Extract Fault Research Articles

Gas path fault diagnosis for gas turbine group based on deep transfer learning

Gas turbines are widely used in power generation. To ensure reliability, data-driven diagnosis has become increasingly popular. However, sufficient historical data are unavailable, especially for newly-run gas turbines. An intuition arises regarding whether data from gas turbine group with long-term operation and abundant historical data, is helpful for newly-run gas turbines. Inspired by transfer learning, this paper proposes a novel method to identify the health states of newly-run gas turbines by transferring shared knowledge from data-rich gas turbines. Convolutional neural network (CNN) is employed to extract fault knowledge from data-rich gas turbines. Then, the trained CNN is finetuned with a few data from newly-run gas turbines. Experiment is presented on six datasets from simulation platform with identical-type and different-type gas turbines. Experiment shows that it improves diagnostic accuracy of newly-run gas turbines by 4.22–7.39% compared with conventional methods. Transferability and visualization analysis reveal the shared knowledge is transferred effectively.

Automatic Bearing Fault Feature Extraction Method via PFDIC and DBAS

Determining the embedded dimension of a singular value decomposition Hankel matrix and selecting the singular values representing the intrinsic information of fault features are challenging tasks. Given these issues, this work presents a singular value decomposition-based automatic fault feature extraction method that uses the probability-frequency density information criterion (PFDIC) and dual beetle antennae search (DBAS). DBAS employs embedded dimension and singular values as dynamic variables and PFDIC as a two-stage objective to optimize the best parameters. The optimization results work for singular value decomposition for bearing fault feature extraction. The extracted fault signals combined with envelope demodulation can efficiently diagnose bearing faults. The superiority and applicability of the proposed method are validated by simulation signals, engineering signals, and comparison experiments. Results demonstrate that the proposed method can sufficiently extract fault features and accurately diagnose faults.

A 1.5D Spectral Kurtosis-Guided TQWT Method and Its Application in Bearing Fault Detection

Bearings are the key parts of rotating machinery, and their operation status is related to the operation safety of the whole equipment. Vibration signals often contain periodic impulse components which can reflect the fault state of bearings. However, due to the interference of signal transmission path and the influence of operating environment noise, the periodic impulse components in the signal are often submerged by the nonperiodic transient impulse components, modulation harmonic components, and noise components. Therefore, the core problem of bearing fault diagnosis theory is used to accurately extract the frequency band of bearing fault state information and suppress the frequency band of interference information. In this paper, the signal is processed by the tunable Q-factor wavelet transform (TQWT), the midfrequency band of the signal is tightly divided by selecting different Q-values, and the 1.5D spectral kurtosis defined in frequency domain is used to select the optimal subband. Simulated analysis shows that this method can avoid low-frequency harmonic interference, nonperiodic transient impulse components, and strong noise components in the time domain. Therefore, it can effectively realize the selection of the subbands of periodic impulse components and effectively extract fault feature information. Through experimental signal analysis, TQWT has good sparsity decomposition characteristics and can reasonably divide the signal frequency band, so as to separate the useful fault characteristic frequency band and interference frequency band. At the same time, compared with the kurtosis index, 1.5D spectral kurtosis has better robustness and resolution for low signal-to-noise ratio signals, which can achieve the purpose of fault characteristic band extraction.

Rolling Bearing Fault Classification Based on Stacked Denoising Auto Encoders

Aiming at the characteristics of large capacity and diversity of rolling bearing fault data, an intelligent rolling bearing fault diagnosis method based on stacked denoising auto encoders was proposed. Firstly, Principal Component Analysis was used to reduce the dimension of the original data, and the redundant information is deleted. Then, three de-noising auto-encoders are created to train the bearing data. Then, a stack de-noising auto-encoder with three hidden layers is stacked with the trained DAE for reverse optimization. Finally, the features are input into soft-max classifier to realize rolling bearing fault diagnosis. The experimental results show that SDAE network can extract fault features effectively, and it is better than back propagation neural network in generalization and fault classification accuracy.

Rotating machinery fault diagnosis based on multivariate multiscale fuzzy distribution entropy and Fisher score

With evaluating the randomness and the nonlinear dynamic change of time sequence effectively, multiscale fuzzy distribution entropy (MFDE) is proposed to extract fault features from vibration signal. However, it is unsuitable for multivariate signals, which contain more abundant fault information. Through changing the coarse-grained and calculation process of MFDE, a novel entropy named as multivariate multiscale fuzzy distribution entropy (MMFDE) is proposed, which not only has the characteristics of MFDE, but also considers the within- and cross-channel correlations. By applying MMFDE to two kinds of simulation signals, the results show the proposed entropy is stable and reliable. A novel fault diagnosis method for rotating machinery is proposed by extracting fault features with MMFDE, selecting sensitive ones with Fisher score (FS), and identifying working state with support vector machine (SVM). Two experiments show the better diagnosis performance of the proposed rotating machinery fault diagnosis method than the related methods with the accuracy of 98.95% and 96.80% respectively.

A Novel Two-Stage Unsupervised Fault Recognition Framework Combining Feature Extraction and Fuzzy Clustering for Collaborative AIoT

Currently, with the development of the Internet of Things (IoTs) and artificial intelligence, a new IoT structure known as the artificial Intelligence of Things (AIoTs) comes into play. With the development of AIoT, a large amount of unlabeled industrial big data has been accumulated. The analysis of large amounts of unlabeled data is labor-intensive and time-consuming for diagnostic personnel. To improve this situation, a novel two-stage unsupervised fault recognition algorithm, namely, deep adaptive fuzzy clustering algorithm (DAFC) is proposed for unsupervised fault clustering in this article. DAFC amalgamates stacked sparse autoencoder (SSAE) into adaptive weighted Gath–Geva (AWGG) clustering to form an unsupervised fault recognition framework for clustering analysis of unlabeled industrial big data. SSAE can extract the highly abstract features of the original data, and adopt different unsupervised strategies to fine-tune the network in two stages. AWGG is an improvement of Gath–Geva clustering, and can adaptively obtain optimal clustering results without presetting the number of clusters. Experimental results on two different datasets show that the proposed DAFC can stably extract fault features from unlabeled data, and automatically obtain the optimal clustering results without knowing the number of clusters in advance. To the best of our knowledge, this article is the first attempt to fine-tune SSAE in an unsupervised manner, and to propose an unsupervised fault recognition framework that requires no prior knowledge or data labels at all. DAFC can be a feasible industrial big data application for collaborative AIoT. Diagnostic personnel analyze the clustering results obtained by DAFC instead of the original unlabeled data, greatly saving time and labor costs.

Time–frequency analysis via complementary ensemble adaptive local iterative filtering and enhanced maximum correlation kurtosis deconvolution for wind turbine fault diagnosis

A complementary ensemble adaptive local iterative filtering (CEALIF) and enhanced maximum correlation kurtosis deconvolution (EMCKD) approach is proposed for weak fault signals in wind turbine bearings, which are easily concealed by strong background noise and susceptible to intermittent interference. The adaptive local iterative filtering (ALIF), as a novel nonstationary signal processing technique, can perform adaptive filtering based on the signal itself characteristics. However, its mode mixing is an annoying problem. To relieve this problem, the noise-assisted CEALIF-based filtering is proposed. Nonetheless, the ambient noise present in the original signal is retained in the component of interest. Maximum correlation kurtosis deconvolution (MCKD) is an effective tool for enhancing periodic pulses. However, its deconvolution parameters need to be set manually, and are more demanding when the bearing failure is weak. To address this circumstance, this paper introduces the particle swarm algorithm (PSO) to solve the optimal deconvolution parameters and proposes EMCKD. Firstly, by employing CEALIF, the original signal is adaptively filtered into a sequence of IMFs. Then, the IMF that best characterizes the fault information is selected based on the weighted kurtosis index (WKI). Finally, the shock components of the selected IMF are enhanced to extract the periodic shock based on EMCKD. The proposed approach can accurately extract fault characteristics by analyzing the whole life cycle signals of bearings and fault signals of a 1.5 MW direct-drive wind turbine within strong background noise. Further, the proposed approach is implemented for the compound fault extraction of bearings, and the compound fault information of the inner race as well as the outer race of the bearings is successfully extracted.

An end-to-end denoising autoencoder-based deep neural network approach for fault diagnosis of analog circuit

Fault diagnosis of analog circuit is critical to improve safety and reliability in electrical systems and reduce losses. Traditional fault diagnosis methods of analog circuit usually rely on the hand design feature extractor and can not generalize well to other diagnosis domains. To address these issues, an end-to-end denoising autoencoder (EEDAE)-based fault diagnosis approach is proposed. The proposed approach includes denoising autoencoder (DAE) and a softmax classifier. The DAE is designed to automatically extract fault features from the raw time series signals without any signal processing techniques and diagnostic expertise, and then the softmax classifier is used to classify the fault mode of analog circuits. Specifically, we design a novel loss function by jointly minimizing reconstruction loss and classification loss to improve training efficiency. The proposed approach just has one training stage, in which the encoder, decoder, and classifier are trained simultaneously. The experimental results demonstrate that compared with traditional methods, the proposed method has higher accuracy and lower requirements on data.

Stockwell‐transform and random‐forest based double‐terminal fault diagnosis method for offshore wind farm transmission line

Due to the difficulty and time-consumption in locating short-distance transmission lines for deep-sea offshore wind farm (DOWF)，this paper proposes a novel double-terminal fault location method by using Stockwell-transform (ST) and random-forest (RF). After the fault type and branch are accurately determined, the accurate transmission line fault location is located. Firstly, Stockwell-transform is employed to extract fault eigenvalues from the collected wind turbine (WT) current signals, which will reduce the sensitivity of eigenvalues to noise. And the Pearson correlation coefficient (PCC) is introduced to remove duplicate eigenvalues. Secondly, the reserved fault eigenvalues are taken as inputs to the different random-forest to classify fault types and identify the fault branch, respectively. Finally, the double-terminal fault location principle is established in fault negative sequence network (only ABCG uses positive sequence components). Newton-Raphson method (NRM) is used to eliminate the influence of asynchrony data, which implies an accurate transmission line fault location for deep-sea offshore wind farm. More than 4000 fault cases data obtained by Simulink simulation verify the feasibility and performance of the proposed method. The results show that the proposed location method has a high fault recognition rate and is immune to fault inception angle, resistance, location and noise.

RETRACTED ARTICLE: Big Data Mining and Analysis Based on Convolutional Fuzzy Neural Network

The purpose of this paper is to solve the shortcomings of traditional big data mining and analysis (BDMA); this paper compares and analyzes the denoise effect of wavelet transform and wavelet packet analysis and chooses to use wavelet packet analysis method to denoise the signal and extract fault feature vector. Finally, the extracted feature vectors are divided into two libraries, including training and testing. The fuzzy convolutional neural network (FCNN) is trained by using the training sample database data, the network weights are constantly modified, and the projection of nonlinear is computed between the input features and output features. After the expected recognition accuracy is achieved, the performance of FCNN is evaluated by using the detection samples. The performance comparison and analysis are conducted with the traditional BDMA algorithm, mainly including the comparison of recognition accuracy and learning convergence speed. Two stages of feature extraction and data mining (DM) are constructed on the big data set. The experimental results prove the effectiveness of wavelet transform feature and FCNN in analyzing and mining in big data. Finally, through the example of BDMA, this paper verifies that the organic combination of wavelet transform feature extraction and FCNN algorithm obviously improves the efficiency, meets the requirements of big data analysis, and provides potential application value for BDMA.

Fault diagnosis of mine asynchronous motor based on MEEMD energy entropy and ANN

The stator current signals of mining asynchronous motor are often non-stationary, making it challenging to extract fault features in the time domain. Therefore, this paper proposes a rotor fault diagnosis method based on the combination of Modified Ensemble Empirical Mode Decomposition (MEEMD) energy entropy and Artificial Neural Network (ANN). Firstly, the stator current signals are decomposed into a series of Intrinsic Mode Function (IMF) components by the MEEMD. Secondly, the IMF components with the most abundant information are selected by the cross-correlation criterion, and their energy entropy is calculated to construct feature vectors. Finally, the feature vectors are input into the ANN for training and state recognition. The faulty motor is modeled by ANSYS Maxwell software to obtain the simulated data. It is verified that the MEEMD-ANN method is feasible for fault diagnosis of mine motors, which can accurately identify the different status of motors, including normal state, broken rotor bars, and air gap eccentricity, the recognition rate can reach 99%. The MEEMD-ANN improves the accuracy by 2% compared with the EEMD-ANN, improves the accuracy by 3.75% compared with the MEEMD-SVM.

Probabilistic Neural Network-Aided Fast Classification of Transmission Line Faults Using Differencing of Current Signal

Electrical power transmission lines are most vulnerable to different faults due frequent atmospheric hazards. Hence, fault detection and classification are imperative to restrict unwanted power outage by isolation of the faulted line. A probabilistic neural network (PNN)-based fault classification methodology is proposed here. This work uses differencing-based modulated fault signals as input to the PNN architecture to extract fault features in terms of three-phase fault intensity index, which are further analyzed for direct classification of faults aided by a decision tree like study. Simulation of faults has been carried out in practical alike simulation environment with variation of fault location, fault resistance and inherent power line noise, which helps to develop robustness of fault analyzer. Classifier accuracy of 99.33% is achieved here, and more importantly, using only (1/6)th fraction of post fault signals, which is considerably low compared to similar contemporary works. Furthermore, the proposed classifier is equipped with the ability of detecting ground fault without inspecting the neutral current. This method uses only 35.7% of the total fault locations for training, and the rest for validating the model, which is also an above average performance. Finally, the analyzer comprises of simple PNN model as the only classifier, hence possess considerably low computational complexity.

A New Health Condition Detection Method for Planetary Gears Based on Modified Distributed Compressed Sensing and Multiscale Symbol Dynamic Entropy

Planetary gear transmission system is an important transmission part of large machinery and is prone to failure. Aiming at the problem of how to extract fault information from vibration signals of nonlinear and nonstationary planetary gearboxes, a performance degradation evaluation index of planetary gearboxes based on improved distributed compressed sensing and modified multiscale symbolic dynamic entropy (DCSMDE) is proposed. DCSMDE combines distributed compression sensing with modified multiscale symbol dynamic entropy and solves the problem of strong nonlinearity and strong vibration signal coupling of the planetary transmission system from the homologous signals of multiple sensors. A distributed compression sensing parameter optimization algorithm based on Rényi entropy is proposed, which uses improved distributed compression sensing technology to simultaneously sample, compress, and denoise the multisource vibration data of rotating machinery. DCSMDE is used to calculate the reconstructed signal, extract the features with higher recognition characteristics, and use the change trend of the DCSMDE value to judge the working status of the planetary gearbox. Experimental results show that DCSMDE can be applied to dynamic evolution and fault identification of mechanical systems and accurately classify actual fault signals. It provides a new idea for the classification of planetary gear faults and the recognition of performance degradation.

An optimized adaptive PReLU-DBN for rolling element bearing fault diagnosis

Rolling element bearings are critical components in industrial rotating machines. Faults and failures of bearings can cause degradation of machine performance or even a catastrophe. Therefore, it is significant to perform bearing fault diagnosis accurately and effectively. Deep Learning based approaches are promising for bearing diagnosis. They can extract fault information efficiently and conduct accurate diagnosis. However, the structure of deep learning networks is often determined by trial and error, which is time consuming and lacks theoretical guidance. Besides, the traditional deep learning approaches have low diagnosis accuracy and learning efficiency. To address these problems, this paper proposes a rolling element bearing fault diagnosis approach based on principal component analysis and adaptive deep belief network with Parametric Rectified Linear Unit activation layers. In the proposed approach, particle swarm optimization is integrated to obtain an optimal DBN structure with high accuracy and convergence rate. Experiments on tapered roller bearings and comparison studies with state-of-the-art methods are conducted to demonstrate the effectiveness and accuracy of the proposed approach.

Cascaded Nonlinear Mass Fluctuation Stochastic Resonance System and Its Application in Bearing Fault Diagnosis

Stochastic resonance (SR) can realize bearing fault signal diagnosis by transferring noisy energy. In order to enhance the output response of the system and realize effective signal extraction, the nonlinear mass fluctuation SR system caused by nonlinear asymmetric dichotomous noise is cascaded to obtain the cascaded nonlinear mass fluctuation SR system. First, the output amplitude gain of the first-stage of the system is derived, and the influence of different parameters on it is explored; then the effects of different parameters of the cascaded system on the output amplitude gain and the output SNR are studied separately, which proves that the cascaded system can effectively double enhance the output response of the system; finally, the adaptive genetic algorithm is used to solve the difficulty of parameter adjustment, and the cascaded nonlinear mass fluctuation SR system is applied to the bearing fault diagnosis. The system proposed in this paper takes into account the effects of nonlinear asymmetric dichotomous noise and cascaded systems and performs waveform smoothing and double enhancement of the output signal. It can better extract fault signals and has effective engineering value.

Sustainable Fault Diagnosis of Imbalanced Text Mining for CTCS-3 Data Preprocessing

At present, the method for fault diagnosis and maintenance of the CTCS-3 (Chinese Train Control System Level 3) electronic equipment relies too heavily on expert knowledge. Moreover, the use of historical fault data is not valued. This paper proposes a sustainable fault diagnosis model based on imbalanced text mining. First, to process fault data from the field recorded in natural language, natural language processing technology is used to extract fault feature words. Then, a term frequency-inverse document frequency model is used to transform the fault feature words extracted from the database into vectors. It is worth noting that imbalance in the fault samples affects the accuracy of this sustainable fault diagnosis model. To solve this problem, we use the borderline-synthetic minority over-sampling technique in the step of predicting train fault components, we also use the backpropagation neural network we proposed and the naive Bayesian model which is commonly used as a classification model, to compare the prediction accuracy of these two algorithms. The experimental results perform well, which proves that the fault diagnosis method using the backpropagation neural network can further assist engineers to complete timely repair and maintenance work. The research in this paper has played a very important role in technical support for intelligent train dispatching and command, and will also play a positive role in technical support for the automatic operation of urban rail transit under the prevention and control of the new coronavirus.

A new bearing fault diagnosis method based on signal-to-image mapping and convolutional neural network

Fault diagnosis is important to ensure the safety and efficience of mechanical equipment. In recent years, data-driven fault diagnosis methods have received extensive attention and research from many scholars. Different from the traditional fault feature extraction methods that rely on expert experience, this paper proposes a feature extraction method based on deep learning (DL). In order to meet the needs of neural networks for the amount of data, data augmentation is emploied to increase the amount of original data. Then, a novel signal-to-image mapping (STIM) is proposed to convert the one-dimensional vibration signals into two-dimensional grey images, which greatly reduce the human involvement. Finally, a convolutional neural network (CNN) model is established to extract fault features from grey images and realize fault classification. The learning process of the CNN model is analyzed and two different bearing experiment datasets are used to verify the effectiveness of the proposed method.

A Rolling Bearing Fault Classification Scheme Based on k-Optimized Adaptive Local Iterative Filtering and Improved Multiscale Permutation Entropy

The health condition of the rolling bearing seriously affects the operation of the whole mechanical system. When the rolling bearing parts fail, the time series collected in the field generally shows strong nonlinearity and non-stationarity. To obtain the faulty characteristics of mechanical equipment accurately, a rolling bearing fault detection technique based on k-optimized adaptive local iterative filtering (ALIF), improved multiscale permutation entropy (improved MPE), and BP neural network was proposed. In the ALIF algorithm, a k-optimized ALIF method based on permutation entropy (PE) is presented to select the number of ALIF decomposition layers adaptively. The completely average coarse-graining method was proposed to excavate more hidden information. The performance analysis of the simulation signal shows that the improved MPE can more accurately dig out the depth information of the time series, and the entropy value obtained is more consistent and stable. In the research application, rolling bearing time series are decomposed by k-optimized ALIF to obtain a certain number of intrinsic mode functions (IMFs). Then the improved MPE value of effective IMF is calculated and input into backpropagation (BP) neural network as the feature vector for automatic fault identification. The comparative analysis of simulation signals shows that this method can extract fault information effectively. At the same time, the experimental part shows that this scheme not only effectively extracts the fault features, but also realizes the classification and identification of different fault modes and faults of different degrees, which has a certain application prospect in the research and application direction of rolling bearing fault identification.

Single-ended line fault location method for multi-terminal HVDC system based on optimized variational mode decomposition

In order to provide a backup plan for double-ended fault location scheme when data communication and synchronization are not effective or available, a single-ended fault location method is proposed using traveling wave (TW) theory. The time-frequency characteristics is utilized to extract fault feature for fault location estimation. Specifically, Variational Mode Decomposition (VMD) is adopted and further improved with singular entropy based parameter optimization and TW arrivals can be clearly identified via intrinsic mode function (IMF) result. In order to solve the issue of highly attenuated TW front, Teager Energy Operator (TEO) is used to visualize the singular points in IMF and the peaks in TEO spectrum indicate the fault TW arrivals. Moreover, frequency-dependent TW propagation velocity is determined by center frequency of selected IMF to enhance accuracy. The fault simulation is performed on the platform PSCAD/EMTDC based on example model of 15-bus CIGRE HVDC system and MATLAB is used for signal processing and algorithm development. The effectiveness and superiority of this method is verified with simulation study considering potential influence.

Fault Diagnosis Approach for Rotating Machinery Based on Feature Importance Ranking and Selection

The key to fault diagnosis of rotating machinery is to extract fault features effectively and select the appropriate classification algorithm. As a common signal decomposition method, the effect of wavelet packet decomposition (WPD) largely depends on the applicability of the wavelet basis function (WBF). In this paper, a novel fault diagnosis approach for rotating machinery based on feature importance ranking and selection is proposed. Firstly, a two-step principle is proposed to select the most suitable WBF for the vibration signal, based on which an optimized WPD (OWPD) method is proposed to decompose the vibration signal and extract the fault information in the frequency domain. Secondly, FE is utilized to extract fault features of the decomposed subsignals of OWPD. Thirdly, the categorical boosting (CatBoost) algorithm is introduced to rank the fault features by a certain strategy, and the optimal feature set is further utilized to identify and diagnose the fault types. A hybrid dataset of bearing and rotor faults and an actual dataset of the one-stage reduction gearbox are utilized for experimental verification. Experimental results indicate that the proposed approach can achieve higher fault diagnosis accuracy using fewer features under complex working conditions.