Gear Fault Diagnosis Research Articles

All time-scale decomposition method and its application in gear fault diagnosis

Adaptive signal decomposition methods, especially without parameters, have become a popular way of diagnosing mechanical faults due to their capability to process mechanical vibration signals adaptively. Empirical mode decomposition (EMD), local mean decomposition (LMD), and local characteristic-scale decomposition (LCD) are typical parameterless adaptive signal decomposition methods currently applied to mechanical fault diagnosis. All of these methods use extreme points to construct baselines, and the mono-component signals are decomposed from an original signal by multiple sift. However, since these methods define time-scale parameters only through extreme points, they are prone to lose the local feature information of an original signal and lead to mode mixing. Aiming at the above problems, the time-scale parameters is defined by using extreme points and zero crossing points simultaneously in this paper. Therefore, we propose a new adaptive signal decomposition method called all time-scale decomposition (ATD). A complex signal can be adaptively decomposed into multiple independent all time-scale components by the ATD method. The baselines of ATD are constructed jointly by extreme points and zero crossing points, so ATD can extract more local feature information of a signal to suppress the mode mixing. First, the principle of ATD is proposed and the method of determining zero crossing points is introduced in this paper. Then, an empirical formula for compensation factor used to determine zero crossing points is deduced. Finally, ATD is verified by the simulation signals and gear signals, respectively. The results indicate that ATD has stronger mode mixing suppression capability and decomposition performance than EMD, LMD, and LCD, and it can be effectively used for gear fault diagnosis.

Fault diagnosis of wind turbine gears based on OCSSA-VMD and WOA-CNN-BiLSTM

Abstract The accuracy of wind turbine gearbox fault diagnosis will be compromised if the fault feature data is not adequately extracted during operation. To enhance fault identification efficiency and mitigate human interference in parameter setting, this paper introduces an optimized mode decomposition algorithm OCSSA-VMD, derived from variational mode decomposition (VMD) and further optimized by osprey-Cauchy-sparrow search algorithm (OCSSA). This algorithm offers two key advantages: (1) automatic optimization of parameters such as the number of modes k and penalty factor α; (2) reduction of feature dimensionality through mean impact value (MIV) algorithm based on minimum envelope entropy principle, resulting in a multi-fault feature vector set from 13 time-domain features in the intrinsic mode function (IMF) optimal component of wind turbine gearbox vibration data. Additionally, a fault diagnosis model WOA-CNN-BiLSTM is proposed based on whale optimization algorithm (WOA) and convolutional neural network-bidirectional long-short-term-memory (CNN-BiLSTM), which demonstrates improved fault classification accuracy to 98.3333% and diagnosis accuracy to 98.3853% under conditions of insufficient data when compared with other models.

Multidisciplinary Framework for Vibration-Based Gear Fault Diagnosis Using Experiments, Modeling, and Machine Learning

Vibration-based gear diagnosis is crucial for ensuring the reliability of rotating machinery, making the monitoring of gear health essential for preventing costly downtime and optimizing performance. This study proposes a multidisciplinary framework to enhance gear diagnosis, that aligns with the new era of digital twins by integrating experiments, dynamic modeling, physical preprocessing, and machine learning. Within this framework, we focus on three core procedures: domain adaptation to reduce discrepancies between measured data and synthetic data generated by dynamic models; physical preprocessing, grounded in in-depth investigations dictating signal processing and feature engineering techniques; and learning algorithms, encompassing the process of training AI-based models. We demonstrate this framework through a comprehensive case study of localized tooth fault diagnosis, using controlled-degradation tests and realistic simulations. First, we detect faults using unsupervised learning algorithms; then, we use zero-shot-learning for classifying between localized and distributed faults; finally, we adopt a one-shot-learning strategy for severity estimation. Above all, this hybrid framework bridges the gaps between physical-based and AI-based approaches by combining physical knowledge and advanced algorithmics with machine learning. This contributes to the PHM field by offering valuable insights into integrating different aspects of research, thereby enhancing performance in gear diagnosis tasks.

An unsupervised transfer learning gear fault diagnosis method based on parameter-optimized VMD and residual attention networks

An unsupervised transfer learning gear fault diagnosis method based on parameter-optimized VMD and residual attention networks

A fault diagnosis method for bearings and gears in rotating machinery based on data fusion and transfer learning

Abstract Rotating machinery is a crucial component of industrial equipment, and the fault diagnosis of bearings and gears, as vital elements of rotating machinery, is essential since they often fail under harsh working conditions, leading to significant property losses and serious personal safety problems. However, fault data for gears and bearings are often sparse in actual condition, and it is a challenge to ensure the reliability and stability of fault diagnosis results by extracting the features of a single data. To solve the above problems, this paper proposes a fault diagnosis method that combines Transfer Learning and data fusion techniques. Firstly, in this method, two kinds of fault signals are transformed into Gramian Angular Difference Fields and Recurrence Plot. Next, a U-shaped feature fusion dual discriminator generative adversarial network is used to fuse two-dimensional images from multiple sensor data. Its feature fusion module deeply integrates the features of the two images, thereby solving the impact of single data on the reliability and stability of fault diagnosis. Moreover, open-source datasets are used for Transfer Learning training to tackle the small sample problem. Finally, a decision-level information fusion classifier, the Dual-Branch Dempster-Shafer Classifier (DB-DSC), classifies the fused images. This classifier incorporates an improved soft threshold function and D-S evidence theory to achieve adaptive gradient changes and improve the robustness and accuracy of classification results. The experimental results show the effectiveness and stability of the proposed method, and the generated images get high score in several metrics. The average classification accuracy of the classification network reaches 93% and 92.5% on the two datasets, Therefore, the proposed method exhibits strong fault diagnosis capabilities under the small sample conditions of bearings and gears.

Research on Typical Fault Diagnosis of Gear based on GHM Multiwavelet Transform and MCKD Algorithm

Background: Gear is the key part of the transmission part of mechanical equipment, which has a high failure rate under complex working conditions or long-term operation, directly affecting the reliability of mechanical equipment. To realize online monitoring of gears and predict gear faults, whether the diagnosis of typical gear faults is accurate and timely is a key link, so it is necessary to study the diagnosis of typical gear faults. Objective: This paper aims to propose a method that can effectively diagnose typical faults of gears and is easy to program, which is beneficial to realize online monitoring of rotating machinery equipment faults. Method: A typical gear fault diagnosis method based on GHM multiwavelet transform and maximum correlation kurtosis deconvolution algorithm (MCKD) is proposed in this paper. The measured vibration signal is decomposed into components with multiple scale spaces and frequency bands through the GHM multiwavelet transform. Then, the MCKD algorithm is run to suppress the noise of GHM multiwavelet transform coefficients, and the periodic components containing fault information are retained and reconstructed. The typical faults such as gear pitting and broken teeth are tested. Results: GHM multiwavelet transform combined with MCKD algorithm can accurately locate the frequency band containing fault information from the vibration signal with multiple modulations, double sideband, and multi-frequency interference and effectively extract the fault frequency and its frequency multiplication. The root means a square error of the extracted fault frequency is 0.08. This method has a good noise suppression effect and high accuracy in extracting fault frequency. Conclusion: The method proposed in this paper is helpful in realizing automatic programming of gear fault diagnosis and is of great significance in realizing real-time monitoring and online diagnosis of rotating machinery equipment faults. More related patents will appear in the future.

Case study: Condition-based maintenance using cyclostationary analysis and numerical modeling with innovative indicators

Early detection and prompt intervention by maintenance engineers to mitigate the impact of breakdowns while enhancing overall operational efficiency remain critical challenges. This study proposes an innovative approach aiming at improving the diagnosis of gear faults. The objective is to assess the sensitivity and performance of traditional indicators in comparison to cyclostationarity, examining their impact on noise levels and vibrational signatures. The initial phase involves simulating gear signals under various conditions such as amplitude, rotation frequency, and meshing frequency, providing the foundation for a thorough analysis of indicator sensitivity and performance. In the second phase, both scalar and cyclostationary indicators were calculated. First, these indicators were compared against simulated signals, and second, their sensitivity and roughness were evaluated using signals measured on the bearings of 101 BJR reducers. This approach revealed that cyclostationary indicators are more sensitive than scalar indicators, suggesting an opportunity to improve the prediction of signal roughness throughout the production process. By introducing new possibilities to enhance the reliability of vibrational measurements, this method contributes to advancing the diagnosis of gear faults.

Current Status of Research on Fault Diagnosis Using Machine Learning for Gear Transmission Systems

Gear transmission system fault diagnosis is crucial for the reliability and safety of industrial machinery. The combination of mathematical signal processing methods with deep learning technology has become a research hotspot in fault diagnosis. Firstly, the development and status of gear transmission system fault diagnosis are outlined in detail. Secondly, the relevant research results on gear transmission system fault diagnosis are summarized from the perspectives of time-domain, frequency domain, and time-frequency-domain analysis. Thirdly, the relevant research progress in shallow learning and deep learning in the field of fault diagnosis is explained. Finally, future research directions for gear transmission system fault diagnosis are summarized and anticipated in terms of the sparsity of signal analysis results, separation of adjacent feature components, extraction of weak signals, identification of composite faults, multi-factor combinations in fault diagnosis, and multi-source data fusion technology.

Enhancing bearing and gear fault diagnosis: A VMD-PSO approach with multisensory signal integration

In the domain of signal analysis for machinery health monitoring and fault diagnosis, this paper introduces a comprehensive methodology that integrates Variational Mode Decomposition (VMD), Particle Swarm Optimization (PSO), and advanced machine learning techniques. The primary objective of this framework is to establish a robust and precise approach for signal decomposition, determining the optimal number of Intrinsic Mode Functions (IMF), and calculating key indicators, including L2/L1, Hoyer Index, and Geometric Mean Improved Gini Index (GMIGI). The methodology initiates with VMD-based signal decomposition, followed by the utilization of PSO to identify the most appropriate number of IMFs for accurate feature extraction. Subsequently, each IMF’s performance is assessed by evaluating its correlation with the input signal, and the IMF with the highest Pearson coefficient is selected as the primary feature for diagnostic purposes. To ensure the robustness and comparability of these indicators, a standardization process is implemented. The standardized indicators are then employed for machinery fault diagnosis, utilizing a diverse set of machine learning algorithms such as support vector machines and discriminant analysis. The proposed methodology undergoes rigorous validation using vibration, acoustic, and current signals, providing a versatile solution for the condition monitoring and diagnosis of mechanical systems. For model validation, we utilize four datasets comprising two vibrational, one acoustic, and one electrical dataset. The experimental results affirm the effectiveness of our approach in accurately detecting and diagnosing faults, thereby contributing to the reliability and maintenance efficiency of industrial machinery.

A gear fault diagnosis method based on reactive power and semi-supervised learning

Abstract In gearbox gear fault diagnosis based on motor current signals, the gear fault characteristic frequency component is often overshadowed by the fundamental frequency component of the current. In addition, the complex working conditions during actual production and use make it difficult to collect gear operation monitoring data containing labeled feature information. To address the above problems, a semi-supervised learning method based on reactive power signals is proposed for gear fault diagnosis of gearboxes. First, the method utilizes the Hilbert transform to process the current signal of the drive motor in the mechanical system, from which the reactive power is constructed. Then, the reactive power signal is analyzed by spectral analysis as a basis for gear fault diagnosis. Subsequently, the GAF-CNN-MTDL(Gramian angular field—convolutional neural network-mean teacher deep learning) fault diagnosis model is proposed to convert the reactive power signal into a two-dimensional image by using the GAF, and the semi-supervised training method of the average teacher is applied to input the fault dataset into the gear fault diagnosis model which is based on the CNN as the main backbone after the fault dataset has been divided into the labeled and the unlabeled dataset in accordance with a certain ratio. Finally, the gear fault dataset is used for method validation. The experimental outcomes demonstrate the method’s proficiency in effectively emphasizing the fault feature information pertaining to the gear part, and the introduced GAF-CNN-MTDL fault diagnosis model enables the utilization of a minimal number of labeled samples to achieve highly accurate gear fault diagnosis.

Nonlinear group constrained mode decomposition and its application in gear fault diagnosis

Abstract Due to the problem of modal confusion and extreme sensitivity to strong noise in signal decomposition using group-sparse mode decomposition, this paper proposes a new signal decomposition method, named nonlinear group constrained mode decomposition (NGCMD). First, NGCMD is based on modified discrete cosine transform, which enables the function to have ‘energy compression’ characteristics, resulting in a more approximate model of complex signals and avoiding the introduction of artifacts in frequency domain analysis. Meanwhile, the objective function adopts norm weighting as the penalty term, making the algorithm not only sparsity but also fast convergence. In addition, to maintain the adaptability of the algorithm, when the algorithm converges, the filter bank is decomposed into a set of non-overlapping filters, where each filter contains an adjacent non-zero term in the filter bank, and a series of sparse group constrained components with complete time–frequency distribution are obtained. Through the simulation signal and gear fault signal, the results show that the NGCMD method has obvious advantages in suppressing modal confusion and robustness, and can effectively diagnose gear fault.

Gear fault diagnosis based on small channel convolutional neural network under multiscale fusion attention mechanism

AbstractDue to the insufficient feature learning ability and the bloated network structure, the gear fault diagnosis methods based on traditional deep neural networks always suffer from poor diagnosis accuracy and low diagnosis efficiency. Therefore, a small channel convolutional neural network under the multiscale fusion attention mechanism (MSFAM‐SCCNN) is proposed in this paper. First, a small channel convolutional neural network (SCCNN) model is constructed based on the framework of the traditional AlexNet model in order to lightweight the network structure and improve the learning efficiency. Then, a novel multiscale fusion attention mechanism (MSFAM) is embedded into the SCCNN model, which utilizes multiscale striped convolutional windows to extract key features from three dimensions, including temporal, spatial, and channel‐wise, resulting in more precise feature mining. Finally, the performance of the MSFAM‐ SCCNN model is verified using the vibration data of tooth‐broken gears obtained by a self‐designed experimental bench of an ammunition supply and delivery system.

Fault Diagnosis of Planetary Gear Train Crack Based on DC-DRSN

To solve the problem that the existing planetary gear train fault diagnosis methods have, namely their low diagnostic accuracy under low signal-to-noise ratio (SNR), a fault diagnosis method based on a double channel-deep residual shrinkage network (DC-DRSN) is proposed. The short-time Fourier transform (STFT) is used to convert the original vibration signal into a two-dimensional time-frequency graph, which effectively enhances the ability to express information. A DC-DRSN model is constructed, and the optimal number of residual shrinkage modules is determined by combining the diagnostic characteristics with different noises, which effectively improves the accuracy and anti-noise ability of fault diagnosis. The results of bearing and planetary gear train crack fault diagnosis show that the diagnosis method based on DC-DRSN has higher diagnostic accuracy while realizing fault diagnosis, which is better than other deep learning diagnosis methods. At the same time, the method can adapt to fault diagnosis in different noise environments, and has good expression ability and generalization ability.

A novel model-based Cauchy-Schwarz divergence condition indicator for gears monitoring during fluctuating speed conditions

Gear monitoring and fault diagnosis are vital for preventing accidents and minimizing economic losses in transportation and industrial systems. Traditional methods use vibration sensors and a two-stage analysis approach: preprocessing data to remove noise and extract relevant components, and generating a condition indicator to detect behavioral anomalies in the gears over time.Time synchronous averaging is a notable tool for monitoring gears at constant speeds. Such a tool filters sensor signals and extracts rotation-related components by using statistical measurements as condition indicators. However, it has limitations in scenarios with time-varying sampling rates and fluctuating speeds, where statistical measures may not fully capture changes in system parameters.This article proposes a novel methodology for monitoring gears in multivariate rotordynamical systems under fluctuating speed conditions. The method integrates time synchronous averaging, system identification algorithms, and statistical tools. It generates a time-synchronous average signal considering speed fluctuations, computes a state–space model of gear behavior in healthy states, extracts residual data from a data-driven model, and generates a condition indicator based on the Cauchy–Schwarz divergence.The proposed methodology was evaluated using experimental data from three rotor dynamical setups under different operational conditions. Validation showed its effectiveness, especially under high-load conditions with significant speed fluctuations.

A data-physic driven method for gear fault diagnosis using PINN and pseudo-dynamic features

A data-physic driven method for gear fault diagnosis using PINN and pseudo-dynamic features

Gear fault diagnosis based on bidimensional time-frequency information theoretic features and error-correcting output codes: A multi-class support vector machine

Gear fault diagnosis based on bidimensional time-frequency information theoretic features and error-correcting output codes: A multi-class support vector machine

Composite fault feature extraction for gears based on MCKD-EWT adaptive wavelet threshold noise reduction

For the strong noise gear fault vibration signal is relatively weak, and the transmission path is complex and variable, in the case of composite faults, the modulation of different fault characteristics of the frequency, coupling, resulting in the actual acquisition of the fault characteristics are difficult to extract and separate. Aiming at fault feature extraction and separation, an adaptive threshold denoising fault detection method based on Maximum correlated kurtosis deconvolution (MCKD) and Empirical wavelet transform (EWT) is proposed. Firstly, envelope entropy and information entropy are used as fitness functions, and the parameters of the MCKD algorithm are optimized by the improved particle swarm algorithm, then the empirical wavelet decomposition is carried out on the signals, and finally adaptive wavelet threshold denoising is carried out on the decomposed Intrinsic mode functions (IMFs) components. The results of experimental data analysis show that compared with the feature extraction methods such as spatial scale threshold EWT-MCKD and Complete Ensemble Empirical Mode Decomposition (CEEMDAN)-MCKD, the proposed method is more suitable for the diagnosis of gear composite faults in a strong background noise environment, the noise interference is effectively suppressed, and the extraction effect of gear composite fault features is more obvious.

Cut-off point (COP) for early gear fault diagnosis via Meshing impact modulation (MIM) analysis

Comparing the number and magnitude differences and variations of the fault-induced features in frequency domain, such as the sidebands of meshing frequency in the raw spectrum, or the rotation frequency of faulty gear in the envelope spectrum between healthy and faulty conditions are the underlying principle of most of classical fault diagnostic methods for gearbox. However, several inevitable factors, for instance, circumferential stiffness anisotropy of gear shaft (including coupling mechanisms like keyways) and gear meshing backlash, can also lead to similar features, which may challenge these classical methods, especially when the fault is in its early stage. In specific, the initial early gear fault may neither lead to the significant increment of sideband or rotation frequency nor stimulate the high amplitude resonance, more than the inevitable factors mentioned above. As such, a meshing impact modulation frequency band is constructed to create a unique feature which does not exist in healthy condition. Considering both the impact nature and the modulation mechanism of gear meshing process, an MIM (Meshing Impact Modulation) index is designed, with which can effectively identify the optimal meshing impact modulation response frequency band with high stability and robustness across entire lifespan of gears. Within the demodulation spectrum of this band, the presence or absence of distinct gear rotation frequency spectral lines and their mirage as sidebands around the meshing frequency can serve as the effective and robust indicators to discern whether the gear is in a healthy "before cut-off point (pre-COP) state " or in an early faulty "after cut-off point (post-COP) state ". The effectiveness and the superiority of the proposed algorithm is testified by the life-test data from a test rig using industrial gearbox.

Wavelet-driven differentiable architecture search for planetary gear fault diagnosis

With the advancement of artificial intelligence and the accumulation of industrial big data, intelligent diagnosis methods based on deep learning have become the mainstream for diagnosing mechanical faults in manufacturing systems. Despite this, developing high-performance neural network models for specific tasks necessitates a substantial amount of expert knowledge. This consumes a considerable amount of time in the trial-and-error process, thereby constraining the progress in neural network development. Neural architecture search (NAS) offers a solution to address this problem. However, the feature extraction operators in the existing NAS search space are primarily imported from other fields, leading to domain bias and a lack of interpretability. To address these challenges, we present a NAS method driven by wavelets. To be specific, the wavelet operators that can extract fault related features from vibration signals is added to the search space, and the gradient optimization strategy is utilized to search the optimal architecture from the hypernet. The effectiveness of the proposed method is validated through a dataset specific to planetary gears. Upon comparison with other models, it is evident that the proposed method exhibits superior performance across all added noise levels.

AdaBoost Ensemble Approach with Weak Classifiers for Gear Fault Diagnosis and Prognosis in DC Motors

This study introduces a novel predictive methodology for diagnosing and predicting gear problems in DC motors. Leveraging AdaBoost with weak classifiers and regressors, the diagnostic aspect categorizes the machine’s current operational state by analyzing time–frequency features extracted from motor current signals. AdaBoost classifiers are employed as weak learners to effectively identify fault severity conditions. Meanwhile, the prognostic aspect utilizes AdaBoost regressors, also acting as weak learners trained on the same features, to predict the machine’s future state and estimate its remaining useful life. A key contribution of this approach is its ability to address the challenge of limited historical data for electrical equipment by optimizing AdaBoost parameters with minimal data. Experimental validation is conducted using a dedicated setup to collect comprehensive data. Through illustrative examples using experimental data, the efficacy of this method in identifying malfunctions and precisely forecasting the remaining lifespan of DC motors is demonstrated.