Multi-class Fault Diagnosis Research Articles

Domain adaptation between heterogeneous time series data: A case study on real-time rotary machinery fault diagnosis

Artificial intelligence (AI) in fault diagnosis has emerged as a significant tool for addressing anomalies in rotary machineries (RMs). However, the success of AI models in this context largely depends on situations where the distribution of samples in the training dataset closely aligns with that of the testing dataset. In addition, practically, it is very difficult, costly, or even impossible to collect sufficient time series samples with proper labeling for AI modeling. To address these issues, this paper presents a study developing a structured methodology called frequency-domain supervised domain adaptation (FDSDA) considering heterogeneous data from versatile sensor technologies. In the context of application, two high-dimensional datasets were generated, each following distinct distributions: one being the vibration signals and the other the acoustic signals. The modeling begins with transforming signals from the time domain to the frequency domain through frequency analysis. Subsequently, a domain adaptation model is implemented, adopting an ensemble learning as an estimator to address the covariate shift challenge. Two experiments are conducted with the proposed method for binary and multi-class fault diagnosis of RMs. The results demonstrate that, despite having limited training and target data, the proposed method can effectively detect RMs fault conditions by outperforming the benchmark method.

Physical Graph-Based Spatiotemporal Fusion Approach for Process Fault Diagnosis.

The rapid development of big data technology and machine learning has increasingly focused attention on fault diagnosis in complex chemical processes. However, data-driven approaches often overlook the inherent physical correlations within the system and lack a robust mechanism for providing trusted explanations for fault diagnosis. To address this challenge, a graph-based fault diagnosis model framework is proposed along with a dependable fault node diagnosis analysis method. In order to enhance the extraction of chemical process features from a spatial perspective, a graph convolution network (GCN)-based node spatial encoding module is integrated. The construction of the adjacency matrix involves combining a priori knowledge of chemical processes with Pearson correlation, thereby incorporating the physical correlations between nodes. Simultaneously, to capture temporal dependencies in fault data, a spatiotemporal feature fusion module based on the long short-term memory network (LSTM) is employed. In terms of model training, a dual-supervision strategy is adopted to ensure stable convergence of the multiclass fault diagnosis model. For model inference, a multi-model voting strategy is designed to mitigate accuracy degradation resulting from model prediction bias. To tackle the interpretability challenge, a fault diagnosis analysis method based on node masking is designed, effectively identifying critical nodes contributing to system faults. Experimental validation on the Tennessee Eastman process demonstrates the effectiveness of the proposed model, achieving high accuracy in fault diagnosis. The average fault diagnosis rate for all fault types reaches 0.9844, showcasing state-of-the-art performance in fault diagnosis.

A novel approach for bearings multiclass fault diagnosis fusing multiscale deep convolution and hybrid attention networks

Insufficient and imbalanced samples pose a significant challenge in bearing fault diagnosis, leading to low diagnosis accuracy. However, the fault characteristics of vibration signals are weak and difficult to extract when faults occur in the early stage. This paper proposes an effective fault diagnosis method that addresses small and imbalanced sample problems under noise interference. First, the number of faulty samples in the form of 1D signals is increased mainly by the sliding split sampling method. The preprocessed data are used to create 2D time–frequency diagrams using the continuous wavelet transform (CWT), which can extract effective features to improve the data quality. Subsequently, the minority samples are oversampled by combining synthetic minority oversampling technique to realize time–frequency conversion augmented oversampling. Moreover, the clustering method and random undersampling method are introduced to prevent the overfitting and underfitting problems respectively. Then, we propose a hybrid attention mechanism to enhance the extraction of effective feature information. This combination, integrating CWT with a multicolumn modified deep residual network, effectively extracts fault characteristics and suppresses noise effects. The experimental results demonstrate the effectiveness of the proposed method by comparison with other advanced methods using two case studies of bearing datasets.

Novel Ensemble Domain Adaptation Methodology for Enhanced Multi-class Fault Diagnosis of Highly-Connected Fleet of Assets

This paper proposes a novel methodology for enhancing multi-class classification accuracy in fault diagnosis problems among domains with highly-connected fleets of assets using time series data. The approach involves appending specially tailored models to an initial model and incorporating domain adaptation techniques to account for domain variations. The methodology is demonstrated through a case study on fault diagnosis of a fleet of hydraulic rock drills, which presents challenges due to variations in sensor data between different fault classes and individual machines. Results show significant improvements in classification accuracy, both in validation and testing, upon employing ensemble models and applying domain adaptation. While the study is limited to one case study, it lays the groundwork for exploring the applicability of the proposed methodology in other contexts.

The Multiclass Fault Diagnosis of Wind Turbine Bearing Based on Multisource Signal Fusion and Deep Learning Generative Model

Low fault diagnosis accuracy in case of the insufficient and imbalanced samples is a major problem in the wind turbine fault diagnosis. The imbalance of samples refers to the large difference in the number of samples of different categories, or the lack of a certain fault sample, which requires good learning of the characteristics of a small number of samples. Sample generation in the deep learning generation model can effectively solve this problem. In this study, we proposed a novel multi-class wind turbine bearing fault diagnosis strategy based on the conditional variational generative adversarial network (CVAE-GAN) model combining multi-source signals fusion. This strategy converts multi-source one-dimensional vibration signals into two-dimensional signals, and the multi-source two-dimensional signals were fused by using wavelet transform. The CVAE-GAN model was developed by merging the variational auto-encoder (VAE) with the generative adversarial network (GAN). The VAE encoder was introduced as the front end of the GAN generator. The sample label was introduced as the model input to improve the model’s training efficiency. Finally, the sample set was used to train encoder, generator and discriminator in the CVAE-GAN model to supplement the number of the fault samples. In the classifier, the sample set is used to do experimental analysis under various sample circumstances. The results show that the proposed strategy can increase wind turbine bearing fault diagnostic accuracy in complex scenarios.

Deep convolutional tree-inspired network: a decision-tree-structured neural network for hierarchical fault diagnosis of bearings

The fault diagnosis of bearings is crucial in ensuring the reliability of rotating machinery. Deep neural networks have provided unprecedented opportunities to condition monitoring from a new perspective due to the powerful ability in learning fault-related knowledge. However, the inexplicability and low generalization ability of fault diagnosis models still bar them from the application. To address this issue, this paper explores a decision-tree-structured neural network, that is, the deep convolutional tree-inspired network (DCTN), for the hierarchical fault diagnosis of bearings. The proposed model effectively integrates the advantages of convolutional neural network (CNN) and decision tree methods by rebuilding the output decision layer of CNN according to the hierarchical structural characteristics of the decision tree, which is by no means a simple combination of the two models. The proposed DCTN model has unique advantages in 1) the hierarchical structure that can support more accuracy and comprehensive fault diagnosis, 2) the better interpretability of the model output with hierarchical decision making, and 3) more powerful generalization capabilities for the samples across fault severities. The multiclass fault diagnosis case and cross-severity fault diagnosis case are executed on a multicondition aeronautical bearing test rig. Experimental results can fully demonstrate the feasibility and superiority of the proposed method.

Machine learning based multi class fault diagnosis tool for voltage source inverter driven induction motor

AC drives are employed in process industries for varying applications resulting in a wide range of ratings. The entire process industry has seen a paradigm shift from manual to automated systems. The major factor contributing to this is the advanced power electronics technology enabling power electronic drives for smooth control of electric motors. Induction motors are most commonly used in industries. Faults in the power electronic circuits may occur periodically. These faults often go unnoticed as they rarely cause a complete shutdown and the fault levels may not be large enough to lead to a breakdown of the drive. An early detection of these faults is required to prevent their escalation into major faults. The diagnostic tool for detection of faults requires real time monitoring of the entire drive. In this work, detailed investigation of different faults that can occur in the power electronic circuit of an industrial drive is carried out. Analysis and impact of faults on the performance of the induction motor is presented. A real time monitoring platform is proposed to detect and classify the fault accurately using machine learning. A diagnostic tool also is developed to display the severity and location of the fault to the operator to take corrective measures.

Tri-axial vibration based collective feature analysis for decent fault classification of VFD fed induction motor

This paper has explored the multi-class fault diagnosis of variable frequency drive (VFD) fed induction motor over a wide operating range (5–50 Hz) using tri-axial vibration signals. The comprehensive study has converged towards an optimum solution of discrete wavelet transform–inverse discrete wavelet transform based feature extraction algorithm using Daubechies2 mother wavelet. In the present investigation, the data-driven approach, supported by experimental verification, has established the horizontal frame vibration to be the most fault informative followed by vertical and axial. However, a combination of typical tri-axial collective features with 36.4% of horizontal (4), 36.4% of vertical (4) & 27.2% of axial (3) features, picked up by RELIEFF based feature selection technique, have demonstrated very satisfactory performances with statistically independent experimental data set. The developed standalone fault diagnosis model over the entire VFD range, outperforms over the contemporary methodologies, even employing shallow classifier like multilayer perceptron. The collective tri axial feature with the data mining technique not only unveiled buried information in the area of vibration signal based motor fault diagnosis, at the same time reduces the computational burden to a great extent. Thus making the proposed algorithm robust and hardware friendly, suitable for low-cost instrumentation.

Spatio-temporal fusion neural network for multi-class fault diagnosis of wind turbines based on SCADA data

Abstract Numerous sensors have been deployed in different locations in components of wind turbines to continuously monitor the health status of the turbine system and accordingly, generate a large volume of operation data by the supervisory control and data acquisition (SCADA) system. Naturally, these sensory data are multivariate time series with high spatio-temporal correlations. It is still challenging to effectively model such correlations and then enable an accurate fault diagnosis. To this end, we proposed a new spatio-temporal fusion neural network (STFNN) for wind turbine fault diagnosis. Specifically, a multi-kernel fusion convolution neural network (MKFCNN) with multiple convolution kernels of different sizes is first designed to extract multi-scale spatial correlations among different variables. Then, we adopt the long short-term memory (LSTM) to further learn the temporal dependence of the learned spatial features. The proposed STFNN model provides an end-to-end fault diagnosis way, which can directly learn spatio-temporal dependency from the raw SCADA data and give the fault diagnosis result. The effectiveness and superiority of the proposed method are evaluated on a generic wind turbine benchmark simulation dataset and a SCADA dataset from a real wind farm. Both experimental results have indicated that the proposed method outperformed several compared methods.

Decent fault classification of VFD fed induction motor using random forest algorithm

AbstractA data-driven approach for multiclass fault diagnosis of drive fed induction motor (IM) using stator current at steady-state condition is a complex pattern classification problem. The applied DWT-IDWT algorithm in this work is reinforced by a novel selection criterion for mother wavelet application and justifies the originality of the work. This investigation has exploited the built-in feature selection process of Random Forest (RF) classifier to resolve the most challenging issues in this area, including bearing and stator fault detection. RF has shown an outstanding performance without application of any feature selection technique because of its distributive feature model. The robustness of the results backed by the experimental verification shows an encouraging future of RF as a classifier in the area of intelligent fault diagnosis of IM.

Cost sensitive active learning using bidirectional gated recurrent neural networks for imbalanced fault diagnosis

Most existing fault diagnosis methods may fail in the following three scenarios: (1) serial correlations exist in the process data; (2) fault data are much less than normal data; and (3) it is impractical to obtain enough labeled data. In this paper, a novel form of the bidirectional gated recurrent unit (BGRU) is developed to underpin effective and efficient fault diagnosis using cost sensitive active learning. Specifically, BGRU is devised to consider the dynamic behavior of a complex process. In the training phase of BGRU, the idea of weighting each training example is proposed to reduce the effect of class imbalance. Besides, in order to explore the unlabeled data, cost sensitive active learning is utilized to select the candidate instances. The effectiveness of the proposed method is evaluated on the Tennessee Eastman (TE) dataset and a real plasma etching process dataset. The experiment results show that the proposed cost senstive active learning bidirectional gated recurrent unit (CSALBGRU) method achieves better performance in both binary fault diagnosis and multi-class fault diagnosis.

Fault Diagnosis of High-speed Train Bogie Based on Deep Neural Network

As an important part of high-speed train (HST), the performance of bogie has a direct impact on the safety of train. Real-time monitoring and evaluation of the operating state of bogie are of great significance for the safe operation of the train. The traditional signal processing methods are difficult to analyze the complex vibration signals of bogie which are collected during train operation. Deep Neural Network (DNN) has been widely used in the field of fault diagnosis due to its good performance in feature extraction of complex data. In this paper, DNN is taken as the overall framework. Failure modes include four states: air springs completely fail, anti-yaw dampers completely fail, lateral dampers completely fail and normal operation. In these four states, HST is running at the speed of 200km/h. Using the data collected in one hour of simulation as the input signal of DNN, the diagnostic accuracy of the four working conditions reached 92.5%. Experimental result shows that DNN has a good performance in multi-class fault diagnosis of bogie.

Improved Classification by Non Iterative and Ensemble Classifiers in Motor Fault Diagnosis

Data driven approach for multi-class fault diagnosis of induction motor using MCSA at steady state condition is a complex pattern classification problem. This investigation has exploit ...

A Novel Approach for Multi Class Fault Diagnosis in Induction Machine Based on Statistical Time Features and Random Forest Classifier

Fault diagnosis and detection is the important area in health monitoring of electrical machines. This paper proposes the recently developed machine learning classifier for multi class fault diagnosis in induction machine. The classification is based on random forest (RF) algorithm. Initially, stator currents are acquired from the induction machine under various conditions. After preprocessing the currents, fourteen statistical time features are estimated for each phase of the current. These parameters are considered as inputs to the classifier. The main scope of the paper is to evaluate effectiveness of RF classifier for individual and mixed fault diagnosis in induction machine. The stator, rotor and mixed faults (stator and rotor faults) are classified using the proposed classifier. The obtained performance measures are compared with the multilayer perceptron neural network (MLPNN) classifier. The results show the much better performance measures and more accurate than MLPNN classifier. For demonstration of planned fault diagnosis algorithm, experimentally obtained results are considered to build the classifier more practical.

Fault diagnosis in spur gears based on genetic algorithm and random forest

There are growing demands for condition-based monitoring of gearboxes, and therefore new methods to improve the reliability, effectiveness, accuracy of the gear fault detection ought to be evaluated. Feature selection is still an important aspect in machine learning-based diagnosis in order to reach good performance of the diagnostic models. On the other hand, random forest classifiers are suitable models in industrial environments where large data-samples are not usually available for training such diagnostic models. The main aim of this research is to build up a robust system for the multi-class fault diagnosis in spur gears, by selecting the best set of condition parameters on time, frequency and time–frequency domains, which are extracted from vibration signals. The diagnostic system is performed by using genetic algorithms and a classifier based on random forest, in a supervised environment. The original set of condition parameters is reduced around 66% regarding the initial size by using genetic algorithms, and still get an acceptable classification precision over 97%. The approach is tested on real vibration signals by considering several fault classes, one of them being an incipient fault, under different running conditions of load and velocity.

Fault diagnosis of spur gearbox based on random forest and wavelet packet decomposition

This paper addresses the development of a random forest classifier for the multi-class fault diagnosis in spur gearboxes. The vibration signal’s condition parameters are first extracted by applying the wavelet packet decomposition with multiple mother wavelets, and the coefficients’ energy content for terminal nodes is used as the input feature for the classification problem. Then, a study through the parameters’ space to find the best values for the number of trees and the number of random features is performed. In this way, the best set of mother wavelets for the application is identified and the best features are selected through the internal ranking of the random forest classifier. The results show that the proposed method reached 98.68% in classification accuracy, and high efficiency and robustness in the models.

Knowledge extraction using data mining for multi-class fault diagnosis of induction motor

The feature extraction capability of rough set theory (RST) and genetic algorithm (GA) are used to extract knowledge from radial frame vibration signal for fault diagnosis of induction motor. This knowledge can assist in selecting scales for continuous wavelet transform (CWT) and mono-components required for Hilbert transform (HT) to extract fault related information from the vibration signal. Thus, the computational complexity of the signal processing tools is considerably reduced making both CWT and HT hardware friendly and suitable for real-time applications. For machine learning based automatic multi-class fault diagnosis, the performance of the classifiers are also considerably improved with significant reduction in computational burden since the redundant and irrelevant information can be effectively removed. The information obtained using data mining technique is successfully used to detect six types of induction motor faults. The results obtained are also verified in presence of high level of noise which has not been attempted earlier. The main contribution of the paper is to combine the advantages of two powerful signal processing tool like CWT and HT to extract hidden information from vibration signal in conjunction with data mining technique making them computationally efficient and easy to implement.

Multi-class fault diagnosis of induction motor using Hilbert and Wavelet Transform

The information extraction capability of two widely used signal processing tools, Hilbert Transform (HT) and Wavelet Transform (WT), is investigated to develop a multi-class fault diagnosis scheme for induction motor using radial vibration signals. The vibration signals are associated with unique predominant frequency components and instantaneous amplitudes depending on the motor condition. Using good systematic and analytical approach this fault frequencies can be identified. However, some faults either electrical or mechanical in nature are associated with same or similar vibration frequencies leading to erroneous conclusions. Genetic Algorithm (GA) is proposed and used successfully to find the most relevant fault frequencies in radial (vertical) frame vibration signal which can be used to diagnose the induction motor faults very effectively even in the presence of noise. The information obtained by Continuous Wavelet Transform (CWT) was found to be highly redundant compared to HT and thus by selecting the most relevant features using GA, the fault classification accuracy has considerably improved especially for CWT. Almost similar fault frequencies were found using CWT+GA and HT+GA for radial vibration signal.

Fault diagnosis model based on fuzzy support vector machine combined with weighted fuzzy clustering

A fault diagnosis model is proposed based on fuzzy support vector machine (FSVM) combined with fuzzy clustering (FC). Considering the relationship between the sample point and non-self class, FC algorithm is applied to generate fuzzy memberships. In the algorithm, sample weights based on a distribution density function of data point and genetic algorithm (GA) are introduced to enhance the performance of FC. Then a multi-class FSVM with radial basis function kernel is established according to directed acyclic graph algorithm, the penalty factor and kernel parameter of which are optimized by GA. Finally, the model is executed for multi-class fault diagnosis of rolling element bearings. The results show that the presented model achieves high performances both in identifying fault types and fault degrees. The performance comparisons of the presented model with SVM and distance-based FSVM for noisy case demonstrate the capacity of dealing with noise and generalization.

Multiclass fault diagnosis in gears using support vector machine algorithms based on frequency domain data

As a dominant machine learning method, the support vector machine is known to have good generalization capability in its application of the multiclass machine–fault classification utility. In this paper, an application of the SVM in multiclass gear–fault diagnosis has been studied when the gear vibration data in frequency domain averaged over a large number of samples is used. It is established that the SVM classifier has excellent multiclass classification accuracy when the training data and testing data are at identical angular speeds. However, this method relies on the availability of both the training and testing data at that particular angular speed of the gear operation. But the training data may not always be available at all angular speeds of the gear. Hence, two novel techniques, namely the interpolation and the extrapolation methods, have been proposed; these techniques that help the SVM classifier perform multiclass gear fault diagnosis with noticeable accuracy, even in the absence of the training data at the testing angular speed. This method is based on interpolating and extrapolating the training data at angular speeds near the speeds of the test data. In this study effects of choice over different kernels and parameters of SVM on its overall classification accuracy has been studied and optimum values for these are suggested. Finally, the effect on length of training data and data density on the SVM accuracy is also presented.