基于LMD和SVM算法的模拟电路故障诊断

对模拟故障电路进行特征提取与分类是模拟电路诊断的两个重要环节。现有方法多对时域响应信号进行小波变换以提取故障特征,并用神经网络或支持向量机方法实现对故障进行分类。为提高模拟电路故障诊断率,提出一种局域均值分解(LMD)与SVM相结合的新算法。该算法运用局域均值算法(LMD),将其自适应地分解为一系列单分量调幅-调频信号(PF),通过提取电路正常和故障状态的特征,运用SVM对其分类,获得诊断效率。仿真实验结果表明,该方法对模拟电路的故障诊断精度达到98%以上,适用于模拟电路的故障诊断。     

基于小波高频系数统计特征的电路故障诊断

对故障电路进行特征提取与分类是模拟电路诊断的两个重要环节。现有方法多对时域响应信号进行小波变换以提取故障特征,并用神经网络或支持向量机方法实现对故障进行分类。为提高模拟电路故障诊断率,提出一种新的特征选取方法:在模拟电路的时域响应中对其进行小波变换,并对变换得到的高频细节系数统计平均值、标准偏差、峭度、熵和偏斜度等统计特征,并建立以支持向量机为分类器的故障诊断系统。以两种常见电路为例,实验结果表明,提出方法对常见电路进行故障诊断,准确率得到提升,精度达到99%以上,优于传统单纯小波系数分析方法,适用于模拟电路的故障诊断。     

Analogue electronic circuit diagnosis based on ANNs

Feed-forward artificial neural networks (ANNs) have been applied to the diagnosis of nonlinear dynamic analogue electronic circuits. Using the simulation-before-test (SBT) approach, a fault dictionary was first created containing responses observed at all inputs and outputs of the circuit. The ANN was considered as an approximation algorithm to capture mapping enclosed within the fault dictionary and, in addition, as an algorithm for searching the fault dictionary in the diagnostic phase. In the example given DC and small signal frequency domain measurements were taken as these data are usually given in device’s data-sheets. A reduced set of data per fault (DC output values, the nominal gain and the 3 dB cut-off frequency, measured at one output terminal) was recorded. Soft (parametric) and catastrophic (shorts and opens) defects were introduced and diagnosed simultaneously and successfully. Large representative set of faults was considered, i.e., all possible catastrophic transistor faults and qualified representatives of soft transistor faults were diagnosed in an integrated circuit. The generalization property of the ANNs was exploited to handle noisy measurement signals.     

Intelligent fault diagnosis system of induction motor based on transient current signal

This paper presents a method for induction motor fault diagnosis based on transient signal using component analysis and support vector machine (SVM). The start-up transient current signal is selected as features source for fault diagnosis. Preprocessing of transient current signal is performed using smoothing and discrete wavelet transform to highlight the salient features of faults. In this work, independent component analysis, principal component analysis and their kernel are performed to reduce the dimension of features and to extract the optimal features for classification process. In this work, the influence of the number of component analysis towards diagnosis accuracy is also studied. SVM multi-class classification using one against all strategy is selected for classification tool due to good generalization properties. Performance of the system is validated by applying the system to induction motor faults diagnosis. According to the result, the system has potential to serve an intelligent fault diagnosis system in real application.     

结合WGAN GP与CNN SVM的滚动轴承故障红外诊断

针对实际工程应用中由于滚动轴承故障状态出现的时间很短而导致数据集不平衡难以采用深度学习算法进行故障诊断的问题,提出了一种基于Wasserstein距离的梯度惩罚生成对抗网络(WGAN GP)和基于支持向量机分类的卷积神经网络(CNN SVM)相结合的滚动轴承故障红外诊断方法。从红外热像图中构建不平衡数据集,通过采用WGAN GP对不平衡数据扩充以达到数据集均衡,之后将CNN SVM模型应用于数据集,提取样本深度特征完成故障分类。实验表明,WGAN GP与CNN SVM相结合的模型在不平衡数据集下表现良好,相较于其他模型有更好的故障诊断能力,并且在故障分类阶段的用时可减少1689以上。     

Feature Vector Selection Method Using Mahalanobis Distance for Diagnostics of Analog Circuits Based on LS-SVM

Multi-fault diagnosis for analog circuits based on support vector machine (SVM) usually used a single feature vector to train all binary SVM classifier. In fact, each binary SVM classifier has different classification accuracy for different feature vectors. However, no one has discussed the optimal or near-optimal feature vector selection problem. Based on Mahalanobis distance, a near-optimal feature vector selection method has been proposed for diagnostics of analog circuits using the least squares SVM (LS-SVM). The selection problems of wavelet types, wavelet decomposition level, and normalization methods have been also discussed. Two filters with parametric faults and a nonlinear half-wave rectifier with hard and parametric faults were used as circuits under test (CUTs). The simulation results showed the following: (1) the accuracies using the feature vector with the maximum MD were better than the average accuracies using all the feature vectors, and were better than most accuracies using a single feature vector. But the computation time using the MD method was an order of magnitude larger than that using a single feature vector; (2) Most the diagnostic accuracies using the maximum MD method were near to the optimal accuracies using the exhaustive method while the computation time was reduced about 20–50?% in comparision to the exhaustive method; (3) the Haar wavelet was the best choice among Daubechie’s wavelet family for all CUTs’ diagnosis; (4) only non-normalization, all-normalization, and part-normalization methods are necessary to be considered for feature vector normalization. The proposed method can obtain a near-optimal diagnostic accuracy in a reasonable time, which is beneficial for analog IC or circuits testing and diagnosis.     

Research on fault diagnosis of airborne fuel pump based on EMD and probabilistic neural networks

Airborne fuel pump is a key component of the airborne fuel system, which once fails will bring a huge negative impact on aircraft safety. Therefore, accurate, reliable and effective fault diagnosis must be performed. However, the current airborne fuel pump has several difficulties: fault samples shortage, high maintenance costs and low diagnostic efficiency. In this paper, after Failure Mode, Effects and Criticality Analysis (FMECA) of airborne fuel pump, an experimental platform of airborne fuel transfusion system is developed and then a fault diagnosis model based on empirical mode decomposition (EMD) and probabilistic neural networks (PNN) is established. Meanwhile, the diagnosis model is verified by practical experiments, and the sensor layout of the experimental platform is optimized. Firstly, the vibration signals and pressure signals under normal state and six types of typical fuel pump faults are acquired on the experimental platform. Then EMD method is applied to decompose the original vibration signals into a finite Intrinsic Mode Functions (IMFs) and a residual. Secondly, the energy of first four IMFs is extracted as vibration signals fault feature, combined with the mean outlet pressure to construct fault feature vectors. Then feature vectors are divided into training samples and testing samples. Training samples are used to train PNN fault diagnosis model and testing samples are used to verify the model. Finally, the experimental results show that only one pressure sensor and one y-axis vibration sensor are needed to achieve 100% fault diagnosis. Furthermore, compared with SVM and GA-BP, the PNN fault diagnosis model has fast convergence, high efficiency and a higher performance and recognition for the typical faults of airborne fuel pump.     

基于随机森林算法的模拟电路故障诊断

为提高模拟电路参变故障的诊断率,提出基于多特征向量提取和随机森林(RF)算法的模拟电路故障诊断新方法。采用时域和频域特征向量组合的多维特征向量以反映不同故障特征,经RF算法进行决策,并对决策树棵数及候选特征向量个数进行优化。故障诊断实验结果表明,所提方法能较好地实现容差模拟电路故障诊断,与支持向量机(SVM)方法相比,表现出更好的分类性能;与小波(包)特征提取方法相比,简化了多维数据特征提取步骤,易于实现在线故障诊断。     

层次支持向量机在数字电路故障诊断中的应用

针对在数字电路故障诊断过程中存在的样本不平衡度严重的问题,采用层次式支持向量机实现对其故障诊断,通过考虑各类样本的数据量来构造以支持向量为叶节点的树,该方法可有效地解决样本不平衡所带来的问题,同时能够减少计算SVM分类器的个数,提高了训练和诊断速度及准确率.针对故障样本集不可能覆盖所有故障状态而出现的未知故障状态的问题...     

Test generation algorithm for analog systems based on support vector machine

In some methods for test generation, an analog device under test (DUT) is treated as a discrete-time digital system by placing it between a digital-to-analog converter and an analog-to-digital converter. Then the test patterns and responses can be performed and analyzed in the digital domain. We propose a novel test generation algorithm based on a support vector machine (SVM). This method uses test patterns derived from the test generation algorithm as input stimuli, and sampled output responses of the analog DUT for classification and fault detection. The SVM is used for classification of the response space. When the responses of normal circuits are similar to those of faulty circuits (i.e., the latter have only small parametric faults), the response space is mixed and traditional algorithms have difficulty in distinguishing the two groups. However, the SVM provides an effective result. This paper also proposes an algorithm to calculate the test sequence for input stimuli using the SVM results. Numerical experiments prove that this algorithm can enhance the precision of test generation.     

Vibration analysis for bearing fault detection and classification using an intelligent filter

This paper proposes an intelligent method based on artificial neural networks (ANNs) to detect bearing defects of induction motors. In this method, the vibration signal passes through removing non-bearing fault component (RNFC) filter, designed by neural networks, in order to remove its non-bearing fault components, and then enters the second neural network that uses pattern recognition techniques for fault classification. Four different categories include; healthy, inner race defect, outer race defect, and double holes in outer race are investigated. Compared to the regular fault detection methods that use frequency-domain features, the proposed method is based on analyzing time-domain features which needs less computational effort. Moreover, machine and bearing parameters, and the vibration signal spectrum distribution are not required in this method. It is shown that better results are achieved when the filtered component of the vibration signal is used for fault classification rather than common methods that use directly vibration signal. Experimental results on three-phase induction motor verify the ability of the proposed method in fault diagnosis despite low quality (noisy) of measured vibration signal.     

基于决策树SVM的某型无人机发射机故障诊断

为能够准确地对无人机数据链发射机进行故障诊断,通过对某型无人机数据链发射机工作原理和故障模型分析,结合长期的维护、保障经验,提出了一种基于故障优先级的决策树支持向量机故障诊断方法,避免了现有的多类分类支持向量机在多类故障诊断方面存在不足,优化了多类分类支持向量机组合策略。经试验,与几种常用的多类分类支持向量机方法对比,该诊断策略有效得提高了故障诊断正确率,能够准确地定位发射机内部故障功能模块,具有一定的实际意义。     

Diagnostics of Analog Circuits Based on LS-SVM Using Time-Domain Features

Most researchers use wavelet transforms to extract features from a time-domain transient response from analog circuits to train classifiers such as neural networks (NNs) and support vector machines (SVMs) for analog circuit diagnostics. In this paper, we have proposed some new feature selection methods from a time-domain transient response, and compared the diagnostic results based on a least squares SVM (LS-SVM) using different time-domain feature vectors. First, we have improved two traditional feature selection methods: (a) using the mean and standard deviation in wavelet transform features, and (b) using the mean, standard deviation, skewness, kurtosis, and entropy in statistical property features. Then, a conventional time-domain feature vector based on the impulse response properties of a control system has been proposed. The simulation experiments for a leapfrog filter and a nonlinear rectifier show that: (1) the two improved methods have better accuracy than the traditional methods; (2) the proposed conventional time-domain feature vector is effective in the diagnostics of analog circuits—over 99 % for both of the two example circuits; (3) the proposed diagnostic method can diagnose soft faults, hard faults, and multi-faults, regardless of component tolerances and nonlinearity effects.     

Application of Multiscale Learning Neural Network Based on CNN in Bearing Fault Diagnosis

With the application of intelligent manufacturing becoming more and more widely, the losses caused by mechanical faults of equipment increase. Identifying and troubleshooting faults in an early stage are important. The process of traditional data-driven fault diagnosis method includes data acquisition, fault classification, and feature extraction, in which classification accuracy is directly affected by the result of feature extraction. As a common deep learning method in image recognition, the convolutional neural network (CNN) demonstrates good performance in fault diagnosis. CNN can adaptively extract features from original signals and eliminate the effect of conventional handcrafted features. In this study, a multiscale learning neural network that contains one-dimension (1D) and two-dimension (2D) convolution channels is proposed. The network can learn the local correlation of adjacent and nonadjacent intervals in periodic signals, such as vibration data. The Paderborn data set is came into use to demonstrate the classification accuracy of the method which is brought forward, which includes three conditions of healthy, outer ring (OR) damage and inner ring (IR) damage. The classification accuracy of the method which is put forward is up to 98.58%. The same dataset was applied to test the classification accuracy of support vector machine (SVM) for comparison. And the proposed multiscale learning neural network demonstrates considerable improvements.

     

Design error diagnosis in digital circuits with stuck-at fault model

In this paper we describe in detail a new method for the single gate-level design error diagnosis in combinational circuits. Distinctive features of the method are hierarchical approach (the localizing procedure starts at the macro level and finishes at the gate level), use of stuck-at fault model (it is mapped into design error domain only in the end), and design error diagnostic procedure that uses only test patterns generated by conventional gate-level stuck-at fault test pattern generators (ATPG). No special diagnostic tests are used because they are much more time consuming. Binary decision diagrams (BDD) are exploited for representing and localizing stuck-at faults on the higher signal path level. On the basis of detected faulty signal paths, suspected stuck-at faults at gate inputs are calculated, and then mapped into suspected design error(s). This method is enhanced compared to our previous work. It is applicable to redundant circuits and allows using incomplete tests for error diagnosis. Experimental data on ISCAS benchmark circuits shows the advantage of the proposed method compared to the known algorithms of design error diagnosis.     

粗糙集和支持向量机在复杂电路系统诊断中的应用

为了解决复杂电路系统故障样本少、特征信息冗杂的问题,提出了一种基于粗糙集属性约简理论和支持向量机分类方法相结合的故障诊断方法.首先采用粗糙集约简故障模式库中的冗余特征属性和矛盾样本,然后提取最简故障特征模式作为支持向量机的学习样本,通过样本训练使构建的支持向量机多分类器能够快速实现故障诊断的目的.最后,通过仿真算例验证了该方法在小样本故障识别上的有效性和可行性.     

An advanced diagnostic method for delay faults in combinational faulty circuits

Due to physical defects or process variations, a logic circuit may fail to operate at the desired clock speed. So, verifying the timing behavior of digital circuits is always necessary, and needs to test for delay faults. When a delay fault has been detected, a specific diagnostic method is required to locate the site of the fault in the circuit. So, a reliable method for delay fault diagnosis is proposed in this paper. Firstly, we present the basic diagnostic method for delay faults, which is based on multivalued simulation and critical path tracing. Next, heuristics are given that decrease the number of critical paths and improve diagnosis results. In the second part of this paper, we provide an approximate method to refine the results obtained with the basic diagnostic process. We compute the detection threshold of the potential delay faults, and use statistical studies to classify the faults from the most likely to be the cause of failure to the less likely. Finally, results obtained with ISCAS'85 circuits are presented to show the effectiveness of the method.     

Synergistic use of soft computing technologies for fault detection in gas turbine engines

In this paper, we present a synergistic approach to startup fault detection and diagnosis (FDD) in gas turbine engines. The method employs statistics, signal processing, and soft computing techniques in a complementary manner to address fault detection at transient conditions. Traditional turbine engine FDD methods are based on engine data collected at steady-state conditions. However, incipient faults are difficult to diagnose using steady-state engine data; only engine faults that are fairly developed can be detected using conventional methods. Because incipient engine component faults are often manifest in the engine startup characteristics, we present a method to characterize the engine transient startup. Engine sensor data during engine startup are recorded in time series format. The sensor profiles corresponding to "good" and "bad" engine startups are sampled using the bootstrap technique. A feature vector is extracted in two steps, and signal processing is followed by the feature vector selection. In the signal processing step, principal component analysis (PCA) is applied to reduce the samples consisting of sensor profiles into a smaller set. In the feature vector selection step, a cost function is defined, and important discriminating features for fault diagnosis are distilled from the PCA output vector. The features obtained from this step are then classified using neural-network-based methods. The "leave-one-out" approach to cross validation is applied to obtain an objective evaluation of the neural network training. The proposed FDD method is evaluated using actual engine startup data, and the results are presented.     

Induction machine fault detection using SOM-based RBF neural networks

A radial-basis-function (RBF) neural-network-based fault detection system is developed for performing induction machine fault detection and analysis. Four feature vectors are extracted from power spectra of machine vibration signals. The extracted features are inputs of an RBF-type neural network for fault identification and classification. The optimal network architecture of the RBF network is determined automatically by our proposed cell-splitting grid algorithm. This facilitates the conventional laborious trial-and-error procedure in establishing an optimal architecture. In this paper, the proposed RBF machine fault diagnostic system has been intensively tested with unbalanced electrical faults and mechanical faults operating at different rotating speeds. The proposed system is not only able to detect electrical and mechanical faults, but the system is also able to estimate the extent of faults.