Fault Probability Research Articles

Imbalance fault diagnosis under long-tailed distribution: Challenges, solutions and prospects

Intelligent fault diagnosis based on deep learning has yielded remarkable progress for its strong feature representation capability in recent years. Nevertheless, in engineering scenarios, machines often operate under normal condition and the probability of various faults is also different, which leads to the extreme class imbalance and long-tailed distribution between different health conditions. Long-tailed distribution usually degrades the performance of the model, since the model tends to focus on dominant classes and exhibits a poor performance on tail classes. To address this problem, numerous researches have been carried out and fruitful achievements have been achieved in recent years. However, a comprehensive summary of the latest achievements is still lacking. In addition, existing studies solely concentrate on the issue of class imbalance, and lack of analysis and future research directions of long-tailed problem. Hence, this paper specifically defines the long-tailed distribution and discusses the relations between long-tailed problem and class imbalance problem. Then, research achievements on class imbalance problems are reviewed. Specifically, we divide the existing research achievements into three categories: class rebalance, information augmentation and module design. Finally, this paper summarizes the challenges of the long-tailed problem in fault diagnosis and provides some future research directions.

Reliability Analysis in the Train Communication Ethernet Based on Multiple Methods

Train communication network (TCN) is an important guarantee for the reliable and safe operation of high‐speed rail train. The higher reliability of the TCN is, the safer operation of high‐speed rail train will be, and the cost of overhaul and maintenance will be greatly reduced. The 250 km/h high‐speed rail TCN being developed in China is the first time that the train communication Ethernet (TCE) is used as train control network. Therefore, it is of great practical significance to analyze the reliability of TCE and improve the reliability of TCN design and vehicle design. Dynamic fault tree (DFT) and fault Petri nets (FPN) are both important methods for system reliability analysis. Among them, DFT is most widely used in the field of engineering analysis. The FPN method also has its unique characteristics and advantages, which can describe fault propagation analysis and dynamic transition process in practical application. In this paper, TCE's structure, working principle, and fault mode are analyzed to sort out the relationship between complex fault events and establish a DFT with TCE failure as the top event. DFT and FPN methods were combined to better express the dynamic propagation of TCE fault and carry out the fault probability evaluation and system reliability based on the known prior failure rate. The analysis results show that improving the reliability of specified network segments of TCE is of great help to improve the reliability of the high‐speed rail train. © 2022 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

NUMERICAL SOLUTION FOR THE DISTRIBUTION OF THE ELECTROMAGNETIC AND THERMAL FIELDS OF AN AIR-CORE TRANSFORMER

The article presents a numerical solution for the distribution of the electromagnetic and thermal fi elds of an air-core transformer using the fi nite element method. The distribution of the magnetic fi eld is determined for the dynamic steady state and magnetic nonlinear core characteristics. The model presented facilitates the establishment of criteria for optimizing transformer operation under various load conditions, environments as well as in the case of failures. Thus, the transformer can operate at maximum capacity while, at the same time, the probability of faults due to overheating is reduced to a minimum.

Mobile robot communication fault diagnosis method based on swarm intelligence algorithm

The current communication fault diagnosis methods mainly focus on the classifier with fault probability, which often leads to the low diagnosis efficiency. In order to overcome the above problems, mobile robot communication fault diagnosis method based on swarm intelligence algorithm is proposed in the paper. Firstly, the abnormal data is extracted after the analysis of the communication data with Kalman filter. Secondly, the supporting decision model was designed to standardize the communication exchange process and locate the fault range. Futher more, the ant colony optimization algorithm combined with particle swarm optimization was used to locate the fault area, and the multi-model hybrid method was adopted to comprehensively judge the communication fault. The different interference ratio was used in the experiment to test the performance of the proposed algorithm compared with the SVM and Bayesian model. Finally, the experiment results show the validity of mobile robort communication fault diagnosis based on swarm intelligence algorithm.

Repetitive Signal Generator Based Detection Method for Faulty Arm With Switch Open-Circuit Failure in High-Voltage Modular Multilevel Converters

Modular multilevel converters (MMCs) in high- voltage applications contain large numbers of power switches, which increases the fault probability inside MMCs. This paper proposes a detection method for the faulty arm containing switch open-circuit failures in high-voltage MMCs to improve the reliable operations of MMCs. The proposed method uses the periodicity of the deviation between the maximum and minimum capacitor voltages to provide the detection reference. To yield a reliable reference, the repetitive signal generator is employed. Then, the fault is identified by the difference between the referenced and measured capacitor voltage deviations. An auxiliary criterion is also designated to avoid the false detection caused by the mutations in MMCs. The proposed method can reduce the difficulty in the selection of thresholds. Moreover, the proposed method is robust in high-voltage MMCs with inferior capacitor voltage balancing performance, which ensures detection rapidity and accuracy. Finally, the feasibility of the method is verified by simulation results in Matlab/Simulink.

Internal fault probability-based time domain differential protection applied to transmission lines connecting battery energy storage stations

Battery energy storage station (BESS) presents disparate fault current characteristics in charge and discharge states. Classic and recently proposed differential protection algorithms, including phasor-based and time-domain algorithms, are heavily influenced by operating state of BESS. Problems such as low sensitivity and ineffectiveness may arise, especially in charge state. In response to this issue, this paper applies Bayes' Theorem to calculate the probability of internal fault occurrence, named as internal fault probability. In this article, fault current characteristics of BESS under different operating states are described, and then the causes of performance problems of classic and recently proposed differential protection algorithms are analyzed. Subsequently, virtual operating current vector is developed, and the difference of virtual operating current vector in time-domain waveform under various conditions is described. On this basis, internal fault probability-based differential protection (IFP-DP) is presented. Besides, its operating performances under internal and external faults are analyzed. The advantageous performance of IFP-DP is fully validated by changing fault location, type, resistance, operating state and considering outliers, current transformer (CT) saturation, and loss of data. Proposed IFP-DP exhibits gratifying performance in operating time, sensitivity and security in the presence of BESS.

Fabrication of structural and compositional heterostructured CoCrNi–CoCrFeMnNi multi-material using direct energy deposition additive manufacturing

Metal additive manufacturing (MAM) is an emerging processing technology that can efficiently produce various metallic materials. Aimed at controlling material properties to suit functional requirements, there has been a considerable demand in recent years to manufacture multi-materials with locally different chemical compositions from one material composition to another. However, due to the technical difficulties of MAM, MAM-processed multi-materials have a fundamental limitation of high defect probability near the interface. In this study, to overcome the limitations of MAM-processed multi-materials, we proposed a new strategy for multi-material fabrication by depositing a single layer of CoCrFeMnNi high-entropy alloy (HEA) on a cold-rolled CoCrNi medium-entropy alloy (MEA) substrate using direct energy deposition. The prepared CoCrNi–CoCrFeMnNi multi-material exhibits both structural and compositional heterogeneity, and no defects were observed. In addition, it has a superior strength-ductility combination compared to the individual properties of homogeneous CoCrNi MEA and CoCrFeMnNi HEA owing to a synergetic strengthening effect with the heterogeneous microstructure.

Assessment of strong earthquake risk in the Chinese mainland from 2021 to 2030

The long-term earthquake prediction from 2021 to 2030 is carried out by researching the active tectonic block boundary zones in the Chinese mainland. Based on the strong earthquake recurrence model, the cumulative probability of each target fault in the next 10 years is given by the recurrence period and elapsed time of each fault, which are adopted from relevant studies such as seismological geology, geodesy, and historical earthquake records. Based on the long-term predictions of large earthquakes throughout the world, this paper proposes a comprehensive judgment scheme based on the fault segments with the seismic gap, motion strongly locked, sparse small-moderate earthquakes, and apparent Coulomb stress increase. This paper presents a comprehensive analysis of the relative risk for strong earthquakes that may occur in the coming 10 years on the major faults in the active tectonic block boundary zones in the Chinese mainland. The present loading rate of each fault is first constrained by geodetic observations; the cumulative displacement of each fault is then estimated by the elapsed time since the most recent strong earthquake.

High-resolution seismic faults interpretation based on adversarial neural networks with a regularization technique

Geologic fault detection at high precision and resolution is the key for fine structure and reservoir modeling. Previous studies using neural networks for fault segmentation mainly focus on the local features of the targets and train the networks using synthetic data sets. To increase the fault segmentation resolution only using a limited amount of seismic field data, we develop an adversarial neural network architecture for high-resolution identification of faults (FaultAdvNet) taking advantage of global feature fusion. The architecture consists of (1) a light-weight segmentation module (approximately 0.49 M parameters), (2) a feature fusion module considering reflectors of faults and surrounding stratums, and (3) a discriminator module acting as a regularization term. Case studies using seismic field data from the Gulf of Mexico show an overwhelming performance improvement of the FaultAdvNet when compared with other fault detection methods. The FaultAdvNet picks all of the faults with sufficiently high confidence and low prediction risk. The predicted faults of the FaultAdvNet have good continuity and show clear boundary with fault probability values mainly ranging from 0.95 to 1. Saliency analysis also suggests that the FaultAdvNet can focus on the target at a sufficiently higher resolution (dozens of meters). Functionality experiments verify the mechanisms of the feature fusion module and the discriminator module in FaultAdvNet. We consider that a neural network (such as the discriminator) can serve as a data-driven regularization term to constrain the target network (the segmentation network) efficiently, especially given a limited amount of seismic data.

Online Reliability-Enhanced Virtual Network Services Provisioning in Fault-Prone Mobile Edge Cloud

Fault-Prone Mobile Edge Cloud (FP-MEC) is a new type of distributed network composed of mobile edge computing and network function virtualization, where virtual network services can be provided in the form of service function chains (SFCs) that are a sequence of virtual network functions (VNFs) on-demand deployed on resource-limited edge servers. FP-MEC has a characteristic that the fault probability of each VNF is dynamic and fluctuates with time and workloads, making SFCs temporarily unreliable. To increase the reliabilities, redundant Backup VNFs (BVNFs) need to be deployed near the VNFs and activated when they experience faults. Different mobile users would request different SFCs with reliability and service time demands to process their data. However, workloads of VNFs are dynamic and unpredictable in FP-MEC due to random arrival of user requests. How to optimally deploy VNFs and corresponding BVNFs on a set of edge servers to form expected SFCs that have higher reliabilities than user demand values, meanwhile throughput of receiving requests is maximized while receiving cost is minimized in real-time, is a challenging problem. The receiving cost is composed of deployment cost of instantiating VNFs and BVNFs, and communication cost of routing data among users, VNFs and BVNFs. In this paper, the long-term provisioning problem is first formulated as an integer linear program and proved to be NP-hard. Then, it is discretized into a sequence of one-slot optimization problems to handle practical time-varying fault probability, where a set of SFC requests are given at each time slot, and receiving or rejecting decisions are executed immediately without any future information. Finally, an online approximation scheme with a constant approximation ratio is proposed to solve the one-slot problems in polynomial time. Theoretical analyses and experiments based on real network topology of CERNET in China demonstrate that the scheme is promising compared to existing works.

Quantitative analysis of diffuse electron scattering in the lithium-ion battery cathode material Li1.2Ni0.13Mn0.54Co0.13O2

In contrast to perfectly periodic crystals, materials with short-range order produce diffraction patterns that contain both Bragg reflections and diffuse scattering. To understand the influence of short-range order on material properties, current research focuses increasingly on the analysis of diffuse scattering. This article verifies the possibility to refine the short-range order parameters in submicrometre-sized crystals from diffuse scattering in single-crystal electron diffraction data. The approach was demonstrated on Li1.2Ni0.13Mn0.54Co0.13O2, which is a state-of-the-art cathode material for lithium-ion batteries. The intensity distribution of the 1D diffuse scattering in the electron diffraction patterns of Li1.2Ni0.13Mn0.54Co0.13O2 depends on the number of stacking faults and twins in the crystal. A model of the disorder in Li1.2Ni0.13Mn0.54Co0.13O2 was developed and both the stacking fault probability and the percentage of the different twins in the crystal were refined using an evolutionary algorithm in DISCUS. The approach was applied on reciprocal space sections reconstructed from 3D electron diffraction data since they exhibit less dynamical effects compared with in-zone electron diffraction patterns. A good agreement was achieved between the calculated and the experimental intensity distribution of the diffuse scattering. The short-range order parameters in submicrometre-sized crystals can thus successfully be refined from the diffuse scattering in single-crystal electron diffraction data using an evolutionary algorithm in DISCUS.

Research on the output coordinated strategy of a DPFC considering device fault probability

A distributed power flow controller (DPFC) can control the line power flow in a flexible and fast way to meet the requirements of a new power system. However, the output of DPFC sub-units is generally distributed by the even distribution method or proportional distribution method at present. The internal health status and output capacity of the device are not taken into account, which affects the efficiency and service life of the device. In this article, a reliability description method of DPFC based on fault probability is proposed. The coordination strategy of output voltage capability and the number of input sub-units is proposed, and the process of the proposed coordinated strategy for device output is presented. The simulation results show that the proposed coordinated output strategy can address the issue of rapid reliability decline caused by the long-term work of some sub-units and improve the overall reliability of the DPFC.

Augmented deep learning workflow for robust fault prediction over multiple tectonic regimes

Abstract Over the past decade, deep learning frameworks such as convolutional neural networks (CNNs) have made major research inroads in upstream oil and gas, with applications in seismic processing/imaging, velocity model building, petrophysics, geological seismic interpretation, all the way to development, production and supply chain logistics. CNN fault prediction centers around the idea of image edge detection and, for improved prediction results, three data-driven steps are recommended. First, pre-condition the seismic data to increase signal-to-noise ratio as much as possible: iterative dip-steered median filtering and principal component filtering are adopted to further improve signal-to-noise ratio and sharpness for edge detection. Second, a fault probability volume obtained through deep learning (DL) -based fault detection using a U-Net architecture, taking synthetic seismic models as samples that can improve DL-based fault prediction in lieu of readily available labeled fault sets that may be prohibitively time-consuming to generate. Finally, edge enhancement is performed on the inference results to improve precision and fault continuity. A comparative analysis between related edge enhancement technologies is also presented. Results for three different faulting modes (normal, reverse and strike-slip) over three real seismic field data sets from China demonstrate the robustness of the proposed workflow.

An Augmented Reality-Assisted Prognostics and Health Management System Based on Deep Learning for IoT-Enabled Manufacturing.

With increasingly advanced Internet of Things (IoT) technology, the composition of workshop equipment has become more and more complex. Based on this, the rate of system performance degradation and the probability of fault have both increased. Owing to this, not only has the difficulty of constructing the remaining useful life (RUL) model increased but also the improvement in speed of maintenance personnel cannot keep up with the speed of equipment replacement. Therefore, an augmented reality (AR)-assisted prognostics and health management system based on deep learning for IoT-enabled manufacturing is proposed in this paper. Firstly, the feature extraction model based on Convolutional Neural Network-Particle Swarm Optimization (PSO-CNN) is proposed with the purpose of excavating the internal associations in large amounts of production data. Based on this, the high-accuracy RUL prediction is accomplished by Gate Recurrent Unit (GRU)-attention, which can capture the long-term and short-term dependencies of time series and successfully solve the gradient disappearance problem of RNN. Moreover, more attention will be paid to important content with the help of the attention mechanism. Additionally, high-efficiency maintenance guidance and visible instructions can be accomplished by AR. On top of this, the remote expert can offer help when maintenance personnel encounters tough problems. Finally, a real case was implemented in a typical IoT-enabled workshop, which validated the effectiveness of the proposed approach.

Exploring factors affecting the performance of deep learning in seismic fault attribute computation

Recently, the computation of seismic fault attribute that may be significant in seismic interpretation is that seismic fault detection is treated as an image segmentation problem using different deep-learning (DL) architectures. To do this, researchers have concentrated on applying cutting-edge DL architectures in computing seismic fault attributes. To explore the factors that may affect the accuracy of seismic fault attribute, we compare the computed fault probability using DL architectures under different scenarios. The designed scenarios aim to highlight the leading factors that may affect the accuracy and resolution of seismic image segmentation. The discussed factors include the dimension and size of training data, training data preparation, ensemble learning, and batch size in DL. The proposed comparisons are applied to one marine seismic survey from New Zealand and one land seismic survey from China. The results demonstrate that properly preparing training data is far more important than choosing a cutting-edge DL architecture in computing seismic fault attribute. We also propose a practical workflow that can include real seismic data and corresponding interpreted fault sticks in training data for a specific seismic survey.

Enhancing carrier transport in flexible CZTSSe solar cells via doping Li strategy

The passivation of non-radiative states and inhibition of band tailings are desirable for improving the open-circuit voltage (Voc) of CZTSSe thin-film solar cells. Recently, alkali metal doping has been investigated to passivate defects in CZTSSe films. Herein, we investigate Li doping effects by applying LiOH into CZTSSe precursor solutions, and verify that carrier transport is enhanced in the CZTSSe solar cells. Systematic characterizations demonstrate that Li doping can effectively passivate non-radiative recombination centers and reduce band tailings of the CZTSSe films, leading to the decrease in total defect density and the increase in separation distance between donor and acceptor. Fewer free carriers are trapped in the band tail states, which speeds up carrier transport and reduces the probability of deep-level defects capturing carriers. The charge recombination lifetime is about twice as long as that of the undoped CZTSSe device, implying the heterojunction interface recombination is also inhibited. Besides, Li doping can increase carrier concentration and enhance build-in voltage, leading to a better carrier collection. By adjusting the Li/(Li + Cu) ratio to 18%, the solar cell efficiency is increased significantly to 9.68% with the fill factor (FF) of 65.94%, which is the highest FF reported so far for the flexible CZTSSe solar cells. The increased efficiency is mainly attributed to the reduction of Voc deficit and the improved CZTSSe/CdS junction quality. These results open up a simple route to passivate non-radiative states and reduce the band tailings of the CZTSSe films and improve the efficiency of the flexible CZTSSe solar cells.

Following Cu Microstructure Evolution in CuZnO/Al2O3(−Cs) Catalysts During Activation in H2 using in situ XRD and XRD‐CT

AbstractUnderstanding how the microstructure of the active Cu0 component in the commercially applicable Cu/ZnO/Al2O3(−Cs2O) low‐temperature water‐gas shift catalyst evolves under various H2 partial pressures in the presence/absence of a Cs promoter during thermal activation has been investigated. Time‐resolved XRD and spatially‐resolved XRD‐CT data were measured as a function of H2 concentration along a packed bed reactor to elucidate the importance of the zincite support and the effect of the promoter on Cu sintering mechanisms, dislocation character and stacking fault probability. The rate of Cu reduction showed a dependency on [Cs], [H2] and bed height; lower [Cs] and higher [H2] led to a greater rate of metallic copper nanoparticle formation. A deeper analysis of the XRD line profiles allowed for determining a greater edge character to the dislocations and subsequent stacking fault probability was also observed to depend on higher [H2], smaller Cu0 (and ZnO) crystallite sizes, increased [ZnO] (30 wt.%, sCZA) and lower temperature. The intrinsic activity of Cu/ZnO/Al2O3 methanol synthesis catalysts has been intimately linked to the anisotropic behaviour of copper, and thus the presence of lattice defects; to the best knowledge of the authors, this study is the first instance in which this type of analysis has been applied to LT‐WGS catalysts.

Assessing the operational design domain of lane support system for automated vehicles in different weather and road conditions

With the growing rate of automated vehicles (AVs) at the lower level of automation, the experimental tests are also in progress with vehicles at higher levels. In the absence of extended digital infrastructures and deployment of level 5 full automated vehicles, the physical infrastructure is required to maintain a fundamental role to enable their introduction in public roads. This paper focuses on lane support system (LSS) whose operational design domain (ODD) is strongly connected to the road characteristics and conditions. An experimental test was carried out with a state of the art, and LSS and advanced technologies were used for road monitoring on different roads under various environmental conditions including dry, wet pavements and rain. We applied the generalized estimation equation for logistic regression to account within-cluster homogeneity which is induced by repeated measures on the same road sections. Statistical models allow the identification of variables that are significant for the LSS fault probability among various effects of road features including marking, pavement distress, weather conditions, horizontal curvature, and cross section. Results pointed out the relevance of the wet retro-reflection of marking (RLw) and the horizontal curvature in the definition of ODD for LSS. Threshold values have been proposed for the tested LSS. Wet pavement doesn't affect the LSS performance when compared to the dry condition. Rain was shown to be critical even with very good road characteristics. • Open on-road LSS test under various weather conditions such as dry, wet and rain. • Generalized estimating equation to highlight correlation among observations. • Average 3.9% fault probability of LSS with high variability in two-lane rural roads. • Comparatively rain increases the fault probability by a factor of 2.75 than dry weather condition. • Marking RLw and curvature 1/ R are the most relevant road factors.

Damping performance and martensitic transformation of rapidly solidified Fe-17%Mn alloy

The rapid solidification of Fe-17%Mn alloy was performed to investigate the influence of cooling rate on its damping performance and martensitic transformation mechanism. A proper heat treatment was also carried out to clarify its coupled effects with rapid solidification. The stacking fault probability and martensitic transformation temperature were determined to demonstrate their relationship with the cooling rate and the heat treatment process. With the increase of cooling rate, the volume fraction of ε-martensite increased and the stacking fault probability of ε-martensite was enhanced. The formation of ε-martensite phase was remarkable for the increase of damping capacity and microhardness. It was found that rapid solidification was beneficial for the formation of ε-martensite and the improvement of damping capacity. This effect can be facilitated by the incorporation of the heat treatment process.

Study on pressure spread characteristics of arc fault in critical area of UHV oil-filled equipment

Large-scale oil-filled equipment such as AC transformers, converter transformers and reactors are important equipment in UHV transmission systems. However, the sudden pressure rise caused by arc faults will cause explosions of oil-filled equipment, which seriously threatens the safe and stable operation of the power system. The tap-changer and ascending flange are the critical areas of UHV oil-filled equipment, with a higher probability of internal arc faults than the rest of the equipment. In order to clarify the pressure spread characteristics of tap-changers and ascending flanges during arc fault, the pressure distribution characteristics were analyzed based on finite element simulation. Based on the gas-liquid two-phase flow model and the compressible model, the generation and transmission of pressure waves were simulated, and the pressure distribution law and evolution process at different times were analyzed after arc energy was loaded at different fault locations. The results show that the pressure near the fault point is the highest, and the pressure peak decreases significantly with the increase of distance. The time to reach the peak pressure of different parts is obviously different, which shows the feasible to judge the arc fault location based on the pressure distribution characteristics of multiple monitoring points.