Incipient Fault Detection Research Articles

Sweep reflection coefficient method for localization of internal faults in transformer windings

Early-stage detection of incipient internal faults like inter-turn fault or winding mechanical fault in a transformer is of great importance to avoid its unplanned outage or expensive consequential failures. Frequency response analysis is widely used for diagnostics of such defects. In this contribution, the sweep reflection coefficient (SRC) has been proposed as a proper frequency response not only for the detection and discrimination of the faults but also for their localization. Therefore, a two-step approach is introduced based on the comparison of the SRCs of the healthy and faulty windings. As the first step, the type of fault is distinguished, and in the second step, an appropriate localization algorithm is employed depending on the type of fault. The ability of this method has been investigated in a model winding as well as a distribution transformer by applying resistive, capacitive, and hybrid turn to ground and inter-turn faults as well as several complex internal faults.

Incipient fault detection and diagnosis of nonlinear industrial process with missing data

BackgroundIn real industrial process, timely detection and diagnosis incipient fault is often more meaningful. At the same time, due to sensor failures or data acquisition system failures, process data may be missing or corrupted, resulting in loss of process information. MethodsIn view of the above problems, a Mixed Kernel function Dissimilarity Neighborhood Preserving Embedding (MKDNPE) method is proposed. Firstly, Low Rank Matrix Decomposition (LRMD) is used to recover the missing data, the recovered low rank matrix contains the main information of the process. Then, the MKDNPE model is developed in the recovered low rank matrix, where the mixed kernel function is composed of a Gaussian radial basis kernel function and a polynomial kernel function. It can simultaneously extract the local information of process data and the global characteristics of data structure, and deal with the nonlinear characteristic of process. Finally, the dissimilarity statistic is introduced for incipient fault detection, and the method based on contribution chart is used for fault diagnosis. Significant findingsA numerical example and two benchmark processes are carried out for simulation verification. The simulation results further verified that the proposed method has good detection and diagnosis capabilities for incipient nonlinear faults in industrial processes with missing data.

Edge-based Explainable Fault Detection Systems for photovoltaic panels on edge nodes

This paper presents an eXplainable Fault Detection Systems (XFDS) for incipient faults in PV panels. The XFDS is realizable on simple edge devices and has four main components: (i) irradiance-based three diode model (IB3DM), (ii) data-based fault classifier, (iii) eXplainable Artificial Intelligence (XAI) application, and (iv) edge node implementing XFDS. The IB3DM is a high fidelity model having nine parameters, modeling the PV panel behavior even at low irradiance conditions, thereby helping incipient fault detection. The fault-classifier uses an Extreme gradient boosting (XGBoost) classifier to classify faults using PV panels’ data streams. Nevertheless, both IB3DM and XGBoost lack the explainability required for the field personnel to understand the fault causes. To achieve this, the IB3DM and XGBoost are used as a base model to build an XAI application using the Local interpretable model-agnostic explanations (LIME) framework. By fusing the XGBoost with XAI, the low amplitude fault signals and their intermittent nature are handled. The edge node implements the XFDS using compact sensors and user interfaces that explain the faults to the field technicians. The proposed XFDS is illustrated using deployment and simulation studies on different PV technologies to demonstrate model accuracy, incipient fault detection, fault explanations, and human augmentation to the system.

Self-supervised pretraining via contrast learning for intelligent incipient fault detection of bearings

Data-driven approaches for prognostic and health management (PHM) increasingly rely on massive historical data, yet annotations are expensive and time-consuming. Learning approaches that utilize semi-labeled or unlabeled data are becoming increasingly popular. In this paper, a self-supervised pre-training via contrast learning (SSPCL) is introduced to learn discriminative representations from unlabeled bearing datasets. Specifically, the SSPCL employs momentum contrast learning (MCL) to investigate the local representation in terms of instance-level discrimination contrast. Further, we propose a specific architecture for SSPCL deployment on bearing vibration signals by presenting several data augmentations for 1D sequences. On this basis, we put forward an incipient fault detection method based on SSPCL for run-to-failure cycle of rolling bearings. This approach transfers the SSPCL pre-trained model to a specific semi-supervised downstream task, effectively utilizing all unlabeled data and relying on only a little priori knowledge. A case study on FEMTO-ST datasets shows that the fine-tuned model is competent for incipient fault detection, outperforming other state-of-the-art methods. Furthermore, a supplemental case on a self-built fault datasets further demonstrate the great potential and superiority of our proposed SSPCL method in PHM.

Incipient fault detection for geological drilling processes using multivariate generalized Gaussian distributions and Kullback–Leibler divergence

Early detection of downhole faults in geological drilling processes can effectively reduce downtime and prevent catastrophic drilling accidents. Considering that the changes of drilling signals in early faults are difficult to observe while the data distribution may have significant deviations, a systematic incipient fault detection method is proposed for drilling processes based on the Multivariate Generalized Gaussian Distributions (MGGDs) and Kullback-Leibler Divergence (KLD). The contributions are twofold: (1) A new dissimilarity index is proposed for incipient fault detection by estimating the KLD between the MGGD of the historical data under the normal condition and that of the real-time data; (2) an adaptive alarm limit design approach is proposed for updating the alarm limits that adapt to the increasing of the drilling depth. Industrial case studies from a real drilling project are presented to demonstrate the effectiveness and practicability of the proposed method.

A Satellite Incipient Fault Detection Method Based on Decomposed Kullback-Leibler Divergence.

Detection of faults at the incipient stage is critical to improving the availability and continuity of satellite services. The application of a local optimum projection vector and the Kullback–Leibler (KL) divergence can improve the detection rate of incipient faults. However, this suffers from the problem of high time complexity. We propose decomposing the KL divergence in the original optimization model and applying the property of the generalized Rayleigh quotient to reduce time complexity. Additionally, we establish two distribution models for subfunctions and to detect the slight anomalous behavior of the mean and covariance. The effectiveness of the proposed method was verified through a numerical simulation case and a real satellite fault case. The results demonstrate the advantages of low computational complexity and high sensitivity to incipient faults.

An incipient fault diagnosis methodology using local Mahalanobis distance: Detection process based on empirical probability density estimation

Incipient fault detection is growing as a challenging and hot topic in industrial and academic areas. It is essential to avoid slight unpermitted changes of a system state that can be aggravated and lead to severe security issues. The main challenge of this problem lies in the fact that tiny changes in the early stage can be blurred with noise and create confusion leading to poor detection performance of typical fault detection methods. To detect subtle deviations buried in noise and cope with the non-Gaussian distributed data condition while keeping with the time series information, a sensitive fault detection methodology combining a specifically tuned Local Mahalanobis Distance (LMD) algorithm and an Empirical Probability Density (EPD) estimation technique is proposed. More specifically, first, a healthy domain estimation is proposed to compute the local Mahalanobis distance with optimally tuned characteristics. To approximate a healthy domain, this work proposes a down-sampling algorithm for anchors generation and a parameter estimation method optimally tuned and based on Generalized Extreme Value distribution (GEV) for the domain margin selection. Subsequently, the EPD cumulative sum technique is applied to the LMD result for improving the detection sensitivity further. The performance analysis based on simulation data shows that our proposal is effective to non-Gaussian data and sensitive for incipient fault detection. A case study based on the Continuous-flow Stirred Tank Reactor (CSTR) further validates the effectiveness of our proposal and highlights its benefit by comparing it with state-of-the-art-based solutions in terms of detection delay, detection probability, false alarm probability, and area under the receiver operating characteristic curve (AUC).

New statistics for simultaneously machine incipient fault detection and monotonic degradation assessment

Machine condition monitoring (MCM) has become an important tool to avoid sudden machine breakdown and gaining more economic profits. Tasks including early fault detection and monotonic degradation assessment are important in MCM. For the incipient fault detection, statistics such as kurtosis, Gini index are widely utilized, but they cannot give an accurately incipient fault detection time, and many fluctuations may exhibit. For the monotonic degradation assessment, root-mean-square are commonly used, however, it is sensitive to energy, and cannot show distinct degradation tendency in an early fault state. Those drawbacks have limited the development of practical MCM algorithms. To address those issues, this paper proposed four parameterized statistics for simultaneously early fault detection and monotonic degradation assessment. The four parameterized statistics can be health indicators and simplify the MCM algorithms, which can be beneficial to the practical MCM applications.

A similarity-based framework for incipient fault detection in underground power cables

Incipient faults are known as the faults mainly occurring on the junctions of underground power cables. Clearing quickly, these faults are difficult to detect by conventional protection equipment, leading to insulation degradation and permanent fault in the power cable over time. In this paper, employing similarity functions, a new method is put forward to distinguish the existing fault from other similar events with both high speed and accuracy. The proposed algorithm consists of a two-stages algorithm including change detection, disturbance identification, and fault detection. The proposed method performs the calculation on the voltage signal that is recorded at the sending end of the line. The cross-correlation function is employed for identifying change detection and determining disturbance time. Afterward, using sinusoidal curve fitting, the algorithm can distinguish the incipient fault from sudden load change, capacitor switching, and harmonic load. The accuracy of the proposed approach is verified by some experimental data and 6272 simulation data obtained from four types of well-known arc models, capacitor switching, and also load variations.

Slow feature analysis-aided detection and diagnosis of incipient faults for running gear systems of high-speed trains

Incipient faults in running gear systems corrupt the overall performance of high-speed trains, increasing the necessity of fault detection and diagnosis whose purpose is to maintain the safe and stable operation of high-speed trains. For this purpose, a novel data-driven method, that utilizes Hellinger distance and slow feature analysis, is proposed in this study. By integrating Hellinger distance into slow feature analysis, a new test statistic is defined for detecting incipient faults in running gear systems. Furthermore, the hidden Markov method is developed for performing reliable fault diagnosis tasks. The salient strengths of the proposed method lie in its satisfactory fault detectability on the one hand and the considerable robustness against high-level noises on the other hand. Finally, the effectiveness of the proposed method is verified through a numerical example and a running gear system of high-speed trains under actual working conditions.

Early Fault Diagnostic System for Rolling Bearing Faults in Wind Turbines

Abstract The operation and maintenance costs of wind farms are always high due to high labor costs and the high replacement cost of parts. Thus, it is of great importance to have real-time monitoring and an early fault diagnostic system to prevent major events, reduce time-based maintenance, and minimize the cost. In this paper, such a two-step system for early stage rolling bearing failures in offshore wind turbines is introduced. First, empirical mode decomposition is applied to minimize the effect of ambient noise. Next, correlation coefficients between a reference signal and test signals are obtained and incipient fault detection is achieved by comparing the results with a threshold value. Through further analysis of the envelope spectrum, sample entropy for selected intrinsic mode functions is obtained, which is further used to train a support vector machine classifier to achieve fault classification and degradation state recognition. The proposed diagnostic approach is verified by experimental tests, and an accuracy of 98% in identifying and classifying rolling bearing failures under various loading conditions is obtained.

Extra Freedom Degrees of Six-Phase Induction Motors Used to Diagnose Stator Inter-Turn Faults

— Multi-phase electrical machines have recently drawn growing attention and usage, among which, 6-phase induction motors (6PIMs) have gained an important part. They are main candidates to be utilized in the oil and gas industries, hot rolling mill drives, pumps, ship propulsion, and traction. Condition monitoring and fault diagnosis are essential technologies, which enable incipient fault detection and removal in order to prevent serious damages to the machines and catastrophic interrupt of them. Among the incipient faults, the stator winding inter-turn fault is a rather frequent one. In recent decades, numerous studies have dealt with this fault in three-phase induction motors, however, a few reports are available concerning 6PIMs. This paper applies the winding function method to model and simulate the 6PIM with inter-turn fault in a stator phase winding. Then, using the extra degrees of freedom in 6PIM, an analytical method is used to define an efficient index for detecting the stator inter-turn faults. This index also identifies the faulty phase of the stator, shows good sensitivity to the fault severity and is almost insensitive to the load torque change. Various simulation studies and corresponding experimental test results confirm effectiveness of the proposed index.

Incipient fault detection benefited from voting fusion strategy on analysis of process variation

Actual industrial processes often have complex system dynamics and operating conditions may vary over time. Thus the constant control limits (CCLs) are loose and not suitable for monitoring the incipient faults. To address this issue, a process variation driven voting fusion strategy (PVVF) method is proposed to detect incipient faults under varying operating conditions. Because the normal control limits cannot be obtained and based on the definition of Q statistic, the proposed method uses the predicted states to generate varying control limits (VCLs) to track the actual system dynamics. Then the process information is produced by measuring the departure between the current actual states and above VCLs. The voting fusion strategy is used to combine process information of each variable to monitor process variations. When tested on a real industrial process, the proposed method is sensitive to the incipient faults under varying operating conditions.

A statistical-based criterion for incipient fault detection in underground power cables established on voltage waveform characteristics

The incipient faults which mainly occur due to the electric arc occurrence in the power cables with insulation defects are hardly detectable by the conventional protective relays, and over time can develop into a permanent fault in the system. Employing Kalman filter, this paper puts forward a method to detect the incipient faults and to discriminate them from other similar incidents in the power system. The proposed method is established on the comparison between the waveform of the measured voltage and fundamental component of the measured voltage, estimated by Kalman filter algorithm in the sending end of the cable during the fault. Employing the difference between the measured and estimated waveforms, the incipient fault detection and discrimination are carried out within two stages. In the first stage, event detection is relegalized by comparing the standard deviation of the obtained error with a certain threshold. The second stage is conducted to find the incipient fault based on the non-attenuating characteristic and quasi-periodic nature of the incipient fault. The feasibility of the proposed method is verified through computer simulation using four different electric arc models and also the acquired experimental data from real incipient faults.

A Multinomial DGA Classifier for Incipient Fault Detection in Oil-Impregnated Power Transformers

This study investigates the use of machine-learning approaches to interpret Dissolved Gas Analysis (DGA) data to find incipient faults early in oil-impregnated transformers. Transformers are critical pieces of equipment in transmitting and distributing electrical energy. The failure of a single unit disturbs a huge number of consumers and suppresses economic activities in the vicinity. Because of this, it is important that power utility companies accord high priority to condition monitoring of critical assets. The analysis of dissolved gases is a technique popularly used for monitoring the condition of transformers dipped in oil. The interpretation of DGA data is however inconclusive as far as the determination of incipient faults is concerned and depends largely on the expertise of technical personnel. To have a coherent, accurate, and clear interpretation of DGA, this study proposes a novel multinomial classification model christened KosaNet that is based on decision trees. Actual DGA data with 2912 entries was used to compute the performance of KosaNet against other algorithms with multiclass classification ability namely the decision tree, k-NN, Random Forest, Naïve Bayes, and Gradient Boost. Investigative results show that KosaNet demonstrated an improved DGA classification ability particularly when classifying multinomial data.

Adaptive Neuro-Fuzzy Inference System (ANFIS) based modelling of incipient steam generator tube rupture diagnosis

Steam generators are one of the primary barriers in pressurized water reactors that play a critical safety role in separating the non-radioactive and the radioactive sides of nuclear power plants. The occurrence of primary to secondary leakage as a result of incipient faults on the tubing could cause a through-wall tube rupture which may lead to the release of undesirable levels of radioactive products into the environment. The efficiency of existing monitoring techniques and models in the detection of incipient faults and the diagnosis of steam generator tubes is a key challenge facing the nuclear industry. This paper bridges this gap by applying an Adaptive Neuro-Fuzzy Inference System (ANFIS) approach that not only has a high sensitivity in the detection of incipient faults but also has superior local interpolation capabilities and high prediction of time series. Data-driven ANFIS models were used to predict uncertain critical parameters that were sensitive to the fault signatures as well as to estimate the severity of ruptures. To evaluate the optimized models, steam generator tube rupture (SGTR) scenarios were simulated in Qinshan I NPP (CNP300 PWR) reactor coolant system using a thermal-hydraulic best estimate code, RELAP5/SCDAP Mod4.0. The parameters estimation efficiencies were evaluated statistically using the mean absolute percentage error (MAPE), coefficient of determination (R2), relative standard deviation (RSD) and visually by residual plots.

Detection of incipient faults in oil-immersed power transformers based on dissolved gas analysis: Case studies

Power transformers are one of the most crucial as well as costly equipment in power grids. Nowadays, a set of factors like the growing load, expanding power consumption, and market liberation affect on the transformer reliability and operational lifetime. At present, numerous approaches are on hand to diagnosis and evaluate their operational conditions. Dissolved gas analysis (DGA) in the insulation oil is an efficient diagnostic scheme which can offer an essential source for transformer failure diagnosis. In this paper, five existing approaches of utilizing dissolved gas data to identify fault kind in power transformers oil are employed. The considered methods are: key gas technique, IEC 60599 technique, Doernenburg ratio scheme, Duval triangle framework, as well as Rogers ratio scheme. These fault diagnosis methods are used to interpret and validate the dissolved gas analysis (DGA) results for four case studies. The experiments are performed on four 15.75 kV/400 kV, 3-Phase, 250 MVA power transformers in a thermal power plant. The obtained DGA results reveal that all considered DGA methods have effectively detected the fault type for each considered case except Duval triangle method which has failed to discover the fault type for two cases.

Periodicity-enhanced sparse representation for rolling bearing incipient fault detection

Incipient fault detection of rolling bearings is a challenging task since the weak fault features are disturbed by heavy background noise. This paper develops a periodicity-enhanced sparse representation method to address this issue. Firstly, periodicity-enhanced basis pursuit denoising (PBPD) is proposed by the theoretical derivation. Fault proportion is defined to quantify the single fault severity of sparse signals, then a periodicity-decision criterion for determining the optimal potential fault period is designed to periodically filter the last sparse signal. Secondly, the suitable linear transformation for PBPD is investigated in comparison and maximal overlapping discrete wavelet packet transform is adopted eventually. Thirdly, adaptive selection strategies are developed for the key parameters of PBPD. Simulations and experimental verifications demonstrate PBPD’s excellent performance in rolling bearing incipient fault detection.

Data-Driven Approaches for Diagnosis of Incipient Faults in Cutting Arms of the Roadheader

Incipient fault detection and identification (IFDI) of cutting arms is a crucial guarantee for the smooth operation of a roadheader. However, the shortage of fault samples restricts the application of the fault diagnosis technique, and the data analysis tools should be optimized efficiently. In this study, four machine learning tools (the back-propagation neural network based on genetic algorithm optimization, the naive Bayes based on genetic algorithm optimization, the support vector machines based on particle swarm optimization, and the support vector machines based on dynamic cuckoo) are applied to address the challenge in the IFDI of cutting arms. The commonly measured current and vibration data cutting arms are used in the IFDI. The experimental results show that the support vector machines based on dynamic cuckoo outperform the other methods. Besides, the performance of the four methods under different operating conditions is compared. The fault cause of cutting arms of the roadheader is analyzed and the design improvement scheme for cutting arms is provided. This study provides a reference for improving the fault diagnosis of the roadheader.

A Novel Parameter-Adaptive VMD Method Based on Grey Wolf Optimization with Minimum Average Mutual Information for Incipient Fault Detection

Recently, variational mode decomposition (VMD) has attracted wide attention on mechanical vibration signal analysis. However, there are still some dilemmas in the application of VMD, such as the determination of the number of mode decomposition K and quadratic penalty term α. In order to acquire appropriate parameters of VMD, an improved parameter-adaptive VMD method based on grey wolf optimizer (GWO) is developed by taking the minimum average mutual information into consideration (GWOMI). Firstly, the parameters (K, α) are adaptively determined through GWOMI. Then, the vibration signal is decomposed by the developed method and effective modes are extracted according to the maximum kurtosis. Finally, the extracted modes are processed by Hilbert envelope analysis to acquire the incipient fault features. With the simulation and experimental analysis, it is clearly found that the developed method is effective and performs better than some existing ones.