Sensitive Fault Features Research Articles

Research on Fault Diagnosis Method of Rolling Bearing Based on Feature Optimization and Self-Adaptive SVM

Aiming at the problems of the difficulty of extracting the fault sensitive features in the mixed domain of bearings under complex working conditions and the lack of self-adaptability of support vector machine (SVM) parameters, to solve these problems, we investigate a diagnosis method combining optimal feature selection and self-adaptive SVM. Firstly, the optimal feature space through multicluster feature selection (MCFS) is constructed. Then, the differential evolution (DE) strategy is used to improve the search performance of gray wolf optimizer (GWO) to improve the diagnosis accuracy of SVM and make it strong self-adaptability. The experimental results show that, with a small number of sensitive features containing significant category differences, the proposed method not only guarantees the construction of the optimal feature space under the minimum feature dimension, but also greatly improves the accuracy of fault recognition. Simultaneously, compared with the traditional feature reduction method (PCA), the proposed MCFS combined diagnosis model improves the accuracy from 98.5% to 100% with 1.5%. Concurrently, compared with MPA-SVM, GWO-SVM, and PSO-SVM, the convergence performance is improved by 61.54%, 78.26%, and 92.64%, respectively. It can seek the best classification performance in the shortest time. The effectiveness and superiority of the proposed method are fully verified.

Imbalanced fault diagnosis of rolling bearing using improved MsR-GAN and feature enhancement-driven CapsNet

Traditional fault diagnosis approaches of rolling bearing often need abundant labeled data in advance while some certain fault data are difficult to be acquired in engineering scenarios. This imbalanced fault data problem limits the diagnostic performance. To solve it, an imbalanced fault diagnosis approach based on improved multi-scale residual generative adversarial network (GAN) and feature enhancement-driven capsule network is proposed in this paper. Firstly, frequency slicing wavelet transform is utilized to extract two-dimensional time–frequency features from original vibration signals. By designing multi-scale residual network structure and hybrid loss function, original GAN model is improved, generating high-quality fake time–frequency features to balance fault data distribution. To increase the attention of the diagnostic model to fault-sensitive features and suppress irrelevant features, a feature enhancement network is designed to dynamically weight the fault features by modeling the feature importance. On this basis, enhanced performance of imbalanced fault classification is achieved. Verification experiments demonstrate that it performs well in processing the imbalanced fault data, and has better stability and diagnostic accuracy than state-of-the-art methods.

Sequential Mechanical Fault Diagnosis in High Voltage Circuit Breaker using Attention Mechanism

To address the shortcomings of traditional shallow-based model in terms of feature extraction and generalization capacity, this paper provides a high-voltage circuit breaker mechanical condition online detection scheme by using attention mechanism to locally weight the sample correlation and combining convolutional neural network (CNN) and long short-term memory (LSTM) network. The network uses convolutional layers for feature transformation of the raw vibration data, combined with the local time-domain feature representation capability of the gated unit, to extract fault-sensitive features. The temporal deep learning model using the attention mechanism enables it to extract global features of long-time sequences while making important features obtain higher weight parameters. The resulting model improves the overall learning and recognition efficiency. The experiments on the vacuum circuit breaker show that the proposed model has better performance than the classical diagnostic method based on the support vector machine.

Research on early fault warning methods for key transmission components of wind turbines

Aimed at the problem that the current wind turbine online monitoring system makes it difficult to achieve early warning of mechanical transmission system faults, an early fault warning method for key transmission components of wind turbines is proposed. This method combines nonlinear global local preservation projection, feature dimensionality reduction and metric learning Mahalanobis distance analysis models, and establishes a nonlinear mapping relationship between sensor signal sample points in the feature space and feature values under different health conditions. According to the early fault-sensitive features extracted by the feature reduction fusion method, the distance between the states under different working conditions is measured, and the double bandwidth coefficient threshold model is used to calculate the fault warning threshold to realize the early fault warning of the mechanical transmission system of the wind turbine. Experimental research and engineering examples show that this method can realize early warning of mechanical faults and play an important role in preventing mechanical accidents of wind turbines.

Convolutional Bi-directional Long Short Term Memory Network based Dynamic Fault Diagnosis for Transformer DGA

Dissolved gas analysis (DGA) method is one of the important methods to detect early internal faults in transformers. Aiming at the shortcomings of feature extraction and online modelling in the analysis model based on shallow layer, a transformer DGA online fault diagnosis method combining convolutional neural network and bidirectional long short term memory network is proposed in this paper. By using multi-time characteristic gas data as the input of the diagnostic model, combined with convolution and pooling, the fault sensitive features were extracted, and the time dimension features were extracted by using the recursive neural network with gated structure. Under various evaluation criteria, it is compared with convolutional neural network, bidirectional long short-term memory (Bi-LSTM) network and dynamic support vector machine. The experimental results show that the online transformer diagnosis model proposed in this paper, which considers both time and space characteristics, has higher prediction accuracy.

Fault diagnosis of rolling bearing using marine predators algorithm-based support vector machine and topology learning and out-of-sample embedding

The long-term safe operation of rotating machinery is closely related to the stability of rolling bearings. This paper proposes a rolling bearing fault diagnosis method based on refined composite multiscale fuzzy entropy (RCMFE), topology learning and out-of-sample embedding (TLOE), and the marine predators algorithm based-support vector machine (MPA-SVM). First, the RCMFE algorithm is used to extract the features of the original rolling bearing fault signal and to construct the original high-dimensional fault feature set. Then, TLOE is used to reduce the dimensionality of the high-dimensional fault feature set. The low-dimensional sensitive fault features are extracted to construct a low-dimensional fault feature subset. Finally, fault-type discrimination is performed using the MPA-SVM. The Case Western Reserve University dataset and data from fault diagnosis experiments performed on 1210 self-aligning ball bearings were used to verify the proposed method. The results demonstrate the effectiveness of the fault diagnosis method, which can diagnose bearing faults with up to 100% accuracy.

Misfire detection of diesel engine based on convolutional neural networks

With the ever-stringent vehicles exhaust emission standard and higher requirements on on-board diagnostic technology, the importance of misfire detection in vehicle emission control is emerging. The performance of a traditional misfire detection algorithm predominantly depends on the features and classifier selected. Fixed and handcrafted features require either a reliable dynamic model of an engine or a large number of experiment data to define the threshold, and then, form a map. Since convolutional neural networks (CNNs) have an inherent adaptive design and integrate the feature extraction with classification functions into a compact learning framework, the misfire fault-sensitive features can be auto-discovered from raw speed signals. Furthermore, CNNs can detect the fault features of the misfire through network training with fewer engine operating conditions. In this paper, the theory and method of the misfire diagnosis based on CNNs are presented. The experimental data for network training and testing are sampled on a six-cylinder inline diesel engine. The misfire patterns containing every one-cylinder and two-cylinder misfire are tested under the wide speed and load conditions of the engine. The results show that when the engine operates under steady-state conditions, one-cylinder or two-cylinder complete misfires can be detected accurately by CNNs. In addition, one-cylinder partial misfire is employed to examine the adaptability of trained 1-D CNN. It turns out that when the partial misfire reaches the same level as half amount of the normal fuel injection quantity, one-cylinder partial misfire can be detected with accuracy more than 96%. At last, the misfire detection under the non-stationary conditions, such as acceleration or deceleration, is conducted. The results show the 1-D CNN performed well in a limited acceleration range, and network failure occurs when the absolute acceleration of the engine speed is more than 100 r/min/s.

An Intelligent Data Mining-Based Fault Detection and Classification Strategy for Microgrid

The specific characteristics and operations of microgrid cause protection problems due to high penetration of distributed energy resources. To resolve these issues, the proposed scheme employs the Hilbert transform and data mining approach to protect the microgrid. First, the Hilbert transform is used to preprocess the faulted voltage and current signals to extract the sensitive fault features. Then, the obtained data set of the extracted features is input to the logistic regression classifier for fault detection. Later, fault classification is done by training the AdaBoost classifier. In the proposed scheme, the simulation results for feature extractions are evaluated on a standard International Electrotechnical Commission (IEC) medium voltage microgrid, compatible with MATLAB/SIMULINK software environment, whereas, Python is used for training and testing of data mining model. The results are evaluated under grid-connected and islanded modes for both looped and radial configurations by simulating various fault and no-fault cases. The results show that the accuracy of the proposed logistic regression and AdaBoost classifier is higher when compared to decision tree, support vector machine, and random forest methods. The results further validate the robustness of the proposed method against the measurement noise.

Incipient fault diagnosis method for DC–DC converters based on sensitive fault features

DC–DC converters play an important role in electrical systems. The fault state of a DC–DC converter has a major impact on the operation of the back-end components and the entire electrical system. Therefore, the fault diagnosis is very necessary when the fault state of DC–DC is in the early stage. It can decrease the likelihood of happening the serious faults that may result in the enormous economic loss. To effectively diagnose the incipient fault in DC–DC converters, an incipient fault diagnosis method based on sensitive fault features is proposed. Firstly, for each type of the incipient fault, this study obtains the statistical features in time domain and the wavelet analysis local energy values in the frequency domain of the output. Then to further improve the incipient fault diagnosis accuracy, this study removes the redundant fault features based on overlap calculation, the unique fault features for each type of incipient fault will be selected. Finally, these sensitive fault features are used to build support vector data description models, which can diagnose the incipient faults. Simulation and hardware experimental results validate the practicability and effectiveness of the proposed method.

SCADA-data-based wind turbine fault detection: A dynamic model sensor method

Fault detection based on data from the supervisory control and data acquisition (SCADA) system, which has been installed in most MW-scale wind turbines, has brought significant benefits for wind farm operators. However, the changes in the features of hardware sensor measurements, which are used in current SCADA systems, often cannot provide reliable early alarms. In order to resolve this problem, in this paper, a novel dynamic model sensor method is proposed for the SCADA data based wind turbine fault detection. A dynamic model representing the relationship between the generator temperature, wind speed, and ambient temperature is derived following the first principles and used as the basic structure of the model sensor. When the model sensor is applied for fault detection, its parameters are updated regularly using the generator temperature, wind speed, and ambient temperature data from the SCADA system. Then, from the updated model, the fault sensitive features of wind turbine system are extracted via performing system frequency analysis and used for the turbine fault detection. This novel model sensor method is applied to the SCADA data of a wind farm of 3 wind turbines currently operating in Spain. The results show that the proposed method can not only detect the turbine generator fault but also reveal the trend of ageing with the wind turbine generator, demonstrating its capability of failure prognosis for wind turbine system and components.

Gas sensor fault diagnosis based on Convolutional Neural Network

Gas sensor often fails due to the influence of temperature, humidity, lighting, dust and toxic gases, and causes unreliable phenomena. Therefore, the fault diagnosis of gas sensors is considered as a weak link. Feature extraction and classification play an important role in gas sensor fault diagnosis. However, there are many problems in traditional feature extraction methods. For example, 1) requirements for expert experience, 2) sensitivity of changes in mechanical systems, 3) limitations of new feature extraction. Therefore, it is meaningful and attractive to develop a method that can discover and learn fault-sensitive features of gas sensors from the original data and classify them effectively according to the sensitive features. Convolutional neural network(CNN) has been widely applied in image analysis and voice recording, and has achieved great success. However, gas sensor fault diagnosis is rarely applied. This paper focuses on using CNN to learn fault features from data, extract features automatically, and then classify them effectively. Experiments show that the CNN method provides a effective solution for gas sensor fault diagnosis.

Multi-Block Fault Detection for Plant-Wide Dynamic Processes Based on Fault Sensitive Slow Features and Support Vector Data Description

This study proposes a multi-block fault detection method based on fault-sensitive slow features for large-scale dynamic industrial processes. Firstly, slow feature analysis (SFA) can effectively extract the process dynamic information. However, the slowest changing features may not contain more fault information. Thus, through the analysis of T 2 statistic in SFA-based process monitoring model, a fault sensitivity coefficient is defined as a new slow feature sorting criterion to select the most sensitive slow features to fault in each variable direction. Then, considering the unknown characteristics of the fault in the real-time monitoring process, the monitoring model is established for each dimension of variables based on the multi-block strategy. Finally, the support vector data description is used as a fusing method to integrate the statistics calculated in each sub-block to obtain an intuitive detection result. The effectiveness and superiority of the proposed strategy are demonstrated by the experiments on Tennessee Eastman benchmark process and an actual blast furnace ironmaking process.

Rolling Element Bearing Fault Diagnosis using Empirical Mode Decomposition and Hjorth Parameters

The condition of rolling element bearing can be monitored by analyzing vibration signatures of machines at regular intervals. Many complex and computationally intensive bearing fault diagnosis schemes have been proposed in the past, extracting various features of vibration signals in time, frequency and time-frequency domains. This paper presents a time efficient and easy to implement scheme to diagnose bearing faults using Hjorth Parameters in time domain. The proposed scheme is experimentally validated on Case Western Reserve University Bearing Data Set, that comprises of normal faultless vibration signals and vibration signals with inner race, outer race and ball bearing faults. The feature vector is computed by calculating Hjorth parameters of representative Intrinsic Mode Functions obtained through Empirical mode decomposition of raw vibration signals. The data is then classified using four Rule-based classifiers. It is observed that PART classifier exhibits the best performance with an accuracy of 96.14% and 93.82 % for training and test datasets respectively. This paper proposes and validates that Hjorth Parameters, when used with Rule based machine learning algorithms, can be effectively used as fault sensitive features for rolling element bearing fault diagnosis.

Fault diagnosis of planetary gearbox using a novel semi-supervised method of multiple association layers networks

Abstract The vibration signal can effectively represent the equipment fault information and be utilized for fault diagnosis by further extracting sensitive features. However, traditional supervised diagnosis requires a huge amount of labeled samples, and the features extraction and selection are mainly done manually, which are costly and time consuming. Deep Learning (DL) can automatically extract fault-sensitive features and effectively improve the recognition rate of fault diagnosis, while Semi-Supervised Learning (SSL) trains a model together with labeled and unlabeled data to increase recognition accuracy. In this paper, a novel deep Semi-Supervised method of Multiple Association Layers Networks (SS-MALN) framework of planetary gearbox vibration-based fault diagnosis is proposed. The SS-MALN model has the advantages of SSL and DL simultaneously, which can reduce the amount of labeled samples and improve the accuracy of recognition. The wavelet packet transform was employed to highlight various impulse components and present time-frequency features of the original vibration signal. Then, the transformed samples were divided into two parts, a major part of which deletes the category information, i.e. the labels. After that, the labeled and unlabeled samples were employed to train the improved SS-MALN model. Validation by Drivetrain Diagnostic Simulator testbed data on several planetary gearbox fault datasets manifests that our SS-MALN semi-supervised method can deliver enhanced performance over other traditional deep networks of planetary gearbox fault classification with less labeled data.

Application of a new EWT-based denoising technique in bearing fault diagnosis

The vibration signal analysis is a popular method for extracting sensitive fault features. The vibration signals are usually contaminated by noise, and therefore the extracted features cannot be providing sufficient information about the bearing faults. In this paper, a new technique is introduced for denoising the vibration signals and recognizing the bearing faults based on the empirical wavelet transform (EWT). Firstly, the vibration signals are decomposed by the EWT method into a set of functions called the empirical modes. Then, the noise-dominate modes have been denoised by an improved thresholding function that has been recently presented. Finally, the kurtosis parameter and the envelope spectrum of the denoised signal are used for early fault detection and diagnosing the fault type, respectively. The result of the simulated signal and different experimental datasets illustrate that the presented work is preferable for the empirical mode decomposition based denoising technique in the early fault detection.

Intelligent Diagnosis of V-Type Marine Diesel Engines Based on Multifeatures Extracted From Instantaneous Crankshaft Speed

This paper presents an intelligent combustion fault diagnosis approach for V-type marine diesel engines with uneven firing intervals based on the instantaneous crankshaft speed (ICS). First, experimental ICSs of a 16-cylinder marine diesel engine under normal and different degrees of fault conditions are investigated to obtain the most sensitive fault features. A new polar representation method named cyclic-polar-map (CPM) is proposed to effectively demonstrate the fault features. The quantitative radii of the CPM are used to indicate the location and degree of fault. Then, a support vector machine (SVM)-based intelligent fault diagnosis system is developed for automatically detecting the combustion faults with the features extracted from the proposed CPM and the frequency analysis of ICS. In addition, for the SVM training set is not always available in practice, a dynamic model of the crankshaft system is developed to simulate different fault patterns. Finally, the efficiency of the proposed intelligent diagnosis approach is validated by experimental data. The results show that the proposed SVM-based intelligent diagnosis system using both CPM features and frequency domain features of the ICS can obtain good accuracies under different fault degrees at various speed and load conditions.

Vibration Signature Extraction of High-Voltage Circuit Breaker by Frequency and Chaotic Analysis

This paper aims at presenting a novel approach for the analysis of vibration signals detected from circuit breaker (CB) opening or closing operations and to mechanical fault identification. An adaptive signal decomposition approach and chaotic nonlinear dynamic technique are used for extracting fault sensitive features of CB. The fault sensitive frequency band is derived using variational mode decomposition (VMD) combining with Hilbert marginal spectrum (HMS). The obtained sensitive frequency band(s) are employed for the reconstruction of the dynamical attractor (chaotic dynamics viewed from a phase space perspective, namely, the attractor-based perspective) following the Taken's embedding theorem. The invariant measures and ergodic quantities such as the largest Lyapunov exponent, correlation dimension, and Kolmogorov entropy which can be estimated on the reconstructed attractor are presented as the fault sensitive features. The CB fault is estimated in a chaotic space using these sensitive features. The effectiveness of the developed method is evaluated by using data sets recorded from several vacuum CBs with the same type.

An Adaptive Weighted Multiscale Convolutional Neural Network for Rotating Machinery Fault Diagnosis Under Variable Operating Conditions

Extracting robust fault sensitive features of vibration signals remains a challenge for rotating machinery fault diagnosis under variable operating conditions. Most existing fault diagnosis methods based on the convolutional neural network (CNN) can only extract single-scale features, which not only loss fault sensitive information on other scales, but also suffer from the domain shift problem. In this work, a novel end-to-end deep learning network named adaptive weighted multiscale convolutional neural network (AWMSCNN) is proposed to adaptively extract robust and discriminative multiscale fusion features from raw vibration signals. The AWMSCNN consists of three main components: the denoising layer, the adaptive weighted multiscale convolutional (AWMSC) block, and the multiscale feature fusion layer. The AWMSC block can learn rich and complementary features on multiple scales in parallel. Then, an adaptive weight vector is introduced to modulate multiscale features to emphasize fault sensitive features and suppress features that are sensitive to operating conditions. The train wheelset bearing dataset and the bearing dataset provided by Case Western Reserve University (CWRU) are used to verify the superiority of the proposed model over the basic CNN and other multiscale CNN models. The experiment results show that the proposed model has strong fault discriminative ability and domain adaptive ability against variable operating conditions.

Early Fault Detection of Machine Tools Based on Deep Learning and Dynamic Identification

In modern digital manufacturing, nearly 79.6% of the downtime of a machine tool is caused by its mechanical failures. Predictive maintenance (PdM) is a useful way to minimize the machine downtime and the associated costs. One of the challenges with PdM is early fault detection under time-varying operational conditions, which means mining sensitive fault features from condition signals in long-term running. However, fault features are often weakened and disturbed by the time-varying harmonics and noise during a machining process. Existing analysis methods of these complex and diverse data are inefficient and time-consuming. This paper proposes a novel method for early fault detection under time-varying conditions. In this study, a deep learning model is constructed to automatically select the impulse responses from the vibration signals in long-term running of 288 days. Then, dynamic properties are identified from the selected impulse responses to detect the early mechanical fault under time-varying conditions. Compared to traditional methods, the experimental results in this paper have proved that our method was not affected by time-varying conditions and showed considerable potential for early fault detection in manufacturing.

A novel fusion diagnosis method for rotor system fault based on deep learning and multi-sourced heterogeneous monitoring data

Deep learning-based fault diagnosis has been acclaimed for its superiority in adaptively mining salient features. The monitoring data used as the input of deep learning typically includes only structured data (e.g. vibration signals, voltage signals, and acoustic emission signals), not unstructured data (e.g. infrared images), which provides another perspective on the mechanical health condition. To apply multi-sourced heterogeneous monitoring data fully, this paper presents a novel fusion diagnosis method integrating structured and unstructured data for rotor system faults. A novel multi-mode convolutional neural network (M-CNN) is first proposed to automatically learn fault-sensitive features from raw multisensory data composed of vibration signals and infrared images. M-CNN equipped with adjustable filter banks can identify data types and adaptively adopt the appropriate convolution mode. Then, t-distributed stochastic neighbor embedding (t-SNE) is introduced to fuse the deep features to further improve the quality of the learned features. Finally, the fused features are employed to conduct fault classification tasks. The effectiveness of the scheme is verified by fault diagnosis experiments in a rotor system, where it achieves a remarkable classification rate of 98.97%. Compared with similar methods, the proposed method exhibits outstanding performance, indicating the feasibility of using multi-sourced heterogeneous data for rotor system fault-diagnosis.