Gear Fault Diagnosis Research Articles

A novel method for helical gear modeling with an experimental validation

Dynamic models are important for developing gear diagnostics methods since they allow physical phenomena occurring during operation to be studied in a relatively simple environment. The main challenge in gear modeling is the calculation of the time-variant gear mesh stiffness, and this challenge is even greater in helical gears. The mechanism of helical gears is more complex than in spur gears; the helix angle both adds an axial component to the contact force and also makes the contact line three-dimensional. This study suggests a novel dynamic model for helical gear vibrations that combines an existing validated dynamic model for spur gears with a unique extension for helical gears. The extension is based on a common method called “multi-slice”, according to which the helical tooth width is divided into infinitesimal slices, and each slice is treated as spur tooth. The suggested model introduces a novel implementation of the multi-slice method that overcomes the aforementioned challenges with only few parameters and calculations, depends on the tooth geometry. Furthermore, for the first time in helical gear modeling, the manufacturing profile errors are integrated to the model to generate scatter in the data that can better reflect the reality. The model is validated experimentally and for two different test-rigs by a qualitative comparison of the RMS of the vibration signal. The simulations and the measured data show similar behavior at different ranges of rotational speed and applied load, emphasizing the potential inherent in the model for future work on gear fault diagnosis.

A recursive multi-head self-attention learning for acoustic-based gear fault diagnosis in real-industrial noise condition

Acoustic-based diagnosis (ABD) is a promising method for rotating machinery fault detection in real-industrial fields due to its advantage of non-contact measurement by air-couple. However, most ABD methods, which rely on single function module of denoising or anti-noise diagnosis, lacks sufficient ability against the strong and highly non-stationary background noise interference in practical industrial application. To deal with the obstacle, a novel two-stage ABD system, which performs denoising and anti-noise diagnosis in a step-by-step fashion, is proposed in this paper. In our method, the denoising sub-system can adaptively track and suppress real-industrial background noise from collected multi-channel signal by our designed recursive multi-head self-attention (RMHSA) mechanism, which take into account the global and long-range information of signal to recursively exclude noise component at each position in signal sequence for overcoming the limitation of traditional method in balancing denoising performance and computational complexity and discontinuity in diagnosis procedure. Subsequently, a novel anti-noise method is explored to automatically track the residual noise components after initial suppression and estimating the noise interference probability within time-frequency (T-F) unit of each signal sample by our designed spatial multi-head self-attention (SMHSA) mechanism in recursive manner. Simultaneously, the anti-noise protocol interacts with diagnosis model based on estimated probability in anti-noise diagnosis sub-system to further improve the diagnosis performance of ABD system under the interference of background noise. Experiment result in both real-industrial noise condition and simulated noise conditions with extremely worse SNRs indicate that the proposed two-stage ABD system is effective in dealing with gear fault diagnosis task under noise condition.

Deep Domain Adaptation with Correlation Alignment and Supervised Contrastive Learning for Intelligent Fault Diagnosis in Bearings and Gears of Rotating Machinery

Deep domain adaptation techniques have recently been the subject of much research in machinery fault diagnosis. However, most of the work has been focused on domain alignment, aiming to learn cross-domain features by bridging the gap between source and target domains. Despite the success of these methods in achieving domain alignment, they often overlook the class discrepancy present in cross-domain scenarios. This can result in the misclassification of target domain samples that are located near cluster boundaries or far from their associated class centers. To tackle these challenges, a novel approach called deep domain adaptation with correlation alignment and supervised contrastive learning (DCASCL) is proposed, which synchronously realizes both domain distribution alignment and class distribution alignment. Specifically, the correlation alignment loss is used to enforce the model to generate transferable features, facilitating effective domain distribution alignment. Additionally, classifier discrepancy loss and supervised contrastive learning loss are integrated to carry out feature distribution alignment class-wisely. The supervised contrastive learning loss leverages class-specific information of source and target samples, which efficiently promotes the compactness of samples of the same class and the separation of samples from different classes. Moreover, our approach is extensively validated across three diverse datasets, demonstrating its effectiveness in diagnosing machinery faults across different domains.

Fault Diagnosis of Gear Based on Multichannel Feature Fusion and DropKey-Vision Transformer

Fault Diagnosis of Gear Based on Multichannel Feature Fusion and DropKey-Vision Transformer

A novel empirical random feature decomposition method and its application to gear fault diagnosis

Sparse random mode decomposition (SRMD) is an emerging signal decomposition method for analyzing time series data. However, SRMD is unable to adaptively select suitable clustering parameters, and its decomposition effect is easily affected by noise. To overcome the above deficiencies, a novel signal decomposition method named empirical random feature decomposition (ERFD) is proposed in this paper. Firstly, ERFD utilizes sparse random feature model to construct the random feature energy spectrum of input signal, and the extraction of signal components can be achieved via spectrum segmentation mechanism. Subsequently, an adaptive continuous envelope segmentation strategy guided by the mean harmonic intensity ratio (MHIR) index is designed for the adaptive frequency band division of random feature energy spectrum, which effectively attenuates the interference of noise extreme points and ensures the accurate separation of interested components. Finally, the random features corresponding to different frequency bands are reconstructed to obtain a series of mutually independent intrinsic random components (IRCs). Furthermore, the proposed ERFD method is applied to gear fault diagnosis, and its feasibility and effectiveness are fully verified through simulation and two experimental cases. The comparative analysis results indicate that ERFD has a better suppression effect on noise and can effectively extract weak fault features of gear.

Comparison of ML Algorithms and Neural Networks on Fault Diagnosis of a Worm Gear

Comparison of ML Algorithms and Neural Networks on Fault Diagnosis of a Worm Gear

Gear fault diagnosis based on complex network theory and error-correcting output codes: Multi class support vector machine

Gearbox failures have a detrimental effect on the machine performance that affects the production capacity and economic benefits of a manufacturing industry in an adverse manner. Early detection of faults in gears helps to prevent sudden machine failures. Previously, many signal processing and artificial intelligence techniques have been successfully applied to diagnose gear faults by analyzing the machine vibrations. These techniques have repeatedly emphasized on the need to extract quality features from gear vibration signals and using artificial intelligence techniques that can identify multiple types of faults simultaneously. As such, in this article, new features called as visibility graph features based on complex network theory have been proposed to extract fault information from vibration signals. In addition, a new artificial intelligence framework for multifault classification known as error-correcting output codes-multiclass support vector machine is proposed to detect gear faults. To the best of authors’ knowledge, these features along with the error-correcting output codes-artificial intelligence framework have been rarely utilized for fault diagnosis purpose. The proposed approach has been applied on experimental gear vibration data to validate its effectiveness. First, the vibration signals are transformed to graphs, and various graph properties are computed to serve as fault features. Then, the fault features are supplied to train the error-correcting output codes-multiclass support vector machine model and learn the fault patterns. Finally, the gear fault classification accuracies are determined for various gear test conditions and compared with those of existing methods. It is observed that the proposed approach provides an average improvement of 4.22%, 9.93% and 11.8% over the time-domain features, time–frequency domain features and voting-based multiclass support vector machine classifier, respectively, for different operating conditions. Furthermore, the suggested technique augments the classification accuracies by 10%, 7.5% and 6.7%, respectively, as compared with the deep-learning models, namely standard stacked sparse autoencoder, deep neural network and convolution neural network.

A gear fault diagnosis method based on variational mode decomposition and multi-scale discrete entropy

Aiming at monitoring of gearbox faults, a gear fault feature extraction method based on variational mode decomposition (VMD) and multi-scale discrete entropy (MDE) is proposed in this paper. Firstly, the gear fault signal is decomposed into a series of intrinsic modal function (IMF) by VMD with selected parameters; Secondly, the decomposed IMF are extracted by MDE feature extraction method to form a feature sample set; Finally, the least square support vector machine (LSSVM) is used to classify the data set after feature extraction. The experiment results show that the proposed method owns the higher fault diagnosis accuracy than the traditional multi-scale entropy methods.

A deep learning approach for health monitoring in rotating machineries using vibrations and thermal features

Gearbox failures can lead to substantial damage, significant financial losses due to maintenance downtimes, and, in some instances, fatalities. This study introduces an intelligent gear fault diagnosis system employing a convolutional neural network (CNN), utilizing vibration and thermal features extracted from healthy, chipped, and broken tooth gear health categories. The performance of the CNN is compared with conventional machine learning models, including Naïve Bayes (NB), random forest (RF), and support vector machine (SVM) classifiers. Experimental investigations highlight CNN’s remarkable performance. With vibration features, the CNN achieved 96.78% accuracy, surpassing SVM (84.89%), NB (81.56%), and RF (85.11%). The CNN attained 100% accuracy when utilizing thermal features, while SVM, NB, and RF achieved 91.11%, 88.89%, and 96.51% accuracies, respectively.

A novel joint denoising method for gear fault diagnosis with improved quaternion singular value decomposition

Singular value decomposition (SVD) has drawn increasing attention in recent years as an effective signal noise reduction method. However, it is inapplicable to multivariate signals with abundant fault information. Existing quaternion singular value decomposition (QSVD) can decompose multivariate signals simultaneously, but the methods based on QSVD dependent on the embedding dimension and reconstructed components order seriously, and they are not suitable to noisy signals. To solve the problem, a novel joint denoising method improved quaternion singular value decomposition (IQSVD) is proposed, which can determine two parameters adaptively. Firstly, to select the embedding dimension, the improved power spectral density (IPSD) is proposed with considering frequency features and time-domine traits together. And then the periodic intensity (PI) is used to determine the reconstructed components by searching the component composing the most abundant periodic part. Finally, combined with envelope spectrum analysis, the gear fault is diagnosed. The results of the gear simulated and experimental multivariate signals verify the effectiveness of proposed method.

Quaternion empirical Ramanujan Fourier decomposition and its application in gear fault diagnosis

In the field of gear health monitoring, the measured signals are traditionally derived from a single sensor or a single direction. However, with the increasing complexity and size of equipment structures nowadays, a single-channel signal often falls short in providing comprehensive status information. Furthermore, fault-induced vibrations are randomly indeterminate in both location and direction, which indicates the necessity of using multi-channel signals for effective gear fault diagnosis. As a single-channel method, Ramanujan Fourier transform (RFT) has been recognized for its effectiveness in extracting periodic components from signals. However, the RFT is limited to handling single-channel signals and lacks the ability to analyze nonstationary signals. To overcome these challenges, this paper proposes a novel multi-channel signal processing method called quaternion empirical Ramanujan Fourier decomposition (QERFD), which extends the RFT to the multi-channel field using the quaternion method. By fusing multi-channel signals while preserving the advantages of RFT, the proposed QERFD method enhances fault information extraction. In addition, QERFD shows high computational efficiency. Therefore, QERFD is of great practical value for the health monitoring of complex equipment in the present industrial context of big data. To substantiate the superiority of QERFD, simulations and experiments are carried out and compared against both single-channel and multi-channel methods.

MESMD and Its Application to Fault Diagnosis of Gear

MESMD and Its Application to Fault Diagnosis of Gear

Research on Instantaneous Angular Speed Signal Separation Method for Planetary Gear Fault Diagnosis

Research on Instantaneous Angular Speed Signal Separation Method for Planetary Gear Fault Diagnosis

A Lightweight Gear Fault Diagnosis Method Based on Attention Mechanism and Multilayer Fusion Network

A Lightweight Gear Fault Diagnosis Method Based on Attention Mechanism and Multilayer Fusion Network

Detecting Mixed Gear Faults Using Scalar and Cyclostationary Indicators in an Industrial Settin

In this paper, an innovative approach is presented to enhance gear fault diagnosis using the cyclostationarity method. The first part of this study focuses on simulating gear signals under various conditions, allowing exploration of signal characteristics in vibration measurements. Spectral analyses and statistical calculations are performed to extract both classical and cyclostationary indicators. In the second part, the cyclostationarity method is applied to signals recorded at the gearbox bearings, clearly revealing the presence of faults. The results from these experiments demonstrate that cyclostationarity indicators can be leveraged to improve the prediction of signal roughness during the production process. This approach thus opens up new possibilities to enhance the reliability of vibration measurements and refine gear fault diagnosis.

Time-Frequency Fusion Features-Based GSWOA-KELM Model for Gear Fault Diagnosis

To improve the accuracy of gear fault diagnosis and overcome the low diagnostic accuracy of the model caused by manual parameter selection, a combined diagnostic model based on time-frequency fusion features is combined with the improved global search whale optimization algorithm (GSWOA) to optimize the fault diagnosis capability of the kernel extreme learning machine (KELM). First, the time-domain and frequency-domain features of the gear fault state are extracted separately, and feature vectors are constructed through feature fusion, which overcomes the limitations of single features. Second, the GSWOA based on three strategies is used to optimize the regularization coefficient C and kernel function parameter γ of KELM, and a GSWOA-KELM fault diagnosis model is built to avoid the problem of low fault diagnosis accuracy caused by the manual selection of KELM parameters. Finally, the public dataset from Southeast University is taken to verify the performance of the proposed model by comparing it with KELM, SSA-KELM, and WOA-KELM models. The experimental results demonstrate that the improved time-frequency fusion features-based GSWOA-KELM model shows faster convergence speed and stronger global search ability. Compared with KELM, SSA-KELM, and WOA-KELM models, the performance of the proposed model has been improved by 11.33%, 8.67%, and 1.33%, respectively.

Fault diagnosis method for planetary gearbox based on intrinsic feature extraction and attention mechanism

In a high-noise environment and with a limited number of faulty samples, it becomes challenging to extract a sufficient amount of useful fault information, which makes gear fault diagnosis more difficult. To address these issues, this paper proposes a fault diagnosis method for planetary gearboxes based on intrinsic feature extraction and attention mechanism. The method utilizes the complementary ensemble empirical mode decomposition algorithm to perform modal decomposition on the fault vibration signal, obtaining a series of modal components. By comparing and selecting the modal components that contain a significant amount of fault features, they are then transformed into two-dimensional images with time–frequency properties using wavelet transform. Additionally, a neural network model based on attention mechanism and large-scale convolution is proposed. The preprocessed images are inputted into the network for feature extraction. During this process, the large-scale convolution with residual structure maximizes the retention of effective feature information, while the attention network further filters the features. Finally, the selected features are used for fault classification. The model is validated using the gear datasets from Southeast University and the University of Connecticut. A comparison is made with the Pro-MobileNetV3, channel attention and multiscale convolutional neural network, multiscale dynamic adaptive residual network, and CBAM-ResNeXt50 models. It is found that the accuracy reaches 100% before adding Gaussian noise and 99.68% after adding noise, which is significantly higher than that of other models.

Influence of idler misalignment fault on contact and dynamic characteristics of helical gear

Due to the advantages of stable helical gear transmission, it is widely used in industry, agriculture, and national defense. Shaft misalignment has a great influence on the vibration, bearing capacity, and life of gear transmission. Based on the idler wheel drive system as the research object, according to the principle of gear meshing, the helical gear time-varying contact wire length model was deduced. The dynamic model of the gear transmission system with faults is established considering the faults of the idler's center distance error and angle error, and the influence law of idler misalignment fault on the vibration of the idler transmission system is revealed. On this basis, the finite element model under angle error is established to analyze the contact stress of the tooth surface under different faults and reveal the contact stress distribution law of the tooth surface. The results show that the misalignment fault will change the length of the helical gear contact line. When the gear shaft misaligns, it will increase the axial force of the gear and increase the axial vibration of the gear. Moreover, the angle error will cause uneven load distribution on the tooth surface of the helical gear, which will reduce the contact area of the gear and increase the contact stress. The phenomenon of off-load of gear accelerates the wear of the gear tooth surface and shortens the life of the gear. The study of idler misalignment fault is of great significance for prolonging the life of gear and providing the basis for gear fault diagnosis.

A novel random spectral similar component decomposition method and its application to gear fault diagnosis

Sparse random mode decomposition (SRMD) is a decomposition approach established by combining sparse random feature model with clustering algorithm. It is not subject to the sampling process of signal and can mitigate mode mixing. However, the performance of SRMD is limited by its own hyperparameters, and it is prone to derive inaccurate clustering decomposition results when processing strong noise interference signal. To overcome these defects, a novel method called random spectral similar component decomposition (RSSCD) is proposed. In RSSCD, the time–frequency localized features produced by randomization and sparsification are adopted to represent the input signal. Subsequently, the initial signal components formed by sparse random features are taken as a whole, and the spectral similarity criterion is defined to adaptively form independent random components (RCs), thus improving the accuracy of decomposition. Furthermore, RSSCD is applied to gear fault diagnosis, and a fault significance measure (FSM) index is designed to guide the selection of parameter in sparse random feature representation, which ensures the fault information richness of the required RCs. Finally, the feasibility and effectiveness of RSSCD are fully validated by simulation signals and two experimental cases. The results indicate that RSSCD has excellent decomposition performance and fault feature extraction ability.

A novel metric-based model with the ability of zero-shot learning for intelligent fault diagnosis

Intelligent fault diagnosis plays an important role in maintaining the safe and reliable operation of rotating machinery. However, the data collected in real engineering scenarios may be severely insufficient, which presents challenges to the intelligent fault diagnosis methods. To address this problem, this paper introduces a metric-based meta learning approach for gear fault diagnosis under zero shot conditions. Firstly, a gear-rotor dynamics model is established to simulate the vibration signals under different fault conditions. And the signals are converted into energy maps through wavelet transformation to provide frequency domain fault features. Secondly, a deep convolutional network is employed as the feature extraction module to construct the prototype representations by calculating the average embedding within each fault class. Then, the distances between the actual signals collected from the gear test rig and the class prototypes are computed. Finally, the softmax is applied to convert these distances into probability distributions for outputting the predicted fault classes. Furthermore, label smoothing technology is introduced to mitigate the probability distribution differences between simulated signals and real signals. The experimental results demonstrate that the average diagnostic accuracy of the proposed model reaches 98.9%, which is better than other models.