Fault Diagnosis Based on Fuzzy Support Vector Machine with Parameter Tuning and Feature Selection

This study describes a classification methodology based on support vector machines(SVMs),which offer superior classification performance for fault diagnosis in chemical process engineering.The method incorporates an efficient parameter tuning procedure(based on minimization of radius/margin bound for SVM's leave-one-out errors)into a multi-class classification strategy using a fuzzy decision factor,which is named fuzzy support vector machine(FSVM).The datasets generated from the Tennessee Eastman process(TEP)simulator were used to evaluate the clas-sification performance.To decrease the negative influence of the auto-correlated and irrelevant variables,a key vari-able identification procedure using recursive feature elimination,based on the SVM is implemented,with time lags incorporated,before every classifier is trained,and the number of relatively important variables to every classifier is basically determined by 10-fold cross-validation.Performance comparisons are implemented among several kinds of multi-class decision machines,by which the effectiveness of the proposed approach is proved.

Model Parameter Transfer for Gear Fault Diagnosis under Varying Working Conditions

Gear fault diagnosis technologies have received rapid development and been effectively implemented in many engineering applications.However,the various working conditions would degrade the diagnostic performance and make gear fault diagnosis(GFD)more and more challenging.In this paper,a novel model parameter transfer(NMPT)is proposed to boost the performance of GFD under varying working conditions.Based on the previous transfer strategy that controls empirical risk of source domain,this method further integrates the superiorities of multi-task learning with the idea of transfer learning(TL)to acquire transferable knowledge by minimizing the discrepancies of separating hyperplanes between one specific working condition(target domain)and another(source domain),and then transferring both commonality and specialty parameters over tasks to make use of source domain samples to assist target GFD task when sufficient labeled samples from target domain are unavailable.For NMPT implementation,insufficient target domain features and abundant source domain features with supervised information are fed into NMPT model to train a robust classifier for target GFD task.Related experiments prove that NMPT is expected to be a valuable technology to boost practical GFD performance under various working conditions.The proposed methods provides a transfer learning-based framework to handle the problem of insufficient training samples in target task caused by variable operation conditions.

A Statistical Parameter Analysis and SVM Based Fault Diagnosis Strategy for Dynamically Tuned Gyroscopes

Gyro's fault diagnosis plays a critical role in inertia navigation systems for higher reliability and precision. A new fault diagnosis strategy based on the statistical parameter analysis (SPA) and support vector machine (SVM) classification model was proposed for dynamically tuned gyroscopes (DTG). The SPA, a kind of time domain analysis approach, was introduced to compute a set of statistical parameters of vibration signal as the state features of DTG, with which the SVM model, a novel learning machine based on statistical learning theory (SLT), was applied and constructed to train and identify the working state of DTG. The experimental results verify that the proposed diagnostic strategy can simply and effectively extract the state features of DTG, and it outperforms the radial-basis function (RBF) neural network based diagnostic method and can more reliably and accurately diagnose the working state of DTG.

Optimal State Estimation and Fault Diagnosis for a Class of Nonlinear Systems

This study proposes a scheme for state estimation and,consequently,fault diagnosis in nonlinear systems.Initially,an optimal nonlinear observer is designed for nonlinear systems subject to an actuator or plant fault.By utilizing Lyapunov's direct method,the observer is proved to be optimal with respect to a performance function,including the magnitude of the observer gain and the convergence time.The observer gain is obtained by using approximation of Hamilton-Jacobi-Bellman(HJB)equation.The approximation is determined via an online trained neural network(NN).Next a class of affine nonlinear systems is considered which is subject to unknown disturbances in addition to fault signals.In this case,for each fault the original system is transformed to a new form in which the proposed optimal observer can be applied for state estimation and fault detection and isolation(FDI).Simulation results of a singlelink flexible joint robot(SLFJR)electric drive system show the effectiveness of the proposed methodology.

Method for Analog Circuit Soft-Fault Diagnosis and Parameter Identification Based on Indictor of Phase Deviation and Spectral Radius

The soft fault induced by parameter variation is one of the most challenging problems in the domain of fault diagnosis for analog circuits.A new fault location and parameter prediction approach for soft-faults diagnosis in analog circuits is presented in this paper.The proposed method extracts the original signals from the output terminals of the circuits under test（CUT） by a data acquisition board.Firstly,the phase deviation value between fault-free and faulty conditions is obtained by fitting the sampling sequence with a sine curve.Secondly,the sampling sequence is organized into a square matrix and the spectral radius of this matrix is obtained.Thirdly,the smallest error of the spectral radius and the corresponding component value are obtained through comparing the spectral radius and phase deviation value with the trend curves of them,respectively,which are calculated from the simulation data.Finally,the fault location is completed by using the smallest error,and the corresponding component value is the parameter identification result.Both simulated and experimental results show the effectiveness of the proposed approach.It is particularly suitable for the fault location and parameter identification for analog integrated circuits.

FAULT DIAGNOSIS FOR ANALOG CIRCUITS WITH TOLERANCE—MINIMUM TOLERANCE ESTIMATION ALGORITHM

Based on the influence of circuit element tolerances to the k-fault diagnosis, a method of fault diagnosis is presented which is called minimum tolerance estimation algorithm and has clear physical meaning. Using this’method, an effective estimation of the equivalent fault sources can be obtained with less computing time. It is especially worthwhile to point out that an adaptive sub-optimum algorithm, which comes from the above method, requires even less computing-labor and is particularly suitable to more complicated circuits as well as real-time fault location.

Fault Diagnosis of Nonlinear Systems Based on Hybrid PSOSA Optimization Algorithm

Fault diagnosis of nonlinear systems is of great importance in theory and practice, and the parameter estimation method is an effective strategy. Based on the framework of moving horizon estimation, fault parameters are identified by a proposed intelligent optimization algorithm called PSOSA, which could avoid premature convergence of standard particle swarm optimization （PSO） by introducing the probabilistic jumping property of simulated annealing （SA）. Simulations on a three-tank system show the effectiveness of this optimization based fault diagnosis strategy.

Fault diagnosis based on high-gain observer with an update law for a class of nonlinear systems

In this paper,by using the well-known high-gain observer design,an update law for the gain and an adaptive estimation of parameters,a new method of fault diagnosis for a class of nonlinear systems is presented.Without resort to any transformation for the parameters,the estimation errors of the states and the parameters are guaranteed to be globally exponentially convergent by a persistent excitation condition.Compared to the existing results,it can be applied to nonlinear systems with nonlinear terms admitting an incremental rate depending on the measured output.A case study further verifies the validity of the proposed research.

Fault Diagnosis of Nonlinear Systems Using Structured Augmented State Models

This paper presents an internal model approach for modeling and diagnostic functionality design for nonlinear systems operating subject to single- and multiple-faults. We therefore provide the framework of structured augmented state models. Fault characteristics are considered to be generated by dynamical exosystems that are switched via equality constraints to overcome the augmented state observability limiting the number of diagnosable faults. Based on the proposed model, the fault diagnosis problem is specified as an optimal hybrid augmented state estimation problem. Sub-optimal solutions are motivated and exemplified for the fault diagnosis of the well-known three-tank benchmark. As the considered class of fault diagnosis problems is large, the suggested approach is not only of theoretical interest but also of high practical relevance.

Adaptive Time-Frequency Distribution Based on Time-Varying Autoregressive and Its Application to Machine Fault Diagnosis

The time-varying autoregressive （TVAR） modeling of a non-stationary signal is studied. In the proposed method, time-varying parametric identification of a non-stationary signal can be translated into a linear time-invariant problem by introducing a set of basic functions. Then, the parameters are estimated by using a recursive least square algorithm with a forgetting factor and an adaptive time-frequency distribution is achieved. The simulation results show that the proposed approach is superior to the short-time Fourier transform and Wigner distribution. And finally, the proposed method is applied to the fault diagnosis of a bearing , and the experiment result shows that the proposed method is effective in feature extraction.

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.

On-line Dynamic Model Correction Based Fault Diagnosis in Chemical Processes

<Abstract>A novel fault detection and diagnosis method was proposed, using dynamic simulation to monitor chemical process and identify faults when large tracking deviations occur. It aims at parameter failures, and the parameters are updated via on-line correction. As it can predict the trend of process and determine the existence of malfunctions simultaneously, this method does not need to design problem-specific observer to estimate unmeasured state variables. Application of the proposed method is presented on one water tank and one aromatization reactor, and the results are compared with those from the traditional method.

Key Issues for Modelling, Operation, Management and Diagnosis of Lithium Batteries: Current States and Prospects

1 Introduction.With the continuous growth of the global population,the energy demand continues to increase.However,due to the dominance of fossil fuels in global energy and fossil fuels are non-renewable,it has led to the global energy crisis[1].Besides,the use of fossil fuels will generate a mass of air pollutants(e.g.,carbon dioxide,sulfur dioxide,etc.),which will cause serious environmental pollution,climate change[2],etc.To resolve the aforementioned issues,countries around the world have implemented a variety of measures hoping to fundamentally adjust the global energy structure and achieve sustainable development.Thereinto,“Paris Agreement”reached in 2015 under the framework of“United Nations Framework Convention on Climate Change”aims to control the increase in the average temperature of the globe to within 2°C below preindustrial levels,and thereafter to peak global greenhouse gas emissions as soon as possible,continuously decreasing thereafter[3].United Kingdom plans to reduce the average exhaust emissions of“new cars”to approximately 50–70 g/km by 20230,which is roughly half of what it is now[4].In addition,China proposed a plan at“United Nations General Assembly”in 2020 to peak carbon dioxide emissions by 2030 and strive to achieve carbon neutrality by 2060.It is a fact that the whole world is committed to changing the current energy structure,protecting the Earth’s ecology,and achieving global sustainable development[5].

A fault locating method for multi-branch hybrid transmission lines in wind farm based on redundancy parameter estimation

In order to solve the problem of‘‘abandoned’’wind caused by short circuit faults in a wind farm,a wind farm fault locating method based on redundancy parameter estimation is proposed.Using the characteristics of the traveling wave,transmission equations containing the position of the fault point are constructed.Parameter estimation from statistical theory is used to solve the redundant transmission equations formed by multiple measuring points to locate the faults.In addition,the bad data error detection capability of the parameter estimation is used to determine bad data and remove them.This improves locating accuracy.A length coefficient is introduced to solve the error enlargement problem caused by a transmission line sag.The proposed fault locating method can solve the fault branch misjudgment problem caused by the short-circuit faults near the data measuring nodes of thewind farm based on the proposed fault interval criterion.It also avoids the requirements to the traveling wave speed of traditional methods,thus its fault location is more accurate.Its effectiveness is verified through simulations in PSCAD/EMTDC,and the results show that it can be used in thefault locating of hybrid transmission lines.

An improved empirical wavelet transform method for rolling bearing fault diagnosis

Empirical wavelet transform(EWT)based on the scale space method has been widely used in rolling bearing fault diagnosis.However,using the scale space method to divide the frequency band,the redundant components can easily be separated,causing the band to rupture and making it difficult to extract rolling bearing fault characteristic frequency effectively.This paper develops a method for optimizing the frequency band region based on the frequency domain feature parameter set.The frequency domain feature parameter set includes two characteristic parameters:mean and variance.After adaptively dividing the frequency band by the scale space method,the mean and variance of each band are calculated.Sub-bands with mean and variance less than the main frequency band are combined with surrounding bands for subsequent analysis.An adaptive empirical wavelet filter on each frequency band is established to obtain the corresponding empirical mode.The margin factor sensitive to the shock pulse signal is introduced into the screening of empirical modes.The empirical mode with the largest margin factor is selected to envelope spectrum analysis.Simulation and experiment data show this method avoids over-segmentation and redundancy and can extract the fault characteristic frequency easier compared with only scale space methods.

Multi‑scale Battery Modeling Method for Fault Diagnosis

Fault diagnosis is key to enhancing the performance and safety of battery storage systems.However,it is challenging to realize efficient fault diagnosis for lithium-ion batteries because the accuracy diagnostic algorithm is limited and the features of the different faults are similar.The model-based method has been widely used for degradation mechanism analysis,state estimation,and life prediction of lithium-ion battery systems due to the fast speed and high development efficiency.This paper reviews the mainstream modeling approaches used for battery diagnosis.First,a review of the battery’s degradation mechanisms and the external factors affecting the aging rate is presented.Second,the different modeling approaches are summarized,from microscopic to macroscopic scales,including density functional theory,molecular dynamics,X-ray computed tomography technology,electrochemical model,equivalent circuit model,distributed model and neural network algorithm.Subsequently,the advantages and disadvantages of these model approaches are discussed for fault detection and diagnosis of batteries in different application scenarios.Finally,the remaining challenges of model-based battery diagnosis and the future perspective of using cloud control and battery intelligent networking to enhance diagnostic performance are discussed.

A new automatic convolutional neural network based on deep reinforcement learning for fault diagnosis

Convolutional neural network (CNN) has achieved remarkable applications in fault diagnosis. However, the tuning aiming at obtaining the well-trained CNN model is mainly manual search. Tuning requires considerable experiences on the knowledge on CNN training and fault diagnosis, and is always time consuming and labor intensive, making the automatic hyper parameter optimization (HPO) of CNN models essential. To solve this problem, this paper proposes a novel automatic CNN (ACNN) for fault diagnosis, which can automatically tune its three key hyper parameters, namely, learning rate, batch size, and L2-regulation. First, a new deep reinforcement learning (DRL) is developed, and it constructs an agent aiming at controlling these three hyper parameters along with the training of CNN models online. Second, a new structure of DRL is designed by combining deep deterministic policy gradient and long short-term memory, which takes the training loss of CNN models as its input and can output the adjustment on these three hyper parameters. Third, a new training method for ACNN is designed to enhance its stability. Two famous bearing datasets are selected to evaluate the performance of ACNN. It is compared with four commonly used HPO methods, namely, random search, Bayesian optimization, tree Parzen estimator, and sequential model-based algorithm configuration. ACNN is also compared with other published machine learning (ML) and deep learning (DL) methods. The results show that ACNN outperforms these HPO and ML/DL methods, validating its potential in fault diagnosis.

Multi-parameter quantitative assessment of 3D geological models for complex fault-block oil reservoirs

In the field of 3 D geologic modeling, researchers often pay more attention to modeling methods and workflows, but neglect the quantitative evaluation of models. If the evaluation is narrowed to the same reservoir type, the comparability and practicality of quantitative assessment will be emerging. The evaluation system should include three parts: data verification, geological understanding and process check. Data verification mainly involves testing the accuracy of local prediction with actual data, and geological understanding is to examine whether the global estimation honors geological principles and prior insights. Process check is also indispensable to avoid occasionality. To this end, we produced a set of assessment criteria, taking complex fault-block sandstone oil reservoir as an example. To be specific, thirteen characteristic parameters were totally selected, setting weights according to their rated importance, formulating three-level evaluation standards in a centesimal system for each characteristic parameter, and obtaining the final assessment based on the cumulative score. The results indicate that such evaluation can not only access the quality of the model objectively and comprehensively, but also identify the aspects in need of improvement through the deduction items.

Hybrid Slip Model for Near-Field Ground Motion Estimation Based on Uncertainty of Source Parameters

The hybrid slip model used to generate a finite fault model for near-field ground motion estimation and seismic hazard assessment was improved to express the uncertainty of the source form of a future earthquake.In this process, source parameters were treated as normal random variables, and the Fortran code of hybrid slip model was modified by adding a random number generator so that the code could generate many finite fault models with different dimensions and slip distributions for a given magnitude.Furth...

基于参数优化VMD和改进LSSVM的道岔故障诊断方法

为了解决道岔设备智能故障诊断中特征指标难以提取以及模型训练时间较长的问题,以ZDJ9型转辙机带动的道岔设备为研究对象,以转辙机功率曲线为数据基础,提出一种基于参数优化变分模态分解(Variational Mode Decomposition,VMD)和改进最小二乘支持向量机(Least Squares Support Vector Machines,LSSVM)的道岔故障诊断方法。首先,采用鲸鱼优化算法(Whale Optimization Algorithm,WOA)优化VMD参数,得到模态(Intrinsic Mode Functions,IMF)分量个数和惩罚因子的最优参数组合。其次,计算IMF分量与功率曲线的相关系数,优选相关性较大的前3阶IMF分量,并计算功率谱熵、模糊熵及包络熵值,建立多特征融合样本数据库。最后,针对麻雀搜索算法(Sparrow Search Algorithm,SSA)易陷入局部最优的问题,通过改进Tent混沌映射初始化策略随机生成种群,正余弦算法(Sine Cosine Algorithm,SCA)更新追随者的位置,并采用改进SSA优化LSSVM算法的惩罚因子和核函数方差,构建基于TSSSA-LSSVM的道岔故障诊断模型。实验结果表明:所提道岔故障诊断方法是可行的,采用多特征融合能够更加全面地提取道岔典型故障特征,反映道岔的真实运行状态,提高了故障诊断准确率,而且较TSSSA-SVM,PSO-LSSVM,GWO-LSSVM以及SSA-LSSVM等方法具有较高的故障诊断准确率、召回率以及较低的漏报率,减少了模型训练时间,完全满足现场道岔故障导向安全的原则,具有更好的故障诊断性能,对现场道岔设备的故障维修具有一定的指导意义。