Fault Diagnosis Results Research Articles

Feature Mining and Sensitivity Analysis with Adaptive Sparse Attention for Bearing Fault Diagnosis

Bearing fault diagnosis for equipment-safe operation has a crucial role. In recent years, more achievements have been made in bearing fault diagnosis. However, for the fault diagnosis model, the representation and sensitivity of bearing fault features have a great influence on the diagnosis output results; thus, the attention mechanism is particularly important for the selection of features. However, global attention focuses on all sequences, which is computationally expensive and not ideal for fault diagnosis tasks. The local attention mechanism ignores the relationship between non-adjacent sequences. To address the respective shortcomings of global attention and local attention, an adaptive sparse attention network is proposed in this paper to filter fault-sensitive information by soft threshold filtering. In addition, the effects of different signal representation domains on fault diagnosis results are investigated to filter out signal representation forms with better performance. Finally, the proposed adaptive sparse attention network is applied to cross-working conditions diagnosis of bearings. The adaptive sparse attention mechanism focuses on the signal characteristics of different frequency bands for different fault types. The proposed network model achieves better overall performance when comparing the cross-conditions diagnosis accuracy and model convergence speed.

Active Broad-Transfer Learning Algorithm for Class-Imbalanced Fault Diagnosis

Knowledge transfer with class-imbalanced data is a challenge in predictive maintenance and fault diagnosis. Deep learning algorithms have provided promising results in fault diagnosis. However, their prediction performance is affected by class-imbalanced data in cross-domain tasks. Broad learning algorithms present promising performance in handling class-imbalanced domain-adaptation (CIDA) problems. In the presence of a domain shift, active broad transfer for class-imbalanced learning (ABTCI), an active broad-transfer learning algorithm for CIDA, is proposed. First, the ABTCI algorithm extracts the time–frequency features and feeds them into a recurrent cell to capture spatial–temporal features. Subsequently, it augments the feature space using a sparse autoencoder and an orthogonal mapping projector. By solving the ridge regression problem, the classifier is initialized. Next, the algorithm samples the target data with reliable pseudo-labels and synthesizes new data using random intraclass interpolation among the minor classes containing source and target knowledge. Finally, the classifier is updated using an incremental continuous learning strategy. The performance of the ABTCI algorithm is validated using three datasets, which include 20 class-balanced and 27 class-imbalanced transfer tasks. The performance of the proposed algorithm benchmarked against other deep-learning algorithms is promising.

CNN parameter design based on fault signal analysis and its application in bearing fault diagnosis

As a representative deep learning network, Convolutional Neural Network (CNN) has been extensively used in bearing fault diagnosis and many good results have been reported. In Prognostics and Health Management (PHM) field, the CNN’s input size is usually designed as a 1D vector or 2D square matrix, and the convolution kernel size is also defined as a square shape like 3 × 3 and 5 × 5, which are directly adopted from the image recognition. Though satisfying results can be obtained, CNN with such parameter specifications is not optimal and efficient. To this end, this paper elaborated the physical characteristics of bearing acceleration signals to guide the CNN design. First, the fault period under different fault types and shaft rotation frequency were used to determine the size of CNN’s input. Next, an exponential function was involved in fitting the envelope of decaying acceleration signal during each fault period, and signal length within different decaying ratios was used to define the CNN’s kernel size. Finally, the designed CNN was validated with the Case Western Reserve University bearing dataset and Paderborn University bearing dataset. Results confirm that the physics-guided CNN (PGCNN) with rectangular input shape and rectangular convolution kernel works better than the baseline CNN with higher accuracy and smaller uncertainty. The feasibility of designing CNN parameters with physics-guided rules derived from bearing fault signal analysis has also been verified.

A New Multisource Domain Bearing Fault Diagnosis Method With Adaptive Dual-Domain Obfuscation Weighting Strategy

In the fault diagnosis of modern industrial equipment, domain adaptation is widely utilized to address the challenge of data distribution changes. However, traditional domain adaptation method only learns single domain knowledge and experiences difficulty in coping with scenarios wherein working conditions change drastically. Accordingly, learning knowledge from multiple different source domains can effectively improve the generalization ability of the model and the reliability of the diagnosis results. Simultaneously, aligning marginal and conditional distributions can weaken data distribution differences caused by drastic changes in operating conditions, enabling more accurate knowledge transfer between source and target domains, and improving the performance of model cross-domain fault diagnosis. Furthermore, when the data comes from multiple different domains, completely aligning all source domains is arduous and impractical, and thus, selectively synthesizing different source domain migration diagnostic results is necessary. To address the aforementioned issues, this study proposes a multibranch weight-adaptive adversarial fault diagnosis method. This method uses multiple branch networks to match each source–target domain pair independently, aligns them as much as possible with two domain confusion networks, and calculates the alignment degree to assign the source domain weight coefficient. The final fault diagnosis result is obtained by synthesizing the domain classifier results that correspond to the weighted coefficient domain. Experiments verify the reliability and stability of this method under various working conditions, such as high span speed and variable speed.

Fault diagnosis of rolling bearing based on deep convolutional neural network and gated recurrent unit

Rolling bearing is an essential part of various rotating machines, and its signal is the typical nonlinear signal. Traditional fault diagnosis usually relies on manual experience to extract the features of signals first. Deep convolutional neural networks (DCNN) can make fuller use of time series than traditional convolutional neural networks (CNN). Because of the low accuracy rate, fault diagnosis using gated recurrent unit (GRU) alone is not unsatisfactory. In order to improve the temporality of one-dimensional convolutional neural networks (1D-CNN) and enhance the accuracy of GRU, a novel fault diagnosis method called deep convolutional neural networks - gated recurrent unit (DCNN-GRU) is first put forward, which combines DCNN with GRU. The original signals without preprocessing are input into the DCNN, and the outputs of DCNN are input into the GRU consequently. Then the faults of rolling bearing can be diagnosed effectively. As the post-processing method, the t-distributed stochastic neighbor embedding (t-SNE) method is applied to visualize the fault diagnosis results. Six different network models, including the DCNN-GRU, are used to train the same fault dataset for comparison. The simulation results show that the proposed method can reach more than 99.9% accuracy stably for the given dataset, which can verify the feasibility and effectiveness of proposed method. And the DCNN-GRU can also be verified with good generalization ability using different dataset.

Gearbox fault diagnosis based on a fusion model of virtual physical model and data-driven method

Planetary gear system is widely used in wind power, construction machinery and helicopters, because of its large transmission ratio, strong load-bearing capacity and small size. Because the gears are usually in high-speed operation, their key components are subject to severe wear and impact damage. When the gearbox is subjected to a failure, it is often difficult to judge the operating status of the gearbox in real time. Since the gearbox system is complex, moreover, it is still difficult for the simulation model to accurately identify the gearbox fault. To address this issue, this paper proposes a digital twin-based signal fusion model to accurately identify gearbox faults at the signal level. The proposed model can identify minor faults occurring in planetary gearboxes by fusing actual and virtual signals. Firstly, on the basis of virtual prototype and the multibody dynamics software, a digital physical model based on Hertz theory and extended finite element method is proposed for gearbox dynamic monitoring and digital simulation, in which the gearbox can be visualized. Secondly, the data-driven method is used to extract and select features from real and simulation signals, and the fault of the planetary gearbox is detected by feature matching. Finally, the real and simulation signals are fused effectively based on variational mode decomposition and phase correction of generalized cross-correlation coefficient. The fault diagnosis result of the gearbox is finally determined by using the envelope spectrum of the fused signal. The experimental results show that the virtual real fusion method effectively improves the performance of the traditional signal diagnosis methods, and the fusion of the actual and virtual signals strengthens the fault features of the fused signals effectively.

Automated broad transfer learning for cross-domain fault diagnosis

Smart manufacturing system pursues automated modeling algorithms for industrial applications in dynamic environments. The prevalent deep transfer learning (DTL) has achieved promising results in cross-domain fault diagnosis. However, most DTL algorithms are dataset- and domain-specific. They require hyperparameter optimization (HPO) calling for prior knowledge to accomplish a promising prediction performance. This dilemma persists when new data from different domains arrive. An automated broad-transfer learning algorithm (AutoBTL) is proposed to improve predictive modeling for cross-domain tasks. AutoBTL includes three components, a broad classifier, an active estimator, and a hyperparameter optimizer for solving the HPO problem in cross-domain fault diagnosis. At each iteration, AutoBTL initiates a gated broad architecture assigned by the optimizer for target prediction. Then, an active estimator samples the target data with reliable pseudo labels for domain adaptation and performance evaluation. Finally, the optimizer updates a surrogate model and optimizes the hyperparameter space. These steps get repeated until satisfactory model is generated. AutoBTL involves a transductive joint validation strategy, which significantly improves the performance of the existing HPO algorithms in cross-domain tasks. The performance of the AutoBTL algorithm is validated with three benchmark datasets, including 14 cross-domain tasks. The computational results have demonstrated the accuracy and efficiency of the proposed algorithm over the widely used predictive modeling algorithms.

Deep hypergraph autoencoder embedding: An efficient intelligent approach for rotating machinery fault diagnosis

Intelligent fault diagnosis based on deep learning (DL) has been widely used in various engineering practices. However, when confronting massive unlabeled industrial data, traditional data-driven intelligent fault diagnosis approaches cannot fully mine the correlation geometric structure information between data, and therefore cannot obtain good fault diagnosis results. To overcome this difficulty, an efficient fault diagnosis method based on deep hypergraph autoencoder embedding (DHAEE) is presented in this study. First, unlabeled vibration signals are converted into hypergraphs by applying the designed hypergraph construction method. Second, a hypergraph convolutional extreme learning machine autoencoder (HCELM-AE) is designed, which can mine the higher-order structural information and subspace structural information of the original unlabeled data by designing hypergraph convolutional and self-representation layers. Furthermore, by stacking multiple HCELM-AE modules in a DL framework, the DHAEE and its corresponding fault diagnosis method is constructed, which not only has the advantage of high computational efficiency of ELM-AE, but also has strong representational learning ability of DL methods. Finally, the effectiveness of the DHAEE based fault diagnosis method is verified by rolling bearing fault data and rotor fault data. Experimental results show that the presented fault diagnosis method has higher accuracy and lower computational complexity than other comparison methods, thus proving that DHAEE is an efficient intelligent massive unlabeled data processing approach for rotating machinery fault diagnosis.

A New Topology-Switching Strategy for Fault Diagnosis of Multi-Agent Systems Based on Belief Rule Base.

Effective fault-diagnosis strategies have been the focus of research on multi-agent systems (MASs). In this paper, the belief rule base (BRB)-based distributed fault-diagnosis problem for MASs is investigated, and a topology-switching strategy is developed to increase the reliability of fault-diagnosis model. Firstly, a BRB-based distributed fault-diagnosis model is constructed for the MAS with multiple faults, then expert knowledge is used to judge whether the agent is faulty. Then, considering that the system may be influenced by the fault or some other factors and thus leading to a decrease in the accuracy of the fault-diagnosis results, a topology-switching strategy based on the average distance of the output diagnosis accuracy is proposed to update the topology of the agent so that the fault-diagnosis results can be more reliable. Note that the topology-switching threshold is designed based on the average distance between the accuracy of the fault diagnosis of each agent. The method proposed in this paper can solve the problem when the fault-diagnosis accuracy of the model is affected by some common factors and thus decreases, and can improve the reliability of the fault-diagnosis model very well. Finally, the effectiveness of the BRB-based distributed fault-diagnosis model and the proposed topology-switching strategy to improve the fault-diagnosis accuracy is verified by simulation examples.

Spatial graph convolutional neural network via structured subdomain adaptation and domain adversarial learning for bearing fault diagnosis

Unsupervised domain adaptation (UDA) has shown remarkable results in fault diagnosis under changing working conditions in recent years. However, most UDA methods do not consider the geometric structure of the data. Furthermore, the global domain adaptation technique is commonly applied, which ignores the relation between subdomains. This paper addresses mentioned challenges by presenting the novel deep subdomain adaptation graph convolution neural network (DSAGCN), which has two key characteristics: First, a graph convolution neural network (GCNN) is employed to model the structure of data. Second, adversarial domain adaptation and local maximum mean discrepancy (LMMD) methods are applied concurrently to align the subdomain’s distribution and reduce structure discrepancy between relevant subdomains and global domains. CWRU, PU, and JNU bearing datasets are used to validate the DSAGCN method’s superiority between comparison models. The experimental results demonstrate the significance of aligning structured subdomains along with domain adaptation methods to obtain an accurate data-driven model.

Multisensor Fusion Time–Frequency Analysis of Thruster Blade Fault Diagnosis Based on Deep Learning

With the rapid development of marine robots, detecting abnormalities in propulsion systems is important during sailing as the unperceived damage of thrusters and propellers can cause substantial losses. In this study, different fault conditions of blades, including healthy, fully broken, half-broken, and simulated biofouling, are discussed. Current and sound signals are collected by a Hall element and hydrophone, respectively, to diagnose the propeller under different rotating speeds. The experiments include an ideal condition (swimming pool) and a noisy condition (lake). The raw data of time-domain signals are transformed into a time–frequency domain and shown as an image. A modified convolutional neural network (CNN) based on merging two signals is proposed to classify the faults. To compare the performance of models, the networks use single and multiple signals as input. The results demonstrate that the proposed multiple signals method achieves the best propeller fault diagnosis results, particularly when two signals are first trained separately and then merge at the end (99.94% in a swimming pool and 99.06% in a lake). Finally, the model was applied to Nvidia Jetson TX2 to verify the computing performance of an embedded system.

Fault diagnosis method based on the multi-head attention focusing on data positional information

In order to make full use of the absolute position information of fault signal, this paper designs a new multi-head attention (MHA) mechanism focusing on data positional information, proposes a novel MHA-based fault diagnosis method and extends it to the fault diagnosis scenario with missing information. Based on the absolute positional information and the trainable parameter matrix of the fault data, a novel attention weight matrix is generated, and the fault features are extracted by a fully connected network with the attention mechanism. By integrating the positional information into the weight matrix, the new MHA mechanism has the ability to extract more effective data features, compared with the traditional MHA method. Furthermore, the proposed method is also developed for the fault diagnosis scenarios with missing information. A special attention weight modified method is designed to reduce the impact of missing data on fault diagnosis results. In the experiment simulations, the data sampled from ZHS-2 multi-function motor flexible rotor test bed and the Tennessee- Eastman process data are utilized to test the performance of the algorithm. The results show that the proposed method can effectively extract fault features and reduce the impact of missing data.

Real-Time Fault Diagnosis for EVs With Multilabel Feature Selection and Sliding Window Control

Real-time fault diagnosis on vehicles can effectively avoid potential accidents, which, however, is difficult and challenging to be widely deployed due to the low computational capability and limited data storage of electric vehicles (EVs). To address this issue, we propose a vehicle-mounted fault diagnosis system with low computational complexity and small data storage, for achieving real-time monitoring of vehicle status. To facilitate the accurate and optimized feature selection, we had been collecting 6.52-GB real data from three EVs in 12 months. Motivated by those data, we first propose a multilabel feature selection algorithm to obtain the feature weights, based on which the optimal number of features is then calculated through the backpropagation neural network (BPNN), thus minimizing the computational cost of real-time fault diagnosis regarding sample dimensions. To further simplify the fault diagnosis system, i.e., reducing the minimum required capacity of data storage, we design a real-time diagnosis sliding window (RDSW) where the window moves forward as new samples arrive and the stale data outside the window are discarded. In particular, we calculate the optimal size of RDSW, which controls the minimum required number of samples to guarantee the accuracy of real-time fault diagnosis. Owing to the mechanism of RDSW, vehicles no longer need to store massive data to guarantee the accuracy of real-time fault diagnosis. In addition, the results of real-time fault diagnosis at each vehicle can be shared with other vehicles in cooperative intelligent transportation systems (C-ITS). Finally, comprehensive simulation is conducted to validate the effectiveness of the proposed diagnosis system in terms of accuracy, complexity and storage capacity.

Application of Modern Multimedia and Sensing Technology in Fault Detection and Diagnosis of Hydraulic Agricultural Machinery

In order to solve the problem of fault diagnosis of agricultural machinery hydraulic system, the authors propose an application of modern multimedia and sensing technology in fault detection and diagnosis of hydraulic agricultural machinery. By analyzing the component faults and system faults of the hydraulic system of agricultural machinery, an expert system for fault diagnosis is constructed, and a knowledge base and inference engine suitable for fault diagnosis of the hydraulic system of agricultural machinery are designed. In order to further improve the diagnostic accuracy of the fault diagnosis expert system, a fault diagnosis based on sparse coding is designed, and the sparse coding fault diagnosis results are integrated with the expert system to improve the diagnostic accuracy. The experimental results show that after sparse coding fusion, the fault diagnosis accuracy can be improved to more than 91%. In conclusion, the model meets the requirements of fault diagnosis and puts forward a new idea for the fault diagnosis of agricultural machinery hydraulic system. Make the agricultural machinery and equipment industry develop healthily.

Application of multi-kernel relevance vector machine and data pre-processing by complementary ensemble empirical mode decomposition and mutual dimensionless in fault diagnosis

Resistance signals of a faulty building electrical system contain a large amount of information about the electrical systems operating status. However, it is difficult to extract the fault features completely because of their characteristics of nonlinearity and non-stationarity which brings a problem of a relatively low fault identification rate of the current fault diagnosis methods based on pattern recognition. Aiming at improving the accuracy of fault diagnosis further, this paper proposes a fault diagnosis method of a building electrical system based on the complementary ensemble empirical mode decomposition and mutual dimensionless index extraction (CEEMD-MDI) combined with the multi-kernel relevance vector machine (MK-RVM). First, the resistance signals of a faulty building electrical system are decomposed into a series of intrinsic mode functions (IMFs) by using an adaptive decomposition ability of the CEEMD. Second, the IMFs are used to extract the MDI and to form a feature vector with the resistance signal. Finally, the processed feature vector is input into the MK-RVM for modeling, and the fault diagnosis result of the building electrical system is provided in the form of a probability output. The experimental results show that the fault diagnosis accuracy rate of the proposed method based on CEEMD-MDI and MK-RVM can reach 97.22%, which has better fault diagnosis performance compared with other methods.

Neural Network Based on Health Monitoring Electrical Equipment Fault and Biomedical Diagnosis.

In order to improve the accuracy of electrical equipment failure diagnosis and keep electrical equipment operating safely and efficiently, this paper proposes to design an electrical equipment failure diagnosis system based on a neural network, analyze the faults of electrical equipment and their causes, and establish knowledge base according to relevant data and expert judgment. The fault knowledge base was introduced into the neural network operation structure, and the fault diagnosis results were classified step by step through multiple subnetworks. In data preprocessing, in order to avoid the redundancy of primary fault information features, the principal component heuristic attribute reduction algorithm was used to select the fault data samples optimally. The neural network learning algorithm is used to calculate the forward direction and error rate of the initial error data, and the reliability function is used to optimize the initial weight threshold of the neural network, propagating the error backwards and high. Experimental results show that adding attribute reduction improves error classification performance, avoids the problem of local minima through neural network operation, and has fewer iteration steps, lower average error, and higher accuracy of fault diagnosis, reaching 95.6%.

Bearing Fault Feature Extraction Method Based on Enhanced Differential Product Weighted Morphological Filtering.

Aimed at the problem of fault characteristic information bearing vibration signals being easily submerged in some background noise and harmonic interference, a new algorithm named enhanced differential product weighted morphological filtering (EDPWMF) is proposed for bearing fault feature extraction. In this method, an enhanced differential product weighted morphological operator (EDPWO) is first constructed by means of infusing the differential product operation and weighted operation into four basic combination morphological operators. Subsequently, aiming at the disadvantage of the parameter selection of the structuring element (SE) of EDPWO depending on artificial experience, an index named fault feature ratio (FFR) is employed to automatically determine the flat SE length of EDPWO and search for the optimal weighting correlation factors. The fault diagnosis results of simulation signals and experimental bearing fault signals show that the proposed method can effectively extract bearing fault feature information from raw bearing vibration signals containing noise interference. Moreover, the filtering result obtained by the proposed method is better than that of traditional morphological filtering methods (e.g., AVG, STH and EMDF) through comparative analysis. This study provides a reference value for the construction of advanced morphological analysis methods.

Ensemble of metric learners for improving electrical submersible pump fault diagnosis

Machine learning classification algorithms play a major role in diagnosing faults in industrial equipment. In this paper, we investigate the use of ensembles composed of deep neural networks for improving the electrical submersible pump fault diagnosis results. The proposed method relies on composing an ensemble of multiple convolutional neural networks trained with a metric function for extracting relevant features directly from the raw data. The final classification is given by a standard voting scheme after a Random Forest is trained for each feature set generated by each deep metric neural network and then the methods are compared with two previous methods already used for the electrical submersible pump fault diagnosis. The experiments were carried out using five different metric functions: Proxy-Anchor loss, CosFace loss, Triplet loss, Lifted Structured loss and Contrastive loss. Results show statistical evidence that the new approach using ensemble methods achieves better performance than the previous solutions. Moreover, results indicate that composing an ensemble of multiple distinct metric losses achieves a high macro F-measure with low variance when compared to an ensemble where all neural networks are trained with the same metric loss.

A multi-scale feature fusion network-based fault diagnosis method for wind turbine bearings

A fault diagnosis method based on a multi-scale feature fusion network (MSFF-CNN) is proposed for the problem that the vibration signals of wind turbine bearings are easily disturbed by noise, and feature extraction is harrowing. Compared with the traditional diagnosis method, which has two stages of manual feature extraction and fault classification, this method combines the two into one. First, based on the characteristics of the bearing vibration signal, the multi-scale kernel algorithm is used to learn features in parallel at different scales. Then, the features extracted at different scales are fused to obtain complementary and rich diagnostic information. Finally, the Softmax classifier is used to output the fault diagnosis results. The simulation is carried out through the bearing vibration data of Case Western Reserve University. The results show that the accuracy of bearing fault diagnosis reaches 99.17%, proving the proposed method’s high accuracy and effectiveness.

Fault detection and diagnosis strategy based on k-nearest neighbors and fuzzy C-means clustering algorithm for industrial processes

Fault detection and diagnosis is crucial in recent industry sector to ensure safety and reliability, and improve the overall equipment efficiency. Moreover, fault detection and diagnosis based on k-nearest neighbor rule (FDD-kNN) has been effectively applied in industrial processes with characteristics such as multi-mode, non-linearity, and non-Gaussian distributed data. The main challenge associated with FDD-kNN is the on-line computational complexity and storage space that are needed for searching neighbors. To deal with these issues, this paper proposes a monitoring approach where the Fuzzy C-Means clustering technique is used to decrease the overall on-line computations and required storage by measuring the neighbors of the clusters’ centres rather than the raw data. After building the monitoring model off-line based on the data clusters’ centres, the faults are detected by comparing the average squared Euclidean distance between the on-line data sample and the clusters’ centres with a predefined threshold. Then, the detected faults can be diagnosed by calculating the contribution of each variable in the fault detection index. Furthermore, for easily analysing the fault diagnosis results, the relative contribution for each sample data vector is considered. A numerical example and the Tennessee Eastman chemical process are used to demonstrate the performance of the proposed FCM-kNN for fault detection and diagnosis.