Learning-based Fault Diagnosis Research Articles

Dilated dynamic supervised contrastive learning framework for fault diagnosis under imbalanced dataset conditions

Deep learning-based fault diagnosis methods have achieved significant results. However, under actual working conditions, the inability of equipment to operate for long periods in a fault state results in a much smaller amount of collected fault data compared to normal data. This leads to existing fault diagnosis models being able to learn only limited diagnostic knowledge, causing a substantial decline in diagnostic accuracy. Therefore, a dilated dynamic supervised contrast learning framework for imbalanced data scenarios is proposed. In this framework, a dilated convolution network is designed as a feature extractor to enhance the feature extraction ability by changing the dilation rate. Then, the dynamic supervision loss function is designed, and the sample compensation factor of each category is given according to the frequency of sample used in the training process to enhance the contribution of the minority class and optimize the feature extractor. Finally, a dynamic cross-entropy loss function is designed to train the network and classifier to achieve accurate classification of faults. Experiments on two open-source datasets show the effectiveness of the proposed model framework.

A novel multi-sensor local and global feature fusion architecture based on multi-sensor sparse Transformer for intelligent fault diagnosis

Deep learning has been widely used for intelligent fault diagnosis of rotating machinery. However, owing to the limitations of training sample data and the complex industrial environments with variable operating conditions and noise interference, the existing deep learning-based fault diagnosis methods have difficulty achieving satisfactory performance. To address these issues, this paper proposes a novel multi-sensor local and global feature fusion architecture for intelligent fault diagnosis. First, through the integration of a stem structure and one-dimensional convolutions, a local feature perception mechanism is constructed to learn the sensor-specific local features within each sensor. Second, a global feature perception mechanism, which incorporates a multi-sensor sparse Transformer and a hierarchical architecture, is established to fully explore the sensor-specific global features within each sensor and the cross-sensor global features among multiple sensors. Third, the sensor-specific and cross-sensor global features are fed into a feature aggregation module to obtain the final multi-sensor global features. Finally, these multi-sensor global features are classified through a multi-sensor feature classifier to obtain diagnostic results. The experimental results obtained for a gear case and an inter-shaft bearing case demonstrate the superior diagnostic performance of the proposed method compared to the state-of-the-art comparative methods under limited training samples and complex environments.

Time-segment-wise feature fusion transformer for multi-modal fault diagnosis

Mechanical fault diagnosis is crucial to ensure the safe operations of equipment in intelligent manufacturing systems. Recently, deep learning based fault diagnosis methods have achieved remarkable advancements with monitored data from a single sensor. However, obtaining satisfactory diagnostic results based on a single sensor is often difficult because the complementary information between different sensors is ignored. Extracting comprehensive fault features from multi-modal data is a problem that remains to be solved. To address these challenges, a time-segment-wise feature fusion Transformer (FFTR) is proposed in this paper. First, the signals from various modalities as multiple channels are normalized channel-by-channel and form a multi-modal sample. Second, a time-segment-wise feature learning network is designed to transform a multi-modal sample into several fusion features through the sequential processes of sample segmentation, segment-level feature extraction and time-aligned feature fusion. Finally, a Transformer network is employed for comprehensive multi-modal feature analysis and fault classification. In addition, a joint loss function is designed to comprehensively train the end-to-end FFTR. The comparison experiment with other baseline methods is conducted on two multi-modal datasets. The experimental results show that FFTR achieves 3.69% and 3.93% higher diagnostic accuracy than baseline on two datasets respectively and can address real-world problems effectively.

Recursive prototypical network with coordinate attention: A model for few-shot cross-condition bearing fault diagnosis

In a practical industrial scenario, the variability in bearing operating conditions complicates the collection of a sufficient number of labeled samples, thereby limiting the effectiveness of traditional deep learning-based fault diagnosis methods. In addition, the influence of abnormal samples on the prototype features also severely limits the performance of prototypical network in few-shot fault diagnosis. To address the above issues, a recursive prototypical network based on meta-learning is proposed for few-shot cross-condition bearing fault diagnosis. Firstly, a feature extractor with coordinate attention mechanism is developed, which is able to deeply extract effective features in complex vibration signals. Furthermore, a recursive prototype computation module is introduced to alleviate prototype bias arising from abnormal samples, thereby achieving a more precise representation of prototypes. Finally, a metric module is utilized to obtain the similarity between the prototypes and the query set samples to achieve an accurate classification of faults. To verify the efficacy and superiority of the proposed method, its performance was evaluated on two bearing vibration datasets. The experimental results demonstrated that the method is significantly better than other deep learning methods with high accuracy and generalization, and greater suitability for few-shot cross-condition bearing fault diagnosis tasks.

Digital twin-assisted fault diagnosis framework for rolling bearings under imbalanced data

The application of deep learning-based fault diagnosis methods is constrained by the imbalanced data. Recently, many studies have suggested integrating dynamic model responses into the training process to address data imbalances. However, significant distribution discrepancies exist between dynamic model responses and real measured data, resulting in suboptimal performance. To address this challenge, this research proposes a digital twin-assisted framework for rolling bearings fault diagnosis under imbalanced data, which minimizes the distribution discrepancies between dynamic model responses and real measured data through information and feature transfer. Firstly, a Digital Twin-assisted Data Fusion Strategy (DTDFS) is proposed to facilitate information transfer from physical entities to dynamic models, generating digital twin data for data augmentation. Subsequently, a Frequency Filter Subdomain Adaptation Network (FFSAN) is proposed to achieve feature transfer between twin data and measured data. Finally, experimental results and engineering applications demonstrate that the proposed framework significantly outperforms existing imbalanced fault diagnosis methods, which is crucial to the application of deep learning-based fault diagnosis in industrial settings.

Bearing fault diagnosis method for unbalance data based on Gramian angular field

In the application of deep learning-based fault diagnosis, more often than not, the network model could perform better with a balanced dataset input, whereby the number of fault data is equivalent to that of normal data. However, in the context of real-world applications, the number of fault data is generally insufficient compared to the normal data. In this study, a new approach for fault diagnosis in unbalanced data sets is proposed using the Gramian angular field (GAF) method. Firstly, the GAF method is employed to convert one-dimensional data into two-dimensional data, which enhances the feature extraction process. Secondly, to balance the sample distribution, fault data is generated using Generative Adversarial Networks (GANs). Finally, the Residual neural network (ResNet) with an attention mechanism is utilized to improve the accuracy of fault diagnosis. The proposed method is experimentally validated using open-source bearing datasets that are published by Case Western Reserve University and the University of Ottawa. The experimental results show that the proposed method has greatly improved fault diagnosis performance in cases of data distribution imbalance, surpassing that of the compared methods.

A review on adversarial–based deep transfer learning mechanical fault diagnosis

Mechanical equipment is a vital foundational support for promoting national economic development and is widely utilized in key sectors such as aerospace, shipping, construction machinery, energy, petrochemicals, and robotics. With the advancement of artificial intelligence and industrial intelligence, industrial big data and its intelligent analysis provide robust support for fault prediction and health management of equipment. Building on existing research, intelligent diagnostics for mechanical equipment based on deep learning have gained significant attention and application. However, the success of big data relies on comprehensive fault data, which is challenging to obtain in practical applications where continuous equipment operation is essential. Moreover, mechanical equipment often operates under varying conditions, leading to different data distributions for training and testing. This discrepancy can result in low diagnostic accuracy or even failure of deep learning methods. Deep Transfer Learning (DTL) is an emerging machine learning paradigm that not only leverages the advantages of deep learning (DL) in feature representation but also harnesses the strengths of transfer learning (TL) in knowledge transfer. Consequently, DTL techniques can make deep learning-based fault diagnosis methods more reliable, robust, and applicable, leading to extensive development and research in the field of intelligent fault diagnosis. This paper primarily introduces adversarial-based deep transfer learning (ADTL) models, which are fundamentally based on Generative Adversarial Network (GAN). We provide a detailed discussion of the main applications of ADTL and its recent developments in intelligent fault diagnosis, along with some future challenges and prospects.

Simulation data-driven adaptive frequency filtering focal network for rolling bearing fault diagnosis

The scarcity of well-labeled samples severely limits the application of deep learning-based fault diagnosis methods. To address this issue, this paper proposes a novel domain-adaptive intelligent diagnostic method, termed a simulation data-driven adaptive frequency filtering focal network, which transfers knowledge from dynamic simulation data to unlabeled measured data. First, a dynamic simulation model is established to simulate the coupled behavior of rolling bearings and generate substantial labeled simulation data. Subsequently, an adaptive frequency filter is introduced to suppress random interference components in the frequency domain, complementing conventional time-domain feature extraction methods, thereby reducing the distribution discrepancy between simulation and measured data. Finally, a domain-aware weighted focusing strategy is proposed to adjust sample weights based on predicted domain labels, helping the model focus on difficult samples with significant distribution discrepancies and reduce misclassification. Experimental results demonstrate that the proposed method effectively integrates simulation and measured data, achieving high-accuracy fault diagnosis of rolling bearings in unlabeled measured data. The proposed approach offers a promising fault diagnosis solution in scenarios where obtaining labeled samples is challenging or impractical.

A Domain Generation Diagnosis Framework for Unseen Conditions Based on Adaptive Feature Fusion and Augmentation

Emerging deep learning-based fault diagnosis methods have advanced in the current industrial scenarios of various working conditions. However, the prerequisite of obtaining target data in advance limits the application of these models to practical engineering scenarios. To address the challenge of fault diagnosis under unseen working conditions, a domain generation framework for unseen conditions fault diagnosis is proposed, which consists of an Adaptive Feature Fusion Domain Generation Network (AFFN) and a Mix-up Augmentation Method (MAM) for both the data and domain spaces. AFFN is utilized to fuse domain-invariant and domain-specific representations to improve the model’s generalization performance. MAM enhances the model’s exploration ability for unseen domain boundaries. The diagnostic framework with AFFN and MAM can effectively learn more discriminative features from multiple source domains to perform different generalization tasks for unseen working loads and machines. The feasibility of the proposed unseen conditions diagnostic framework is validated on the SDUST and PU datasets and achieved peak diagnostic accuracies of 94.15% and 93.27%, respectively.

Prior knowledge embedding convolutional autoencoder: A single-source domain generalized fault diagnosis framework under small samples

The proposed transfer learning-based fault diagnosis models have achieved good results in multi-source domain generalization (MDG) tasks. However, research on single-source domain generalization (SDG) is relatively scarce, and the limited availability of small training samples is seldom considered. Therefore, to address the insufficient feature extraction capability and poor generalization performance of existing models on single-source domain small sample data, a novel single-source domain generalization fault diagnosis (SDGFD) framework, the prior knowledge embedded convolutional autoencoder (PKECA), is proposed. During the training phase, first, single-source domain data are used to construct prior features based on the time domain, frequency domain, and time-frequency domain. Second, a prior knowledge embedding structure based on the convolutional autoencoder is built, which compresses the prior knowledge and original vibration data into a high-dimensional space of consistent dimensions, embedding the prior knowledge into the features corresponding to the vibration data using a mean squared error loss function. Subsequently, the proposed centroid-based self-supervised learning (CBSSL) strategy further constrains high-dimensional features, improving the generalization ability. The designed sparse regularized activation (SRA) function significantly enhances the regularization effect on features. During the testing phase, it is only necessary to input the data from the unknown domain to identify the fault types. The experimental results show that the proposed method achieves superior performance in fault diagnosis tasks involving cross-speed, time-varying speed, and small sample data in SDGFD, demonstrating that PKECA has strong generalizability. The code can be found here: https://github.com/John-520/PKECA. © 2024 Elsevier Science. All rights reserved

Federated learning for decentralized fault diagnosis of a sucker-rod pumping system with class imbalance data

The development of modern oilfields has entered the middle and late stages, transforming towards digitalization and intelligence. However, the distribution of the sucker-rod pumping systems is decentralized, and the working condition information is skew-distributed. This situation poses a significant challenge to existing centralized fault diagnosis mechanisms. To address the existing practical challenge in the oilfield, a federated learning-based fault diagnosis framework for class imbalance in decentralized sucker-rod pumping systems (FL-CI) is proposed. This framework incorporates a parameter anonymization-ratio upload mechanism to mitigate the risk of gradient tracking. Then, a monitoring mechanism is leveraged to reversely infer global class-imbalance data using trained parameters uploaded by the clients. In addition, a ratio loss function is designed to calibrate the influence of class imbalance on the global system. After conducting comparative analysis, ablation analysis, and sensitivity analysis on a rod-pumping unit dataset (RPUD), as well as comparative and ablation analyses on the Case Western Reserve University bearing dataset (CWRU), the experimental results demonstrate that the FL-CI framework achieves superior diagnostic performance on the RPUD, with eight out of twelve evaluation metrics significantly outperforming seven state-of-the-art methods. A similar trend is observed on the CWRU, further validating the effectiveness and generalizability of the FL-CI.

A General and Efficient Fault Diagnosis Approach Using Hyperdimensional Computing Enhanced by Local Feature Extraction

Abstract Ensuring the safe and reliable operation of industrial equipment is of great importance, and accordingly, machine fault diagnosis has attracted considerable attention. Recently, various deep learning-based fault diagnosis models have been widely developed due to the powerful ability of feature learning and nonlinear modeling of deep learning models, and these methods heavily rely on large amounts of data and high memory and resources. However, it is difficult to obtain enough data in practical applications, thus leading to the unsatisfactory performance of deep learning-based models and high cost. In this paper, we propose a new and efficient fault diagnosis approach based on Hyperdimensional computing (HDC), which can learn high-dimensional, random, and holographic distributed representations of input features. It has the advantages of single training, low training cost, and does not require a large amount of data. Therefore, we apply HDC to the field of few-shot fault diagnosis for the first time, and propose a simple and efficient approach for few-shot fault diagnosis based on hyperdimensional computing with local feature extraction(L-HDC). Instead of using original data, the model firstly extracts the local features of original data by hand-designed features, and then carries out HDC-encoding, training, and test classification. Through the comparison of various models and the analysis of various data, the effectiveness of our proposed model is verified. In the case of few-shot learning, the recognition accuracy is as high as 90\% and the training time of ordinary HDC model is 5-10 times that of L-HDC model, which is superior to other models.

Deep learning-based fault diagnosis of servo motor bearing using the attention-guided feature aggregation network

This paper introduces a novel approach to fault detection in the servo motor bearings of industrial robots within the context of Industry 4.0 prognostics and health management. The proposed solution leverages the innovative feature aggregation network for robotic fault detection in the application of smart factory. Overcoming challenges associated with traditional techniques that include handcrafted features, transfer learning, and deep learning models, the proposed approach offers a hierarchical information aggregation mechanism. The model is customized through hyperparameter tuning, resulting in a streamlined architecture with significantly fewer parameters. This parameter efficiency is notably distinct when compared to off-the-shelf transfer learning models that commonly feature extensive parameter counts in the range of hundreds of thousands or millions. The proposed model subjected to rigorous validation across diverse experimental scenarios that affirm its adaptability and robust performance. The model showcases accuracy in fault detection under both simple and welding motion scenarios, while its generalization capabilities are demonstrated as it successfully predicts health states in welding motion, showcasing versatility and reliability across various operational scenarios.

Multi-source subdomain negative transfer suppression and multiple pseudo-labels guidance alignment: A method for fault diagnosis under cross-working conditions

Extensive researches have been conducted on transfer learning based fault diagnosis. However, negative information transfer may arise due to significant differences in the subdomain distribution across multiple source domains (MSDs). Most existing methods focus solely on the impact of subdomains from a single source domain (SSD) on the target domain (TD). Therefore, this paper proposed a novel multi-stage alignment multi-source subdomain adaptation (MAMSA) method. The global feature extractor is designed to extract domain-invariant features. Three domain-specific feature extractors capture high-level fault features from different domains with a customized adaptation strategy, which combines adversarial learning and distribution alignment based on multiple pseudo-label-guided local maximum mean discrepancy (MP-LMMD) to learn subdomain-invariant features. MP-LMMD utilizes pseudo-labels generated from all classifiers in the TD to guide the alignment of subdomains, suppressing negative transfer from the source domains (SDs). The experimental results indicate that the MAMSA method has excellent capabilities to suppress negative transfer, and the diagnostic performance can be greatly promoted with MAMSA under cross-working conditions.

Granularity knowledge-sharing supervised contrastive learning framework for long-tailed fault diagnosis of rotating machinery

The long-tailed distribution of monitoring data poses challenges for deep learning-based fault diagnosis (FD). Recent efforts utilizing supervised contrastive learning (SCL) and reweighted loss have made progress, but have overlooked two key issues: 1) prevailing random undersampling introduces sample influence bias and suboptimal model learning; and 2) focusing only on improving the FD average accuracy compromises fundamental fault judgement (FJ), heightening missed-detective and false-alarm risks unsuitable for real-world deployment. To fill these research gaps, this paper proposes a granularity knowledge-sharing SCL (GKSSCL) framework for long-tailed FD, encompassing GKS supervised contrasting and GKS classification stages. In the former, normal data are clustered into multiple fine-grained subclasses that are similar in size to the fault categories for balanced contrasting. Moreover, a mixed-granularity contrastive loss facilitates knowledge sharing across granularities. In the latter, FJ and FD tasks were concurrently trained through a knowledge graph-based adaptive sharing strategy. Experiments on two fault datasets showed that the GKSSCL can effectively harness all normal data, eliminate sample influence bias, and enhance FD precision without sacrificing FJ reliability.

Machine Learning-based Fault Prediction and Diagnosis of Brushless Motors

Machine Learning-based Fault Prediction and Diagnosis of Brushless Motors

Few-Shot Learning-Based Fault Diagnosis Using Prototypical Contrastive-Based Domain Adaptation Under Variable Working Conditions

Few-Shot Learning-Based Fault Diagnosis Using Prototypical Contrastive-Based Domain Adaptation Under Variable Working Conditions

An interpretable multiscale lifting wavelet contrast network for planetary gearbox fault diagnosis with small samples

Previous deep learning-based fault diagnosis methods for planetary gearbox require numerous training samples and lack the necessary interpretability. Aiming at the problems of insufficient interpretability of deep models and the absence of feature mining capability with small samples, this study presents an interpretable multiscale lifting wavelet contrast network for planetary gearbox fault diagnosis. First, an interpretable multiscale lifting wavelet network is designed to achieve comprehensive and credible features mining from fault signals. Secondly, an interactive channel attention mechanism is constructed to choose feature maps with different frequency components. It can further confirm the interpretability of the lifting wavelet layer while improving the accuracy of the model. Finally, a time-frequency contrast loss is designed to simultaneously optimizing the distribution of time-frequency domain features. The effectiveness and interpretability of the model are analyzed through various visualization approaches. Experimental results on two planetary gearbox datasets indicate that our method is an interpretable and effective fault recognition method with small samples, and it holds a promising future for engineering applications.

Fault Diagnosis of Centrifugal fan Bearings Based on I-CNN and JMMD in the Context of Sample Imbalance

The fault diagnosis is an effective technical means to improve the reliability of centrifugal fan bearings. The serious imbalance of data is one of the important issues facing bearing fault diagnosis.In this paper, a transfer learning-based fault diagnosis method for Centrifugal fan bearings is proposed, utilizing the improved CNN (I-CNN) and Joint Maximum Mean Discrepancy (JMMD) algorithms. The raw vibration signals of the bearings are enhanced through fast Fourier transform for feature representation. The signals are then processed by parallel multi-scale CNNs with an embedded Squeeze-and-Excitation (SE) attention to focus on key features. Furthermore, the JMMD is introduced as a metric for quantifying the disparity between the source and target domains, thereby mitigating domain shift. In the loss function, weight factors and scaling factors are introduced to increase attention on minority samples and easily confused samples within the imbalanced dataset. The proposed method is validated on the Centrifugal fan bearing dataset from Jiangnan University and the CWRU dataset.

A coarse and fine-grained deep multi view subspace clustering method for unsupervised fault diagnosis of rolling bearings

Abstract To address the issue of insufficient characterization of fault features in inherent vibration data that affects the performance of unsupervised learning-based fault diagnosis, a coarse and fine-grained deep multi view subspace clustering method (CFG-DMVSC) for unsupervised fault diagnosis of rolling bearings is proposed. The proposed method designs a convolutional autoencoder network based on the Gramian angular field transformation for multi-signal analysis domains. A multi-view coarse-grained self-expressive method based on information entropy is designed to handle differences in information across different views. Furthermore, a fine-grained common and independent information separation loss function based on mutual information is proposed to ensure compactness among multiple views. Both the Case Western Reserve University rolling bearing dataset and privately built bearing fault test bench data demonstrate that, compared to existing methods, the proposed method can perform coarse and fine-grained division in multi-view subspaces, achieving better clustering diagnosis performance on the extracted common information among views.