Gear Fault Diagnosis Research Articles

Generative Adversarial Networks for Gearbox of Wind Turbine With Unbalanced Data Sets in Fault Diagnosis

Signal measurement and diagnosis of wind turbine gearbox are very important for equipment maintenance. Generative adversarial networks (GAN) are particularly outstanding in data generation due to its game mechanism. An improved gear fault diagnosis method based on GAN for unbalanced data sets is proposed in this paper. Firstly, the fault data is encoded by binary vectorization based on kurtosis perceptron. Secondly, the individuals that fit the fitness are selected by the deterministic selection and the macro factor code string is crossed by multiple points. Then, gaussian mutation is focused on searching for local fault points. Finally, the nonlinear decision boundary is established by the logistic regression auxiliary classifier. The effectiveness of the proposed method was verified by three groups of comparison experiments. Compared with the existing methods, the proposed method has a better ability for fault feature generation, classification, and diagnosis accuracy under unbalanced data sets.

Egram Based SVD Method for Gear Fault Diagnosis

The performance of singular value decomposition (SVD) mainly relies on selection of the effective singular values. However, most of the existing methods for selecting the effective singular values have some problems such as unsatisfactory performance of denoising. For selecting the effective singular values more accurately, a novel energy diagram (Egram) based SVD method is proposed. Novelties of the proposed method are divided into two parts. Firstly, the form of the effective singular values is determined by selection criteria of embedding dimension of Hankel matrix in SVD. The selection criteria are defined based on the special relationship between the form of the effective singular values and the embedded dimension of Hankel matrix for the amplitude modulation-frequency modulation (AM-FM) signal, and the special relationship is found that, when appropriate embedding dimension is selected, the AM-FM signal is mainly decomposed into two adjacent singular values with approximately equal numerical magnitude, which are called singular value pairs ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SVP</i> ). Secondly, the Egram is developed to adaptively locate effective <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">SVP</i> by estimating the energy of frequency band and to detect gear fault. The Egram-based SVD method is verified using the simulated signal and experiment signal that measured from the gearbox test bench, and analysis results show that the form and location of the effective singular values that contains gear fault can be determined. Comparisons with the difference spectrum, relative change rate of singular values, relative change rate of singular kurtosis and fast Kurtogram show that the Egram-based SVD has much better ability to detect gear fault.

Gearbox fault diagnosis under nonstationary condition using nonlinear chirp components extracted from bearing force

The main challenges in gearbox condition monitoring and fault diagnosis are the heavy noises caused by harsh operating environments and the complicatednonstationary condition during the runtime. It is extremely difficult and even impossible to obtain sensitive faultcharacteristics through direct measurements. Thus, it is ofgreat significance of faultcharacteristic inversion based on the structure vibration transfer paths for complicated mechanical systems fault diagnosis. This paper proposes a gearbox fault diagnosis method under nonstationary condition through using nonlinear chirp components extracted from bearing force (BFNCC). In the proposed method, the bearing dynamic forces are first identified based on the vibration transfer function matrix, and the intrinsic signal components are then decomposed through a nonlinear chirp mode decomposition method. The proposed method improved the quality of vibration signals by removing the effect of the structure vibration transfer paths. In comparison with order tracking, the BFNCC is capable of clearly reflecting the magnitude increase phenomenon under gear fault condition. Numerical simulation and experimental studies show that the proposed method is effective for gear fault diagnosis under nonstationary condition. The proposed method is demonstrated to be superior to the existing approaches and robust to the location of measuring points, and thus shows potential in machinery fault diagnosis under the interference of complicated structure transfer paths.

An adaptive anti-noise gear fault diagnosis method based on attention residual prototypical network under limited samples

Deep learning networks are widely used to realize the intelligent diagnosis of gear faults. However, the problem of the insufficient number of typical fault samples and strong noise often occur in practical applications. Thence, this paper proposes an adaptive anti-noise gear fault diagnosis method based on attention residual prototypical network (ARPN) under the limited sample. In order to maximize the characterization of implicit classification information under fewer samples, frequency slice wavelet transform is applied to convert the vibration signal into a time–frequency image. Then, the Bayesian optimization algorithm is introduced to automatically adjust hyperparameters to meet different application conditions. And the feature embedding stage combined with the improved Non-Local-Pooling-Attention module is constructed to capture the effective feature information better under strong noise conditions. Finally, the internal principle of the proposed model is analyzed based on the visualization process. Meanwhile, the initial application of an interpretable deep learning network in the classification of gear health status has been realized. The ARPN is verified on the Connecticut standard gear data set and the data set collected actually by the laboratory. The results show that the diagnostic accuracy of the ARPN is higher and has stronger recognition ability under different loads.

A novel wind turbine fault diagnosis based on deep transfer learning of improved residual network and multi-target data

In industrial production, problems such as lack of data, complex fault types, and low generalizability of deep learning models seriously affect the fault diagnosis of wind turbines. Therefore, we have developed a fault diagnosis model for wind turbines under harsh conditions to address the above problems. First, the collected one-dimensional vibration data is reshaped into two-dimensional form by using the Gramian Angular Field. The two-dimensional form not only extends the spatial structure of the data, but also effectively improves the information expression of the data. In addition, the data is classified into large-scale data, medium-scale data, small-scale data, class-imbalanced data, and heterogeneous data based on the data type. Then, the deep residual network structure is redesigned to improve the diagnostic performance of the model based on the sensitivity of the reshaped data to the size of the convolutional kernel, and the new structure of the network is employed to implement transfer learning. Finally, we adopt the developed fault diagnosis model to achieve the fault diagnosis of bearings and gears in the wind turbine gearbox. Meanwhile, an automatic hyperparameter search mechanism was added to improve the partial hyperparameter optimization in this study. It is demonstrated that the model proposed in this study has excellent diagnostic performance with multi-target data for wind turbines, and has excellent generalizability and reliability.

A novel wind turbine fault diagnosis method based on compressed sensing and DTL-CNN

This paper describes the development of a fault diagnosis method for identifying different fault conditions in the rolling bearings and gears of wind turbines. For the fault signal, the compressed sensing (CS) technology is used to perform noise reduction and feature extraction. The noise reduction process consists of sparse compression and reconstruction of the signal. After the data is processed by the compressed sensing technology, the noise and redundant parts of the signal can be greatly removed, and the real operating state signal of the wind turbine can be restored to the maximum. The fault diagnosis scheme is based on a combination of deep transfer learning and convolutional neural network (DTL-CNN), which is able to perform fault type identification with a small batch of rolling bearing data samples and gear samples. In this study, a new CNN structure was developed and the structure was used to achieve bearing-to-bearing and bearing-to-gear transfer fault diagnosis. Finally, the reliability and superiority of the proposed method in wind turbine rolling bearing and gear fault diagnosis are shown by the experimental results.

Fault feature extraction method of gear based on spectral amplitude modulation and autocorrelation analysis

Due to the strong noise interference and amplitude modulation effect, it is difficult to extract the impact characteristics of the fault gear from the spectrum of the fault signal. To tackle these issues, a gear fault feature extraction method based on spectral amplitude modulation (SAM) and autocorrelation analysis is proposed. First, the SAM method is used to decompose the signal into different components according to energy, and the optimal component with the most fault information is determined by combining kurtosis index and magnitude order selection range. Then, the optimal component is denoised using the autocorrelation function. Finally, the fault feature frequency is extracted by calculating the squared envelope spectrum of the denoised signal. The superiority of the method is verified by simulating the signal. Furthermore, the effectiveness and superiority of the method in gear fault diagnosis are verified by comparison with the fast kurtogram, cepstrum pre-whitening, and SAM.

A Modulation Signal Bispectrum Enhanced Squared Envelope for the detection and diagnosis of compound epicyclic gear faults

Epicyclic gearboxes are prevalent in a variety of important engineering systems such as automotive, aerospace, wind turbines, civil equipment and industrial robots owing to the merits of compact structure and high power density. To ensure the productivities of such important systems, condition monitoring techniques are being actively studied to resolve the challenge of varying transmission paths and multiple modulations in vibration signals acquired from ring gear housing. This paper proposes a new Modulation Signal Bispectrum (MSB) Enhanced Squared Envelope method to analyse the vibration signals acquired by a special On-Rotor Sensing (ORS) transducer that is mounted on the shaft of an epicyclic gearbox. A vibration signal model of ORS measurements from epicyclic gearboxes is presented to show the multiple modulation influences. Moreover, inevitable noise influences are also investigated on the conventional squared envelope and the state-of-the-art spectral correlation analysis. On this base, MSB is introduced to suppress these influences for accurately extracting equally spaced harmonics in the squared envelope and is then integrated to isolate fault signatures, allowing effective fault detection and diagnosis of epicyclic gearboxes, which lead to novel contributions of an MSB Enhanced Squared Envelope (MSB-ESE) approach. An experimental study of compound epicyclic gear faults was conducted to demonstrate the superior performance of MSB-ESE along with the special ORS technique in detecting and diagnosing the compound faults on sun and planet gears.

Nonconvex regularized sparse representation in a tight frame for gear fault diagnosis

A gear’s vibration signal consists of multiple components, so it is therefore difficult to accurately extract the transient components of gear faults. Currently, sparse representation is capable of separating fault components from multicomponent noisy vibration signals. However, sparse representation methods still suffer problems with poor computational efficiency and the underestimation of amplitude. To tackle these challenges, this paper proposes a nonconvex regularized sparse representation in a tight frame. The tunable Q-factor wavelet transform (TQWT) is proposed as a sparse dictionary, which can portray the waveform characteristics of the gear’s vibration signal. TQWT satisfies the tight-frame condition, hence it can efficiently reduce the amount of calculations. The minimax concave function is used as the penalty function since it stands out from various penalty functions with the ability to maintain amplitude. The simulation and experimental analysis show that this method has a shorter operation time and a better ability to maintain the amplitude.

Maximum reweighted-kurtosis deconvolution: a fully blind and adaptive method for restoration of gear fault impulse trains

Deconvolution based on vibration signals has been proven to be an effective tool in gear fault diagnosis. However, for many common methods, precisely restoring the fault impulse train is still a challenging task due to the great dependence on prior knowledge and the empirical determination of filter parameters. In this paper, a fully blind and adaptive method termed maximum reweighted-kurtosis deconvolution (MRKD) is proposed. A new deconvolution criterion, i.e., reweighted-kurtosis, is defined. This criterion possesses great robustness to impulse interferencesand thus has great potential to solve the problem of previous kurtosis-based methods in which a single dominant impulse is deconvolved instead of the impulse train induced by a localized fault. Furthermore, a parameter-adaptive strategy is developed to adaptively determine the appropriate filter parameters. As such, the proposed method does not require any prior knowledge of the target fault impulse train and addresses the critical issue of many common methods specifying filter parameters empirically. The proposed method is validated through simulated and real vibration signals. Comparison with the most popular deconvolution methods indicates that MRKD outperforms other methods for the restoration of a gear fault impulse train.

Fault Diagnosis Method for Wind Power Equipment Based on Hidden Markov Model

The aging of mechanical equipment brings a lot of inconvenience to its application. Therefore, research on fault diagnosis technology of mechanical equipment is of great significance for maintaining equipment safety, improving production efficiency and reliability. Bearings and gears are one of the key components of mechanical equipment, and their working conditions seriously affect the changes in equipment performance. Therefore, the fault diagnosis and performance degradation assessment of bearings and gears have always been the research focus of equipment fault diagnosis. This paper proposes a multichannel information fusion method based on the coupled hidden Markov model, discusses the application of the coupled hidden Markov model in bearing fault diagnosis and performance degradation assessment, and studies the multichannel information fusion method based on the coupled hidden Markov model. Channel monitoring data performance degradation evaluation modeling and performance index calculation method are used, and the calculation method of adaptive alarm limit is given by using the performance index. Finally, the natural failure test data and accelerated fatigue test data of gears and rolling bearings are used to analyze and verify the effectiveness of the coupled hidden Markov model for evaluating the performance degradation of complete and incomplete data. The results prove that the selected performance indicators can be quantified, reflecting the degree of bearing performance degradation.

A gear fault diagnosis method based on improved accommodative random weighting algorithm and BB-1D-TP

As an essential component of a gearbox, gears can damage a structure or even an entire gear transmission system in case of failures. As a result, advanced fault diagnosis methods are crucial to system's operation. Currently, single-signal-driven gear fault diagnosis techniques have been applied in many fields, but multipath noise and single-sensor sampling errors inevitably affected the accuracy of diagnosis. This paper proposes a gear fault diagnosis method based on a novel accommodative random weighting theory and a balanced binary one dimension ternary pattern (BB-1D-TP) model. It can accurately diagnose the types of gear failures under the circumstances of multiple channels and strong background noise. The novel accommodative random weighting algorithm reduces the total mean-square error (MSE) by adaptively adjusting the proportional connection between a measured value at a present state and a historical state. Then the balanced binary algorithm extracts texture features of fault signals for signal enhancement. In the end, the classification is done by using Support Vector Machine (SVM) method. The result of experiments demonstrated that the method in this article effectively improves accuracy and efficiency of gear fault identification.

Symplectic geometry packet decomposition and its applications to gear fault diagnosis

There are many signal decomposition methods in gear fault diagnosis at present, such as ensemble empirical mode decomposition (EEMD), wavelettransform (WT), singular spectral analysis (SSA) and symplectic geometry mode decomposition (SGMD). However, these methods have some defects. Especially, when analyze gear fault signals with strong background noise, the noise reduction performance of EEMD and WT cannot meet the requirements. SSA and SGMD can delete fault information as noise, which seriously affect the accuracy of fault diagnosis. Therefore, a novel multi-layer decomposition method, symplectic geometry packet decomposition (SGPD) is proposed. In essence, SGPD combines symplectic geometry theory and multi-layer decomposition idea of wavelet packet to decompose the signal into a series of independent components containing the main fault information. SGPD not only has excellent signal decomposition ability, but also can minimize noise while retaining the fault information of the original signals in the process of sufficiently decomposing the non-steady signal. The analysis results of the emulational and experimental signals indicate that SGPD has strong signal decomposition capabilities and noise robustness.

Application of EMD based statistical parameters for the prediction of fault severity in a spur gear through vibration signals

ABSTRACT Health monitoring of Mechanical drives had been a challenging task for the researchers, to estimate the fault severity for necessary corrective actions. Signal processing techniques made the task of finding faults quickly in the easiest way, to predict machine condition. Steps of preventive maintenance would always help in reducing the machine down time. Vibrations in mechanical systems, found to be the most common source for degrading the performance, and uncontrollable vibrations usually lead to catastrophic situations. Hence regulating machine vibrations would solve majority of the problems and prevents the state of being abnormal. In this context an empirical mode decomposition technique is utilised to study the vibration severity of a spur gearbox. This method decomposes a complex raw signal into a finite number of Intrinsic mode function (IMFs), based on the local characteristic time scale. These IMFs divulge the intrinsic oscillation modes entrenched in the signal. This enables the fault diagnosis of gears and to take essential corrective action. The present study includes acquisition of vibration signals, decomposition into IMF’s, extraction of respective statistical parameters and analysis for the quantification of various faults. The work also demonstrates the effectiveness of the processed signal over the raw signals of vibration.

A Novel Impact Feature Extraction Method Based on EMD and Sparse Decomposition for Gear Local Fault Diagnosis

Sparse decomposition has been widely used in gear local fault diagnosis due to its outstanding performance in feature extraction. The extraction results depend heavily on the similarity between dictionary atoms and fault feature signal. However, the transient impact signal aroused by gear local defect is usually submerged in meshing harmonics and noise. It is still a challenging task to construct high-quality impact dictionary for complex actual signal. To handle this issue, a novel impact feature extraction method based on Empirical Mode Decomposition (EMD) and sparse decomposition is proposed in this paper. Firstly, EMD is employed to adaptively decompose the original signal into several Intrinsic Mode Functions (IMFs). The high-frequency resonance component is separated from meshing harmonics and part of the noise. Then, the IMF with the prominent impact features is selected as the Main Intrinsic Mode Function (MIMF) based on the kurtosis. Accordingly, the modal parameters required for impact dictionary are identified from the MIMF by correlation filtering. Finally, the transient impact component is extracted from the original signal by Match Pursuit (MP). The proposed method was adequately evaluated by a gear local fault simulation signal, and the single-stage gearbox and five-speed transmission experiments. The effectiveness and superiority of the proposed method is validated by comparison with other feature extraction techniques.

Enhanced multiclass support vector data description model for fault diagnosis of gears

This work reports a study aimed at identifying the operating state of gear through an enhanced multiclass support vector data description (eMSVDD) model where the multiple hyperspheres corresponding to different operating states of gears are established. The model addresses the following issues: (1) the overlap of multiple hyperspheres deteriorates the diagnostic performance for boundary samples and severely weakens the generalization ability of the model; (2) the original support vector data description can only be applied to handle binary classification tasks; and (3) the performance of the SVDD depends heavily on the penalty factor C and kernel parameter δ. In the proposed model, we first implement the multi-label classification task by building multiple hyperspheres. Then, a novel fitness function for adaptive simplified chaotic particle swarm optimization is proposed to determine the model parameters, which considers the empirical risk minimization and model generalization capability. Finally, experimental results demonstrate that the eMSVDD outperforms deep neural networks and support vector machine by 3.4% and 0.5%, respectively, and this performance improvement respectively reaches 17.8% and 2.1% in the case of a very small sample size. The proposed approach provides a means for gear fault diagnosis with small samples.

An effective centre frequency selection scheme based on variational mode extraction and its application to gear fault diagnosis

The penalty factors and number of modes in variational mode decomposition (VMD) have to be set empirically. In addition, it is a challenge to select the most useful mode from the multiple mode components decomposed using this method. To solve these problems, an effective gear fault diagnosis method using variational mode extraction (VME) and envelope analysis is proposed. Due to the blindness in determining the centre frequency of the desired mode, this study proposes a method to select the centre frequency based on the maximum signal-to-noise index among the meshing frequency and its harmonics. The desired mode obtained according to the selected centre frequency has the maximum signal energy and relatively minimum noise energy. Two case studies verify that the presented approach is effective for gear fault feature extraction. Comparisons with VMD and empirical mode decomposition (EMD) further highlight the superiority of the method.

Multivariate local characteristic-scale decomposition and 1.5-dimensional empirical envelope spectrum based gear fault diagnosis

Most of the existing gear fault diagnosis methods only use the single-channel signal for processing. In order to extract more fault information and realize more comprehensive and accurate fault analysis, it is necessary to process the collected multi-channel signals. In this paper, a novel multivariate signal decomposition method, multivariate local characteristic-scale decomposition (MLCD) is proposed to decomposes multi-channel signal simultaneously. Comparing MLCD with multivariate empirical mode decomposition (MEMD), the results show that both methods are suitable for multivariate signal decomposition, but MLCD is superior to MEMD in computational efficiency, suppression of endpoint effect and decomposition accuracy. In order to highlight gear fault characteristic frequency, 1.5-dimensional empirical envelope spectrum (1.5D EES) is proposed. 1.5D EES combines the advantages of empirical envelope method and 1.5-dimensional spectrum, which can effectively reduce the noise of envelope signal and highlight the fault characteristics of signal. Based on the above two methods, a new gear fault diagnosis method, multivariate local characteristic-scale decomposition and 1.5-dimentional empirical envelope spectrum (MLCD-1.5D EES) is proposed and applied to multi-channel gear fault signal decomposition and fault feature extraction. Simulation and experimental results demonstrate the effectiveness and superiority of MLCD-1.5D EES.

Application of instantaneous phase detection technology based on laser displacement sensor in fault diagnosis of spur gear of rotation vector reducer

Aiming at the problem of a spur gear fault of the rotation vector (RV) reducer in industrial robots, an instantaneous phase detection technology based on eccentric modulation and laser displacement sensor is proposed. Taking the RV-20E reducer as the object, through the analysis of the RV reducer gear fault mechanism, an RV reducer fault diagnosis test bed is developed to simulate the actual operating conditions of the RV reducer of industrial robot. The instantaneous phase information of the RV reducer is collected, and the adaptive permutation entropy is used to process the instantaneous phase data, which can judge the gear fault type quickly and effectively. The test results show that the fault diagnosis test bench based on this detection technology can effectively and quantitatively judge the fault type of the spur gear of the RV reducer.

Gear fault diagnosis based on CS-improved variational mode decomposition and probabilistic neural network

Increasing the rate of gear fault diagnosis is crucial to research on gear fault diagnosis methods. The existing signal processing methods have modal aliasing phenomena and poor adaptability. Moreover, the neural network method has serious problems, such as complex training and low accuracy. Based on these problems, this study aims to improve the shortcomings of existing gear fault diagnosis methods, thereby enhancing the adaptability and accuracy of gear fault diagnosis systems. This study proposes a method based on the combination of the parameter optimisation of variational mode decomposition (VMD) with cuckoo search (CS) and the probabilistic neural network (PNN) for intelligent identification of gearbox faults. Firstly, the energy parameters of each mode of the signals are extracted by the method of CS-improved VMD, and a feature matrix is constructed. Then, the optimal training parameters of the PNN are selected and the PNN is trained, and the performance is evaluated by the parameters RMSEC and RMSEP. Use the data set from Southeast University of China and the experimental data, and compare with the diagnosis classification effect of four other fault diagnosis models. The diagnostic results of the experimental data show that the fault diagnosis accuracy of the method proposed in this paper can reach an average of 98.5%, proving the advancement and effectiveness of this method over existing fault diagnosis technologies.