Deep Learning-based Fault Diagnosis Research Articles

Self-supervised bi-classifier adversarial transfer network for cross-domain fault diagnosis of rotating machinery

In real industrial scenarios, deep learning-based fault diagnosis has been a popular topic lately. Unfortunately, the source-trained model typically usually underperforms in target domain owning to changeable working conditions. To resolve this problem, a novel self-supervised bi-classifier adversarial transfer learning (SBATL) network by introducing self-supervised learning (SSL) and class-conditional entropy minimization is presented. Concretely, the SBATL is made up of a feature extractor, a discrepancy detector of two classifiers, and a clustering metric based on SSL, which jointly conducts self-supervised and supervised optimization in a two-stream training procedure. In the self-supervised stream, target pseudo labels obtained by SSL are used to construct the topological clustering metric for target feature optimization. In the supervised stream, the feature extractor and classifiers compete with each other in adversarial training, which bridges the discrepancy between two classifiers. Additionally, the class-conditional entropy minimization of target domain is further embedded into both streams to amend the decision boundaries of two classifiers to pass low-density regions. The results indicate that the SBATL gets better cross-domain fault diagnosis performances when compared with other popular methods.

Multi-mode data augmentation and fault diagnosis of rotating machinery using modified ACGAN designed with new framework

As one of the representative unsupervised data augmentation methods, generative adversarial networks (GANs) have the potential to solve the problem of insufficient samples in fault diagnosis of rotating machinery. However, the existing unsupervised GANs are usually incapable of simultaneously generating multi-mode fault samples and have some shortcomings such as mode collapse and gradient vanishing. To overcome these deficiencies, a supervised model called modified auxiliary classifier GAN (MACGAN) designed with new framework is proposed in this paper. Firstly, a new ACGAN framework is developed by adding an independent classifier to improve the compatibility between the classification and discrimination. Secondly, the Wasserstein distance is introduced in the new loss functions to overcome mode collapse and gradient vanishing. Finally, to achieve stable training, a spectral normalization is used to replace the weight clipping to constrain the weight parameters of discriminator. The proposed method is applied to fault diagnosis of bearing and gear. Compared with the existing GANs, the proposed method can more efficiently generate multi-mode fault samples with higher qualities, which can be used to assist the training of deep learning-based fault diagnosis models with high accuracy and good stability.

A Deep Learning-Based Fault Diagnosis of Leader-Following Systems

This paper develops a multisensor data fusion-based deep learning algorithm to locate and classify faults in a leader-following multiagent system. First, sequences of one-dimensional data collected from multiple sensors of followers are fused into a two-dimensional image. Then, the image is employed to train a convolution neural network with a batch normalisation layer. The trained network can locate and classify three typical fault types: the actuator limitation fault, the sensor failure and the communication failure. Moreover, faults can exist in both leaders and followers, and the faults in leaders can be identified through data from followers, indicating that the developed deep learning fault diagnosis is distributed. The effectiveness of the deep learning-based fault diagnosis algorithm is demonstrated via Quanser Servo 2 rotating inverted pendulums with a leader-follower protocol. From the experimental results, the fault classification accuracy can reach 98.9%.

An Improved Triplet Network for Electromechanical Actuator Fault Diagnosis Based on Similarity Strategy

Deep learning-based methods have been widely used and achieved state-of-art performance in fault diagnosis of electro-mechanical actuator (EMA). Traditional deep learning methods face three major challenges, i.e., when the training dataset is limited, unbalanced, and/or when the model is applied in a different working conditions from the training dataset. In this paper, we propose an improved deep learning-based fault diagnosis framework for EMAs based on triplet network with coupled cluster losses. Unlike the traditional approaches, the proposed framework learns to predict similarity between samples, rather than the fault labels directly. The trained model is used to calculate the distance between testing samples and a set of benchmark samples with known labels. Fault diagnosis is, then, conducted based on K-nearest neighbor algorithm. Experiments on a real-world EMA dataset from NASA show that the developed framework can improve the performance of traditional deep learning-based approaches under the three challenges.

Data Augmentation via Randomized Wavelet Expansion and Its Application in Few-Shot Fault Diagnosis of Aviation Hydraulic Pumps

In general, deep learning-based fault diagnosis methods need a large number of training samples, which are often not available in real applications. Aiming at this problem, this article develops a new data augmentation method, i.e., randomized wavelet expansion (RWE), to generate a set of synthesis samples that share similar characteristics with the original sample. The first key point is that the amplitudes of wavelet coefficients at a randomly selected frequency band are enlarged through random expansion. Another key point is that the synthesis samples are processed to have the same mean values and standard deviations as their corresponding original sample. Afterward, the synthesis samples are used as the training dataset to train a deep convolutional neural network (CNN) for implementing the few-shot fault diagnosis of aviation hydraulic pumps. Finally, the performance has been validated through a series of experiments.

Fault Diagnosis of Aeroengine Control System Sensor Based on Optimized and Fused Multidomain Feature

Improving the efficacy and dependability of aeroengines requires timely and effective sensor fault diagnosis. Deep learning-based fault diagnosis method is a current research hotspot. To overcome some of the method’s existing shortcomings and improve the reliability of fault diagnosis, this paper proposes a novel intelligent fault diagnosis framework with higher quality features and more effective fault classifiers. The proposed plan includes three stages. Firstly, multidomain features (time and frequency domain features) are extracted to describe the sensor’s health from several dimensions. Secondly, the advanced Henry gas solubility optimization algorithm (HGSO) is applied to improve classification accuracy through feature selection, and the operating conditions and the features extracted by the network are fused as fault indicators. Finally, an adaptive deep belief network (ADBN) with relu-softsign combination activation layers, variable learning rate, and optimized network structure is proposed as the fault identifier. The advantages of the first two stages lie in the complete utilization of information and reducing the data dimension. In addition, the detection performance and the convergence speed is enhanced by the proposed ADBN. The experimental data are derived from a combination of measured and simulated data generated from the aeroengine model. The experimental results indicate that the improved method can produce better performance and outcomes than the unimproved methods for all fault scenarios, with a higher diagnostic accuracy of 98.1% and a reduced time of 98 s. The efforts of this study provide a efficient and adaptable way to aeroengine sensor fault diagnosis.

Moment matching-based intraclass multisource domain adaptation network for bearing fault diagnosis

Deep learning based fault diagnosis methods assume that training and testing data with sufficient labels are available and share a same distribution. In practical scenarios, this assumption does not generally hold due to variable working conditions of rotating machineries and the difficulty in labeling vibration data under all working conditions. Transfer learning (TL) overcomes this problem by utilizing knowledge learned from the source domain to help accomplish tasks on the target domain. Although TL based fault diagnosis has been considerably studied, most studies mainly focus on single-source TL. Since multisource domains with labeled samples from which more useful knowledge can be extracted are available, in this paper, a novel multisource TL model, called the moment matching-based intraclass multisource domain adaptation network, is proposed. This model uses a feature learner to generate features of each source and target domain data to enable the joint weight classifier to predict target labels. It also introduces a moment matching-based distance metric to reduce the distance among all source domains and the target domain. During the training of the model, an intraclass alignment training strategy is applied to match the marginal and conditional distributions of each domain simultaneously. Experiments on two datasets are performed, wherein the proposed method is used to identify bearing fault types under four load conditions. Experiment results, such as high diagnostic accuracies support the reliability and generalizability of the proposed model.

An enhanced non-local weakly supervised fault diagnosis method for rotating machinery

Despite the considerable success achieved by deep learning-based fault diagnosis methods, a powerful deep learning model must require a substantial set of training data to obtain a strong generalization ability in practical application. Focusing on the fault diagnosis case under small data sets, an enhanced non-local weakly supervised fault diagnosis method is proposed in this paper. In the proposed method, the few-shot learning strategy is presented to perform learning task augmentation to alleviate the overfitting problem caused by scarce training data, and the non-local operation is investigated to further enhance the feature extraction ability of convolutional neural network through capturing the long-range dependency. Two fault diagnosis experiments which were conducted on rolling bearing and bevel gear are carried out. The proposed method obtains 97.23% and 99.76% fault diagnosis accuracy in bearing data sets and gear data sets, which outperforms the traditional fault diagnosis methods.

Synthetic Data Augmentation and Deep Learning for the Fault Diagnosis of Rotating Machines

As failures in rotating machines can have serious implications, the timely detection and diagnosis of faults in these machines is imperative for their smooth and safe operation. Although deep learning offers the advantage of autonomously learning the fault characteristics from the data, the data scarcity from different health states often limits its applicability to only binary classification (healthy or faulty). This work proposes synthetic data augmentation through virtual sensors for the deep learning-based fault diagnosis of a rotating machine with 42 different classes. The original and augmented data were processed in a transfer learning framework and through a deep learning model from scratch. The two-dimensional visualization of the feature space from the original and augmented data showed that the latter’s data clusters are more distinct than the former’s. The proposed data augmentation showed a 6–15% improvement in training accuracy, a 44–49% improvement in validation accuracy, an 86–98% decline in training loss, and a 91–98% decline in validation loss. The improved generalization through data augmentation was verified by a 39–58% improvement in the test accuracy.

Efficient federated convolutional neural network with information fusion for rolling bearing fault diagnosis

In the past year, various deep learning-based fault diagnosis methods have been designed to guarantee the stable, safe, and efficient operation of electromechanical systems. To achieve excellent diagnostic performance, the conventional centralized learning (CL) approach is adopted to collect as much data as possible from multiple industrial participants for deep model training. Due to privacy concerns and potential conflicts, industrial participants are unwilling to share their data resources. To solve the issues, this study proposes a fault diagnosis method based on federated learning (FL) and convolutional neural network (CNN), which allows different industrial participants to collaboratively train a global fault diagnosis model without sharing their local data. Model training is locally executed within each industrial participant, and the cloud server updates the global model by aggregating the local models of the participants. Specifically, an adaptive method is designed to adjust the model aggregation interval according to the feedback information of the industrial participants in order to reduce the communication cost while ensuring model accuracy. In addition, momentum gradient descent (MGD) and dropout layer are used to accelerate convergence rate and avoid model overfitting, respectively. The effectiveness of the proposed method is verified on a non-independent and identically distributed (non-iid) rolling bearing fault dataset. The experiment results indicate that the proposed method has higher accuracy than traditional fault diagnosis methods. Moreover, this study provides a promising collaborative training approach to the fault diagnosis field.

A deep learning based intelligent approach in detection and classification of transmission line faults

Detection and classification of transmission line (TL) faults are key factors for the fault root cause analysis and rapid restoration of the power network. Deep learning can extract representative features automatically from big data, thus becoming an important breakthrough in fault analysis. Existing deep learning-based fault diagnosis models rely on a large number of data obtained from a variety of fault conditions to be generalized. Indeed, in the TL domain, it is difficult to define and, therefore to collect the fault incepted features of all possible fault scenarios. This paper proposes an unsupervised framework for fault detection and classification (FDC) of TL based on a capsule network (CN). Instead of using the baseline CN, an extension to this with a sparse filtering technique is adopted in this work. The capsule network with sparse filtering (CNSF) voluntarily learns the expensive fault features and significantly improves the model performance without involving a large number of data. The proposed scheme receives 1/2 cycle post-fault three-phase signals and encodes them into a single image that is defined as the input for the proposed CNSF model. The effectiveness of the proposed CNSF model is corroborated by four different TL topology confirming the model’s adaptability to a topology change in response to intended control action or the switching actions due to cascading faults. Further assessment of the model’s performance against noise, high impedance faults (HIF), and line parameter variations are also carried out in order to confirm the high reliability of the proposed model. In addition, a rigorous comparative study is conducted to guarantee the state-of-the-art performance of the proposed model.

An explainable one-dimensional convolutional neural networks based fault diagnosis method for building heating, ventilation and air conditioning systems

Due to the frequently changed outdoor weather conditions and indoor requirements, heating, ventilation and air conditioning (HVAC) experiences faulty operations inevitably throughout its lifespan. Therefore, it is important to monitor and diagnose HVAC fault operations. Recently, deep learning methods have attracted more attentions for their guarantee of better diagnosis performance under various system configurations and operating conditions. However, these methods are black-box models which though highly accurate for fault diagnosis but are extremely hard to explain. To overcome the disadvantage of poor interpretability of deep learning black-box models, this study therefore proposes a novel explainable deep learning based fault diagnosis method that is suitable for HVACs. To maintain HVAC operational information and variable locations of all chiller input data samples, proposed method is established with three characteristics: 1) the pooling layer is excluded, 2) the size of convolution filter kernel is set as 1, and 3) use softsign as an activation function. Considering the resulting impacts of HVAC faults on system operating variables, a new Absolute Gradient-weighted Class Activation Mapping (Grad-Absolute-CAM) method is proposed to visualize the fault diagnosis criteria and make the model explainable by providing the fault-discriminative information. The proposed method is validated using fault experimental dataset of a typical building HVAC system (i.e., chiller) from the ASHRAE research project 1043 (RP-1043). The fault diagnosis accuracy is over 98.5% for seven chiller faults. Results indicates that it is capable of interpreting the model work mechanism by activation feature maps and explaining the fault diagnosis criteria by Grad-Absolute-CAM.

Multi-source transfer learning network to complement knowledge for intelligent diagnosis of machines with unseen faults

Most of the current successes of deep transfer learning-based fault diagnosis require two assumptions: 1) the health state set of source machines should overlap that of target machines; 2) the number of target machine samples is balanced across health states. However, such assumptions are unrealistic in engineering scenarios, where target machines suffer from fault types that are not seen in source machines and the target machines are mostly in a healthy state with only occasional faults. As a result, the diagnostic knowledge from source machines may not cover all fault types of target machines nor address imbalanced target samples. Therefore, we propose a framework, called a multi-source transfer learning network (MSTLN), to aggregate and transfer diagnostic knowledge from multiple source machines by combining multiple partial distribution adaptation sub-networks (PDA-Subnets) and a multi-source diagnostic knowledge fusion module. The former weights target samples by counter-balancing factors to jointly adapt partial distributions of source and target pairs, and the latter releases negative effects due to discrepancy among multiple source machines and further fuses diagnostic decisions output from multiple PDA-Subnets. Two case studies demonstrate that MSTLN can reduce the misdiagnosis rate and obtain better transfer performance for imbalanced target samples than other conventional methods.

Vibration Images-Driven Fault Diagnosis Based on CNN and Transfer Learning of Rolling Bearing under Strong Noise

Deep learning-based fault diagnosis of rolling bearings is a hot research topic, and a rapid and accurate diagnosis is important. In this paper, aiming at the vibration image samples of rolling bearing affected by strong noise, the convolutional neural network- (CNN-) and transfer learning- (TL-) based fault diagnosis method is proposed. Firstly, four kinds of vibration image generation method with different characteristics are put forward, and the corresponding pure vibration image samples are obtained according to the original data. Secondly, using CNN as the adaptive feature extraction and recognition model, the influences of main sensitive parameters of CNN on the network recognition effect are studied, such as learning rate, optimizer, and L1 regularization, and the best model is determined. In order to obtain the pretraining parameters, the training and fault classification test for different image samples are carried out, respectively. Thirdly, the Gaussian white noise with different levels is added to the original signals, and four kinds of noised vibration image samples are obtained. The previous pretrained model parameters are shared for the TL. Each kind of sample research compares the impact of thirteen data sharing schemes on the TL accuracy and efficiency, and finally, the test accuracy and time index are introduced to evaluate the model. The results show that, among the four kinds of image generation method, the classification performance of data obtained by empirical mode decomposition-pseudo-Wigner–Ville distribution (EP) is the best; when the signal to noise ratio (SNR) is 10 dB, the model test accuracy obtained by TL is 96.67% and the training time is 170.46 s.

Attention Recurrent Autoencoder Hybrid Model for Early Fault Diagnosis of Rotating Machinery

Early fault diagnosis of rotating machinery is crucial in the industry. The network parameters of the traditional deep learning-based fault diagnosis method are optimized only by the single loss function, and the extracted features are usually not the most valuable features of the input signal. This article proposes a novel method, attention recurrent autoencoder (AE) hybrid model classification algorithm, for early fault diagnosis and severity detection of rotating machinery. The AE can learn the most valuable features in an unsupervised way. By adjusting the weight proportion of the two loss functions and optimizing the multibranch network simultaneously, the proposed method enables the network to extract the most valuable features of the input signals. Moreover, the proposed method can take the raw 1-D vibration signal as the input and does not need time-frequency conversion. By introducing long short-term memory networks in AE, the time-dependent features of the data can be extracted effectively. The proposed method was verified by five kinds of gears at different pitting degrees under six load conditions. The results indicate that the method can provide simultaneous accurate fault diagnosis and severity detection for different pitting degrees.

Deep Learning-Based Fault Diagnosis of Photovoltaic Systems: A Comprehensive Review and Enhancement Prospects

Photovoltaic (PV) systems are subject to failures during their operation due to the aging effects and external/environmental conditions. These faults may affect the different system components such as PV modules, connection lines, converters/inverters, which can lead to a decrease in the efficiency, performance, and further system collapse. Thus, a key factor to be taken into consideration in high-efficiency grid-connected PV systems is the fault detection and diagnosis (FDD). The performance of the FDD method depends mainly on the quality of the extracted features including real-time changes, phase changes, trend changes, and faulty modes. Thus, the data representation learning is the core stage of intelligent FDD techniques. Recently, due to the enhancement of computing capabilities, the increase of the big data use, and the development of effective algorithms, the deep learning (DL) tool has witnessed a great success in data science. Therefore, this paper proposes an extensive review on deep learning based FDD methods for PV systems. After a brief description of the DL-based strategies, techniques for diagnosing PV systems proposed in recent literature are overviewed and analyzed to point out their differences, advantages and limits. Future research directions towards the improvement of the performance of the DL-based FDD techniques are also discussed. This review paper aims to systematically present the development of DL-based FDD for PV systems and provide guidelines for future research in the field.

Fault Diagnosis of System-Level Equipment with a Deep Learning Framework

In this paper, a deep learning-based fault diagnosis framework is proposed to improve the fault diagnosis accuracy of system-level equipment such as condenser systems in nuclear power plants. The condenser system signals are non-vibrating, slowly time-varying, and multi-dimensional in nature. Therefore, in this paper, we propose a deep learning-based fault diagnosis framework, which adopts the idea of combining data warehouse modeling and deep learning fault diagnosis, and establishes the data set required for deep learning through accurate simulation modeling of typical condenser faults, so as to make full use of the feature extraction capability of deep learning under large-scale samples. Based on this, an end-to-end deep learning model is developed for accurate diagnosis of multiple condenser faults under multiple system conditions. Through the fault diagnosis experiments on the validation set data under various system conditions, the fault diagnosis accuracy is as high as 0.9584, which verifies the effectiveness of the proposed framework in fault diagnosis of system-level equipment.

Deep Learning-Based Bearing Fault Diagnosis Method for Embedded Systems.

Bearing elements are vital in induction motors; therefore, early fault detection of rolling-element bearings is essential in machine health monitoring. With the advantage of fault feature representation techniques of time–frequency domain for nonstationary signals and the advent of convolutional neural networks (CNNs), bearing fault diagnosis has achieved high accuracy, even at variable rotational speeds. However, the required computation and memory resources of CNN-based fault diagnosis methods render it difficult to be compatible with embedded systems, which are essential in real industrial platforms because of their portability and low costs. This paper proposes a novel approach for establishing a CNN-based process for bearing fault diagnosis on embedded devices using acoustic emission signals, which reduces the computation costs significantly in classifying the bearing faults. A light state-of-the-art CNN model, MobileNet-v2, is established via pruning to optimize the required system resources. The input image size, which significantly affects the consumption of system resources, is decreased by our proposed signal representation method based on the constant-Q nonstationary Gabor transform and signal decomposition adopting ensemble empirical mode decomposition with a CNN-based method for selecting intrinsic mode functions. According to our experimental results, our proposed method can provide the accuracy for bearing faults classification by up to 99.58% with less computation overhead compared to previous deep learning-based fault diagnosis methods.

Deep Transfer Learning-Based Fault Diagnosis for Gearbox under Complex Working Conditions

In the large amount of available data, information insensitive to faults in historical data interferes in gear fault feature extraction. Furthermore, as most of the fault diagnosis models are learned from offline data collected under single/fixed working condition only, this may cause unsatisfactory performance for complex working conditions (including multiple and unknown working conditions) if not properly dealt with. This paper proposes a transfer learning-based fault diagnosis method of gear faults to reduce the negative effects of the abovementioned problems. In the proposed method, a cohesion evaluation method is applied to select sensitive features to the task with a transfer learning-based sparse autoencoder to transfer the knowledge learnt under single working condition to complex working conditions. The experimental results on wind turbine drivetrain diagnostics simulator show that the proposed method is effective in complex working conditions and the achieved results are better than those of traditional algorithms.

Bearing Fault Diagnosis Based on Multilayer Domain Adaptation

Bearing fault diagnosis plays a vitally important role in practical industrial scenarios. Deep learning-based fault diagnosis methods are usually performed on the hypothesis that the training set and test set obey the same probability distribution, which is hard to satisfy under the actual working conditions. This paper proposes a novel multilayer domain adaptation (MLDA) method, which can diagnose the compound fault and single fault of multiple sizes simultaneously. A special designed residual network for the fault diagnosis task is pretrained to extract domain-invariant features. The multikernel maximum mean discrepancy (MK-MMD) and pseudo-label learning are adopted in multiple layers to take both marginal distributions and conditional distributions into consideration. A total of 12 transfer tasks in the fault diagnosis problem are conducted to verify the performance of MLDA. Through the comparisons of different signal processing methods, different parameter settings, and different models, it is proved that the proposed MLDA model can effectively extract domain-invariant features and achieve satisfying results.