignored. Based on the long short term memory(LSTM) network, this study proposes an attention-based highway bidirectional long short term memory (AHBi-LSTM) network for fault diagnosis based on the raw vibration signal. By increasing the Attention mechanism and Highway, the ability of the network to extract features is increased. The bidirectional LSTM network simultaneously extracts the raw vibration signal in positive and inverse time-domains to better extract the fault features. Six deep groove ball bearings with different health conditions were used to validate the AHBi-LSTM method in an experiment. The results showed that the accuracy of the proposed method for bearing fault diagnosis was over 98%, which was 8.66% higher than that of the LSTM model. The AHBi-LSTM model is also better than other relevant models for bearing fault diagnosis.

Since rolling bearings determine the stable operation of industrial equipment, it is necessary to diagnose thir faults. To improve fault diagnosis accuracy, this paper proposes a method based on a stacked sparse autoencoder (SSAE) combined with a softmax classifier. First, SSAE is used to extract the frequency-domain features of vibration signals. Then, an improved K-fold cross-validation is employed to obtain the features’ pre-train set, train set, and test set. Finally, the SSAE-model is constructed via the pre-train set, while the tuned model is built via the train set. The model performance is evaluated based on accuracy, macro-precision, macro-recall, and macro-F1 score. The proposed model is validated by the Case Western Reserve University and XJTU-SY data with 99.15% and 100% accuracy, respectively.