Few-Shot Fault Diagnosis of Rolling Bearings Using Generative Adversarial Networks and Convolutional Block Attention Mechanisms

Under actual working conditions, the initial faults of rolling bearings are difficult to be predicted effectively because of the lack of prior fault knowledge, weak fault information, and strong noise interference. How to enhance the fault characteristics while removing the vibration signal noise of rolling bearings has become the key challenge to the field of early fault diagnosis of rolling bearings. Based on the above problems, a rolling bearing initial fault diagnosis model based on second-order cyclic autocorrelation (Soca) and deep auto-encoder (AE) combined with transfer learning (TL) was proposed to improve the overall performance of rolling bearing fault identification. First, the Soca is used to estimate the second-order cyclic statistics of the vibration signal, which can better highlight the periodic pulse characteristics of rolling bearings and suppress random noise to a certain extent. After that, the dataset containing the valuable health state information of the rolling bearing is input into the denoising and contraction auto-encoder (DCAE). The advantages of denoising AEs (DAEs) and contraction AEs are fully utilized to extract the initial fault characteristics of rolling bearings. Finally, the balanced distribution adaptation (BDA) is used to reduce the distribution difference and class spacing of the transferable features extracted from the original datasets and the target datasets, and the initial fault diagnosis of the target rolling bearing is diagnosed according to the feature knowledge of the source rolling bearing. In three experimental datasets, six TL experiments were carried out to verify that the Soca-Denoising auto-encoder, Contraction auto-encoder and Transfer learning (DCT) model can effectively enhance the initial fault characteristics of rolling bearings, and has higher diagnostic accuracy and robustness compared with the existing initial fault diagnosis methods of rolling bearings. This method has a good application prospect in the field of early fault diagnosis of rolling bearings.

To address the problems of existing 2D image-based imbalanced fault diagnosis methods for rolling bearings, which generate images with inadequate texture details and color degradation, this paper proposes a novel image enhancement model based on a dual-branch generative adversarial network (GAN) combining spatial and frequency domain information for an imbalanced fault diagnosis of rolling bearing. Firstly, the original vibration signals are converted into 2D time–frequency (TF) images by a continuous wavelet transform, and a dual-branch GAN model with a symmetric structure is constructed. One branch utilizes an auxiliary classification GAN (ACGAN) to process the spatial information of the TF images, while the other employs a GAN with a frequency generator and a frequency discriminator to handle the frequency information of the input images after a fast Fourier transform. Then, a shuffle attention (SA) module based on an attention mechanism is integrated into the proposed model to improve the network’s expression ability and reduce the computational burden. Simultaneously, mean square error (MSE) is integrated into the loss functions of both generators to enhance the consistency of frequency information for the generated images. Additionally, a Wasserstein distance and gradient penalty are also incorporated into the losses of the two discriminators to prevent gradient vanishing and mode collapse. Under the supervision of the frequency WGAN-GP branch, an ACWGAN-GP can generate high-quality fault samples to balance the dataset. Finally, the balanced dataset is utilized to train the auxiliary classifier to achieve fault diagnosis. The effectiveness of the proposed method is validated by two rolling bearing datasets. When the imbalanced ratios of the four datasets are 0.5, 0.2, 0.1, and 0.05, respectively, their average classification accuracy reaches 99.35% on the CWRU bearing dataset. Meanwhile, the average classification accuracy reaches 96.62% on the MFS bearing dataset.

To realize accurate fault diagnosis of rolling bearing under random noise, a novel fault diagnosis method based on ensemble empirical mode decomposition (EEMD) and optimized Elman_AdaBoost is proposed in this paper. First, the EEMD method is used to decompose the original vibration signal into several intrinsic mode functions (IMFs). Then, the correlation coefficient and kurtosis of IMFs are used to remove any excess ingredients, and the signal is reconstructed by using the rest of IMFs. In addition, based on the analysis of the time–frequency characteristics of the reconstructed signal, the root-mean-square value and the power spectrum center are extracted as a 2-D feature vector from each reconstructed signal and regarded as the input characteristic vector of the Elman neural network. Combined with the AdaBoost classification algorithm, the Elman_AdaBoost algorithm is improved and proposed to establish the strong classifier. Finally, based on the vibration data recorded in the bearing center of the Case Western Reserve University, three kinds of bearings with different faults and normal bearings are selected as the sample data and the training samples and test samples are constructed. Experimental results show that the proposed EEMD method and optimized Elman_AdaBoost model can effectively diagnose the rolling bearing under random noise, and has the advantages of fast speed, small error, and stable performance. Comparing with the traditional Elman_AdaBoost and single Elman neural network algorithm, this paper with fewer characteristics can get better accuracy and real-time processing performance, and presents a simple and practical resolution for the fault diagnosis of rolling bearings.

As critical components, rolling bearings are widely used in a variety of rotating machinery. It is necessary to develop a suitable fault diagnosis method to prevent malfunctions and breakages of bearings during operation. However, the current methods for the fault diagnosis of rolling bearings are too cumbersome to be applied in practical engineering. In addition, the working condition of rolling bearings is generally tough, complex, and especially variable. These conditions cause fault diagnosis methods to be less effective. This paper aims to provide a simple and effective method for the fault diagnosis of rolling bearings under variable conditions. The main contribution of this paper is as follows: (1) The refined composite multiscale fuzzy entropy (RCMFE) is applied in bearing fault feature extraction because of its simplicity and high efficiency; (2) The improved support vector machine (ISVM), based on the whale optimization algorithm (WOA), is proposed to identify the fault pattern of rolling bearings. The ISVM is proposed in this paper to solve the problem that parameter setting affects the classification effect of SVM. In the ISVM, the WOA is employed to optimize both the regularization and kernel parameters of the SVM. Compared with the traditional optimization methods, the WOA has the advantages of high optimization speed and better optimization ability; (3) Combining the RCMFE and the ISVM to diagnose bearing fault under variable working conditions. The effectiveness of the RCMFE-ISVM has been validated via experimental vibration signal of bearings faults under variable working conditions.

Enhancing the operational reliability of rotary machinery relies significantly on the effective diagnosis of faults in rolling bearings. This study introduces an innovative method to improve the accuracy of fault diagnosis of rolling bearings during operation. First, we propose a sine empirical mode decomposition (SEMD) designed to effectively mitigate mode mixing and decompose the vibration signals of rolling bearings into a series of intrinsic mode functions. Subsequently, we constructed and optimized a kernel extreme learning machine classifier (KELMC) using the improved sparrow search algorithm (ISSA). Within ISSA, the opposition-based Learning method is refined and applied to enhance the optimization performance of the sparrow search algorithm. Finally, the paper presents a novel method for the fault diagnosis of rolling bearings based on SEMD and ISSA-KELMC, which can effectively extract the fault features and accurately recognize the fault types of rolling bearings by taking advantage of the SEMD and ISSA-KELMC. The effectiveness of the proposed method was verified through two simulation and fault diagnosis experiments. The results demonstrated the efficiency of the method in diagnosing faults in rolling bearings under both consistent and variable working conditions. This approach is valuable for fault diagnosis and condition monitoring of rotating machinery.

This work presents the application of a new signal processing technique, the Hilbert-Huang transform and its marginal spectrum, in analysis of vibration signals and fault diagnosis of roller bearings. The empirical mode decomposition (EMD), Hilbert-Huang transform (HHT) and marginal spectrum are introduced. First, the vibration signals are separated into several intrinsic mode functions (IMFs) by using EMD. Then the marginal spectrum of each IMF can be obtained. According to the marginal spectrum, the localized fault in a roller bearing can be detected and fault patterns can be identified. The experimental results show that the proposed method may provide not only an increase in the spectral resolution but also reliability for the fault detection and diagnosis of roller bearings.

In modern industrial systems, rolling bearings serve as critical components of rotating machinery, and their operational status directly impacts equipment safety and production efficiency. Early fault diagnosis and maintenance are essential to ensure uninterrupted equipment operation. This paper focuses on fault diagnosis of rolling bearings under time-varying conditions during the start-up phase. To achieve this goal, a fault diagnosis model based on the fusion of time-domain and frequency-domain features is proposed. First, a parallel RNN-Transformer architecture is constructed to extract local and global features from the time and frequency domains respectively, followed by their integration. Within the model structure, multi-scale local RNN modules precisely capture local signal characteristics. For global feature extraction, an adaptive causal convolution attention mechanism (ADCC) is incorporated into the Transformer module, significantly enhancing feature modelling efficiency. Secondly, a bidirectional gated cross-attention feature fusion module is developed to achieve deep complementary integration of time-domain and frequency-domain features, as well as local and global characteristics. Finally, a data preprocessing scheme combining the autocorrelation function (ACF) and Welch method is proposed to effectively suppress noise interference, providing high-quality data support for subsequent fault diagnosis. To validate the model’s efficacy, diagnostic experiments were conducted on three datasets: varying fault severity during start-up, time-varying rotational speed, and an independently collected small-sample variable-condition dataset from the start-up phase. Results demonstrate that the proposed diagnostic model outperforms existing methods in accuracy, generalisation capability, and noise resistance, establishing its advantage for fault diagnosis during the start-up phase.

Rolling bearings play a significant role in rotating machinery. Due to the failure of these components, the operations of the whole machinery are compromised and get out of service, which ultimately causes significant workload overhead and monetary loss. Many techniques are proposed for the diagnosis of faults in the rolling bearing; these techniques are manual and require a large amount of time for identification and correction of faults, which is not suitable for routine maintenance and operability of this rotating machinery. Therefore, there is a need for promising techniques for autonomous and reliable fault diagnosis in rolling bearings. Furthermore, the proposed deep learning model for fault diagnosis is not able to provide efficient and reliable results and is vulnerable to degradation problems and lack of multi-scale feature extraction. This proposed model faces the issue of the degradation problem due to the disappearance of the gradient, which ultimately compromises the optimization process. To solve these issues above, the authors proposed a novel three-dimensional (3D) Kronecker convolution feature pyramid (KCFP), which efficiently inputs the acquired data without converting time-frequency domain and pixel loss. In our model, the single dilation rate is replaced by 3D Kronecker convolution, and 3D Feature Selection (3DFSC) is used for the local learning of features. The proposed research enhances feature representation and classification accuracy while mitigating model degradation. Authors evaluate the model on the Paderborn University and MFPT bearing datasets. KCFP achieves round (99.6%) classification accuracy, outperforming the MFF-DRN (98.6%) and standard CNN (97.5%) models for the Paderborn University dataset. KCFP achieves round (97.6%) classification accuracy, outperforming the MFF-DRN (97.0%) and standard CNN (95.7%) models for the MFPT dataset. These results demonstrate the potential of KCFP for reliable and efficient rolling bearing fault diagnosis in industrial applications.

Fault diagnosis for rolling bearing based on VMD-FRFT

Aiming at the problems of low fault diagnosis accuracy caused by insufficient samples and unbalanced data sample distribution in bearing fault diagnosis, this paper proposes a fault diagnosis method for rolling bearings referencing conditional deep convolution adversarial generative networks (C−DCGAN) for efficient data augmentation. Firstly, the concept of conditional constraints is used to guide and improve the sample generation process of the original generative adversarial network, and specific constraints are added to the data generation model to perform a balanced expansion of muti-category fault data for small sample data sets. Secondly, aiming at the phenomena of training instability, gradient disappearance and gradient explosion in the imbalanced sample set, it is proposed to optimize the structure of the generative network by using the structure of self-defined skip connections and spectral normalization, while using the Wasserstein distance with penalty term instead of cross entropy. The function is used as the loss function of the generative adversarial network to improve the stable feature extraction ability of the generative network and the effect of the training process; in this way, simulation sample data with only a small variation from the real data distribution can be generated. Finally, the complete fault data set (after mixing the original data with sufficient fault category and sample number) and the generated data are input into the one-dimensional convolution neural network for fault diagnosis of rolling bearing. The experiment’s results show that the diagnosis method in this paper can improve the fault classification effect of rolling bearings by generating balanced and sufficient sample data.

Rolling bearings are the core components of rotating mechanical equipment, and it is more and more important to fault diagnosis for the operation state and safety of the mechanical equipment. Deep learning has been widely applied to fault diagnosis of rolling bearings, but it is limited with the high complexity of deep neural networks and the dependence on hardware resources. Time varying filter empirical mode decomposition (TVFEMD) method based on Subtraction-Average-Based Optimizer (SABO) optimization is applied to the subtle feature extraction of rolling bearings. Improved MobileNetV3-Large lightweight model based on Exponential Linear Units (ELU) activation function and Efficient Channel Attention (ECA) mechanism is proposed to fault diagnosis. TVFEMD-MobileNet-V3 algorithm with lightweight property is built to subtle fault diagnosis and recognition of rolling bearings. It has been validated on the bearing dataset from Case Western Reserve University (CWRU) with 99.28% accuracy and 3.57 M reference count. The anti-noise performance of TVFEMD-MobileNet-V3 has been verified, and the average accuracy is 93.83% when signal-to-noise ratio is 2 dB. TVFEMD-MobileNet-V3 algorithm has great potential to subtle feature recognition with limited sample sizes of the rolling bearing.

Intelligent fault diagnosis of rolling bearing based on wavelet transform and improved ResNet under noisy labels and environment

Intelligent acoustic-based fault diagnosis of roller bearings using a deep graph convolutional network

Rolling bearings being important components of mechanical equipment, the accurate fault diagnosis method of rolling bearings is of great importance to ensure production safety. Permutation entropy is a nonlinear measure of the irregularity of time series, which involves calculating permutation patterns, that is, defining permutations by comparing adjacent values of the time series. When using graph signal processing technology to analyze the vibration signal of rolling bearing, the natural visibility graph (NVG) can better reflect the dynamic characteristics of the vibration signal than path graph (PG). In this paper, the multiscale permutation entropy (MPE) is defined on NVG, and it is used to characterize the different fault characteristics of rolling bearings. The sand cat swarm optimization (SCSO) algorithm is employed to optimize the parameters of support vector machine (SVM); The MPEs of different faults of rolling bearing which defined on NVG are regarded as the fault feature set input into optimized SVM, and it is applied to characterize the different fault characteristics of rolling bearings, realizing fault diagnosis of rolling bearing. The proposed method is used to analyze the experimental data which contain both normal and faulty rolling bearings. The experiment results show that the proposed method can diagnose the bearing faults effectively. The MPE based on NVG is superior to MPE based on PG and MPE based on the vibration signal in distinguishing the different damage states of rolling bearings. The classification accuracy of optimized SVM based on SCSO algorithm is higher than other classical models. The effectiveness and feasibility of defining entropy on the graph signal and as the fault feature vectors for rolling bearing to realize fault diagnosis is validated. The results indicate that the proposed method can effectively detect bearing faults, and demonstrate its effectiveness and robustness for rolling bearing fault diagnosis.

PurposeThe purpose of this paper is to provide a fault diagnosis method for rolling bearings. Rolling bearings are widely used in industrial appliances, and their fault diagnosis is of great importance and has drawn more and more attention. Based on the common failure mechanism of failure modes of rolling bearings, this paper proposes a novel compound data classification method based on the discrete wavelet transform and the support vector machine (SVM) and applies it in the fault diagnosis of rolling bearings.Design/methodology/approachVibration signal contains large quantity of information of bearing status and this paper uses various types of wavelet base functions to perform discrete wavelet transform of vibration and denoise. Feature vectors are constructed based on several time-domain indices of the denoised signal. SVM is then used to perform classification and fault diagnosis. Then the optimal wavelet base function is determined based on the diagnosis accuracy.FindingsExperiments of fault diagnosis of rolling bearings are carried out and wavelet functions in several wavelet families were tested. The results show that the SVM classifier with the db4 wavelet base function in the db wavelet family has the best fault diagnosis accuracy.Originality/valueThis method provides a practical candidate for the fault diagnosis of rolling bearings in the industrial applications.