WO2019090878A1 - 基于向量值正则核函数逼近的模拟电路故障诊断方法 - Google Patents

基于向量值正则核函数逼近的模拟电路故障诊断方法 Download PDF

Info

Publication number
WO2019090878A1
WO2019090878A1 PCT/CN2017/114969 CN2017114969W WO2019090878A1 WO 2019090878 A1 WO2019090878 A1 WO 2019090878A1 CN 2017114969 W CN2017114969 W CN 2017114969W WO 2019090878 A1 WO2019090878 A1 WO 2019090878A1
Authority
WO
WIPO (PCT)
Prior art keywords
vvrkfa
training sample
fault
sample set
training
Prior art date
Application number
PCT/CN2017/114969
Other languages
English (en)
French (fr)
Inventor
何怡刚
何威
尹柏强
李兵
李志刚
佐磊
张朝龙
Original Assignee
合肥工业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 合肥工业大学 filed Critical 合肥工业大学
Priority to US16/497,444 priority Critical patent/US11287470B2/en
Publication of WO2019090878A1 publication Critical patent/WO2019090878A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/2851Testing of integrated circuits [IC]
    • G01R31/2855Environmental, reliability or burn-in testing
    • G01R31/2872Environmental, reliability or burn-in testing related to electrical or environmental aspects, e.g. temperature, humidity, vibration, nuclear radiation
    • G01R31/2879Environmental, reliability or burn-in testing related to electrical or environmental aspects, e.g. temperature, humidity, vibration, nuclear radiation related to electrical aspects, e.g. to voltage or current supply or stimuli or to electrical loads
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/316Testing of analog circuits
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/2851Testing of integrated circuits [IC]
    • G01R31/2855Environmental, reliability or burn-in testing
    • G01R31/286External aspects, e.g. related to chambers, contacting devices or handlers
    • G01R31/2868Complete testing stations; systems; procedures; software aspects
    • G01R31/287Procedures; Software aspects
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/2851Testing of integrated circuits [IC]
    • G01R31/2884Testing of integrated circuits [IC] using dedicated test connectors, test elements or test circuits on the IC under test
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01RMEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/28Testing of electronic circuits, e.g. by signal tracer
    • G01R31/316Testing of analog circuits
    • G01R31/3163Functional testing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10Complex mathematical operations
    • G06F17/17Function evaluation by approximation methods, e.g. inter- or extrapolation, smoothing, least mean square method
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N10/00Quantum computing, i.e. information processing based on quantum-mechanical phenomena
    • G06N10/60Quantum algorithms, e.g. based on quantum optimisation, quantum Fourier or Hadamard transforms
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • G06N20/10Machine learning using kernel methods, e.g. support vector machines [SVM]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/004Artificial life, i.e. computing arrangements simulating life
    • G06N3/006Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00Computing arrangements based on specific mathematical models
    • G06N7/02Computing arrangements based on specific mathematical models using fuzzy logic
    • G06N7/04Physical realisation
    • G06N7/043Analogue or partially analogue implementation

Definitions

  • the invention belongs to the technical field of analog circuit fault diagnosis, and particularly relates to an analog circuit fault diagnosis method based on vector value regular kernel function approximation.
  • ANN Artificial Neural Network
  • SVM Support Vector Machine
  • the technical problem to be solved by the present invention is to overcome the above-mentioned drawbacks existing in the prior art, and to provide an analog circuit fault diagnosis method based on vector value regular kernel function approximation with high diagnostic speed and high accuracy.
  • the invention firstly collects the time domain response signal of the test circuit, and after wavelet packet decomposition, calculates the energy value of each node as the sample feature parameter, applies QPSO to optimize the regular parameters and kernel parameters of the VVRKFA model and constructs a fault diagnosis model based on VVRKFA. Finally, test data is entered into the model to identify the fault category.
  • the present invention first proposes to introduce VVRKFA into the fault diagnosis of an analog circuit.
  • a fault diagnosis classifier is constructed by applying VVRKFA to locate a fault.
  • VVRKFA When VVRKFA is used to distinguish fault types, you need to select its regular parameters and nuclear parameters to achieve the best classification performance. Identifying faults by applying VVRKFA results in higher fault diagnosis accuracy than other existing methods. And when dealing with large-scale data, VVRKFA only requires shorter training and testing time.
  • VVRKFA Vector-Valued regularized kernel function approximation
  • the original sample feature data is divided into two parts: the training sample set and the test sample set.
  • Quantum particle swarm optimization (QPSO) is used to optimize the regular parameters and kernel parameters of the Vector-Valued Regularized Kernel Function Approximation (VVRKFA) mathematical model.
  • VVRKFA Vector-Valued Regularized Kernel Function Approximation
  • the analog circuit to be tested has only one input end and one output end, the input end inputs a sinusoidal signal, and the output end samples a voltage signal.
  • the db10 wavelet packet transform is applied to the collected output signal, that is, the voltage signal.
  • VVRKFA-based fault diagnosis model is constructed as follows:
  • the Gaussian kernel function K(x i , x j ) exp( ⁇
  • 2 ) is selected as the kernel function of VVRKFA to establish the mathematical model of VVRKFA, and ⁇ is the width factor of the Gaussian kernel function;
  • VVRKFA The mathematical model of VVRKFA is as follows:
  • K(.,.) is a kernel function.
  • the kernel function K(.,.) is the optimal kernel function obtained in step (3.b);
  • C is a regular parameter, and in regular expression (2), Parameter C is the optimal kernel parameter obtained in step (3.b); Is the base vector, N is the number of fault categories;
  • m is the dimension of the training samples in the training sample set;
  • n is the total number of all training samples in the training sample set.
  • step (3.d) Apply the mapping function constructed in step (3.c) to establish the decision function of VVRKFA.
  • the decision function of VVRKFA is expressed as:
  • x t is the test sample within the test sample set; Is the center point of all training samples in the j-th fault of the training sample set, d M represents the Mahalanobis distance, and x i is the training sample in the training sample set.
  • a mapping projection of the sample x i in the feature subspace To test the projected projection of the sample x t in the feature subspace.
  • the establishment of the decision function is completed based on the VVRKFA fault diagnosis model.
  • the quantum particle swarm optimization algorithm is used to optimize the regular parameters and the kernel parameters of the VVRKFA mathematical model, and the optimal regular parameters and nuclear parameters of the VVRKFA mathematical model are obtained.
  • the specific steps are as follows:
  • each particle is a two-dimensional vector
  • the first dimension is the VVRKFA mathematical model.
  • the regular parameter, the second dimension is the kernel parameter of the VVRKFA mathematical model.
  • the test sample data set is input to the fault diagnosis model, and the circuit fault category is identified, and the fault category of each test sample in the test sample set is obtained, thereby obtaining the diagnostic correct rate of each type of fault, and completing Diagnosis of the analog circuit being tested.
  • the invention uses the Vector-Value regularized kernel function approximation (VVRKFA) method to significantly reduce the time and space overhead while maintaining a high diagnostic accuracy, which is obviously superior to the traditional fault classification method.
  • VVRKFA Vector-Value regularized kernel function approximation
  • the present invention has the following advantages:
  • VVRKFA has higher classification accuracy and less time and space overhead.
  • QPSO algorithm Propose the application of QPSO algorithm to optimize the regularization parameters and kernel parameters of VVRKFA. Compared with the traditional grid search method, the QPSO algorithm can obtain the optimal parameters and significantly improve the performance of VVRKFA.
  • FIG. 1 is a flowchart of an analog circuit fault diagnosis method based on a vector value regular kernel function approximation method according to the present invention.
  • FIG. 2 is a circuit diagram of a video amplifying circuit selected in the present invention.
  • Figure 3 shows the training process of the parameters of the quantum particle swarm optimization algorithm for vector value regular kernel approximation method.
  • the present invention consists of four steps: Step 1, extracting the time domain response signal of the analog circuit under test.
  • step 2 the signal is subjected to 3-layer db10 wavelet packet decomposition, and the third layer node is extracted, and the energy of 8 nodes is used as the original sample feature data.
  • the original sample feature data is divided into two parts: the training sample set and the test sample set.
  • Step 3 Based on the training sample set, the QPSO algorithm is used to optimize the parameters of the VVRKFA mathematical model, and a fault diagnosis model based on VVRKFA is constructed.
  • Step 4 Input a test sample set to the constructed VVRKFA-based fault diagnosis model to identify the circuit fault category.
  • step 1 the time domain response signal of the analog circuit under test is obtained, and the input terminal is excited by a sinusoidal signal having an amplitude of 5 volts and a frequency of 100 Hz, and the output terminal samples the voltage signal.
  • step 2 the energy calculation method of each node is as follows:
  • the signal is projected onto the space formed by a set of mutually orthogonal wavelet basis functions, and the signal is decomposed into two parts: high frequency and low frequency, and the low frequency and high frequency parts are simultaneously decomposed in the next layer of decomposition. It is a more elaborate method of analysis.
  • the wavelet packet function ⁇ j,k (t) is defined as:
  • j ⁇ Z is the number of decomposition layers
  • k ⁇ Z is the number of frequency band data points
  • t is the time point.
  • h(k-2t), g(k-2t) are respectively low-pass filter coefficients and high-pass filters in the corresponding multi-scale analysis; Is the kth wavelet decomposition coefficient point in the nth frequency band of the jth layer; Is a wavelet decomposition sequence of the 2nth frequency band of the j+1th layer; Is the wavelet decomposition sequence of the 2n+1th band of the j+1th layer; k ⁇ Z is the number of band data points; t is the time point.
  • is the translation parameter.
  • the Gaussian kernel function K(x i , x j ) exp( ⁇
  • 2 ) is selected as the kernel function of VVRKFA to establish the mathematical model of VVRKFA, and ⁇ is the width factor of the Gaussian kernel function;
  • VVRKFA The mathematical model of VVRKFA is as follows:
  • Is a regression coefficient matrix the purpose is to map the feature inner product space to the label space; N is the number of fault categories; m is the dimension of the training sample in the training sample set; Is the dimension of the sample in the dimensionally reduced sample data set after the dimensionality reduction of the training sample set; Is the base vector; C is the regular parameter; Is a slack variable.
  • K(.,.) is a kernel function.
  • the kernel function K(.,.) is the optimal kernel function obtained in step (3.b);
  • C is a regular parameter, and in regular expression (2), Parameter C is the optimal kernel parameter obtained in step (3.b); Is the base vector, N is the number of fault categories;
  • m is the dimension of the training samples in the training sample set;
  • n is the total number of all training samples in the training sample set.
  • step (3.d) Apply the mapping function constructed in step (3.c) to establish the decision function of VVRKFA.
  • the decision function of VVRKFA is expressed as:
  • x t is the test sample within the test sample set; Is the center point of all training samples in the j-th fault of the training sample set, d M represents the Mahalanobis distance, and x i is the training sample in the training sample set.
  • a mapping projection of the sample x i in the feature subspace To test the projected projection of the sample x t in the feature subspace.
  • the establishment of the decision function is completed based on the VVRKFA fault diagnosis model.
  • each particle is a two-dimensional vector
  • the first dimension is the regular parameter of the VVRKFA mathematical model
  • the second dimension is the kernel parameter of the VVRKFA mathematical model.
  • the particle position update formula in the QPSO algorithm is:
  • X i (t+1) P i '(t) ⁇
  • P' i (t) P i (t-1)+(1- ⁇ )P g (t-1), N is the population size, Mbest is the average point of the individual optimal position of all particles; ⁇ max is the maximum inertia weight; ⁇ min is the minimum inertia weight, P j and P g are the optimal position and global optimal position of particle j respectively X is the position of the particle; t is the current number of iterations, ⁇ is the compression expansion factor; u and ⁇ are randomly distributed random numbers between [0,1]; P' i is the updated position of particle i; P i Is the current position of particle i.
  • the test sample data set is input to the fault diagnosis model, and the circuit fault category is identified, and the fault category of each test sample in the test sample set is obtained, thereby obtaining the diagnostic correct rate of each type of fault, and completing the correctness of the fault. Test the diagnosis of the analog circuit.
  • Figure 2 shows the video amplifier circuit.
  • the nominal values and tolerances of the components are marked on the graph.
  • the excitation source has an amplitude of 5V.
  • a sinusoidal signal with a frequency dimension of 100 Hz, and a fault response time domain signal is sampled at the output of the circuit.
  • R1, R2, R3, C4, R4, R5, R6, R8 and Q1 were selected as test objects.
  • Table 1 Table 2 gives the fault code, nominal value and fault value of each test component, where ⁇ and ⁇ respectively indicate that the fault value is higher and lower than the nominal value.
  • a fault-free condition is also considered a fault and the fault code is F0.
  • Sixty data samples were sampled for each fault category, divided into two parts. The first 30 were used to establish a fault diagnosis model for VVRKFA, and the last 30 data were used to test the performance of the fault diagnosis model.
  • the population size and the number of iterations are set to 10 and 100, respectively, the maximum inertia weight is set to 1, and the minimum inertia weight is 0.5.
  • the regularization parameters and kernel width factors obtained by the optimization are 1.0076 ⁇ 10 -4 and 1.0095, respectively.
  • the training process of QPSO optimizing VVRKFA is shown in Figure 3.
  • the optimal regularization parameter and kernel width factor obtained by optimization are used to construct a fault diagnosis model based on VVRKFA, and the test data is input for fault identification.
  • the diagnostic results are shown in Table 3.
  • the GMKL-SVM fault diagnosis model selected by the PSO algorithm parameter correctly identifies all F0, F1, F3, F4, F7, F8, F11, F11, F12, F13, F14, F15 and F14 faults;
  • One F2 fault is identified as an F5 fault
  • one F5 fault is erroneously identified as an F3 fault
  • one F6 fault is erroneously identified as an F0 fault.
  • the fault diagnosis model of VVRKFA after the regularization parameter and the width factor is optimized by QPSO algorithm has achieved good diagnostic results in fault diagnosis. After calculation, the overall correct diagnosis rate of the fault of the analog circuit can reach 98.82%.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Microelectronics & Electronic Packaging (AREA)
  • Computer Hardware Design (AREA)
  • Health & Medical Sciences (AREA)
  • Environmental & Geological Engineering (AREA)
  • Algebra (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Fuzzy Systems (AREA)
  • Automation & Control Theory (AREA)
  • Databases & Information Systems (AREA)
  • Computational Linguistics (AREA)
  • Toxicology (AREA)
  • Condensed Matter Physics & Semiconductors (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Tests Of Electronic Circuits (AREA)
  • Other Investigation Or Analysis Of Materials By Electrical Means (AREA)

Abstract

基于向量值正则核函数逼近的模拟电路故障诊断方法,包括如下步骤:(1)获取模拟电路的故障响应电压信号;(2)对采集的信号进行小波包变换,计算小波包系数能量值作为特征参量;(3)应用量子粒子群算法优化向量值正则核函数逼近的正则化参数和核参数,训练故障诊断模型;(4)应用训练完成的诊断模型,对电路故障进行识别。所述基于向量值正则核函数逼近的模拟电路诊断方法的分类性能优于其它的分类算法,同时应用量子粒子群算法优化参数亦优于传统获取参数的方法,所述诊断方法可高效地诊断出电路的元件故障。

Description

基于向量值正则核函数逼近的模拟电路故障诊断方法 技术领域
本发明属于模拟电路故障诊断技术领域,具体涉及一种基于向量值正则核函数逼近的模拟电路故障诊断方法。
背景技术
随着集成电路技术的发展,大规模数模混合集成电路广泛应用于电子消费品,工业控制及航空航天等领域。一旦其发生故障,将会影响到设备的性能和功能,甚至导致灾难性后果。另据资料报道,虽然模拟部分仅占集成电路芯片的5%,但其故障诊断及测试成本却占总成本的95%。因此,极有必要开展模拟电路故障诊断及测试技术的研究。
现阶段,已有诸多学者采用人工神经网络(Artificial Neural Network,ANN)和支持向量机(Support Vector Machine,SVM)等方法对模拟电路故障进行诊断。然而,ANN方法存在学习速度慢、容易陷入局部最优解及过训练等问题,难以适应诊断实时性要求。支持向量机(support vector machine,SVM)建立在结构风险最小原理基础上,可较好地解决分类中的小样本和非线性问题,但在面临不平衡数据集及大样本时,难以达到最佳分类能力。
发明内容
本发明要解决的技术问题是,克服现有技术存在的上述缺陷,提供一种诊断速度快、准确率高的基于向量值正则核函数逼近的模拟电路故障诊断方法。
本发明首先采集测试电路的时域响应信号,进行小波包分解后,计算各结点的能量值作为样本特征参量,应用QPSO优化VVRKFA模型的正则参数及核参数并构建基于VVRKFA的故障诊断模型,最后向该模型输入测试数据,识别故障类别。
本发明首次提出将VVRKFA引入模拟电路的故障诊断中。本发明中通过应用VVRKFA构建故障诊断分类器,定位故障。应用VVRKFA区分故障类型时,需要选择其正则参数及核参数,以使其达到最佳分类性能。通过应用VVRKFA识别故障,取得相对于其他现有方法更高的故障诊断正确度。并且在处理大规模数据时,VVRKFA仅需更短的训练及测试时间。
本发明解决其技术问题所采用的技术方案是:
基于向量值正则核函数逼近(Vector-Valued regularized kernel function approximation,VVRKFA)的模拟电路故障诊断方法,包括以下步骤:
(1)提取被测试模拟电路各节点的时域响应电压信号,即采集输出信号。
(2)对采集的输出信号进行小波包分解,计算各节点的能量作为原始样本特征数据。并将原始样本特征数据平均划分为训练样本集及测试样本集两部分。
(3)基于训练样本集,应用量子粒子群算法(Quantum particle swarm optimization,QPSO)优化向量值正则核函数逼近(Vector-Valued Regularized Kernel Function Approximation,VVRKFA)数学模型的正则参数及核参数,构建基于VVRKFA的故障诊断模型。
(4)向构建的基于VVRKFA的故障诊断模型输入测试样本集,对电路故障类别进行识别。
进一步,所述步骤(1)中,被测模拟电路只有一个输入端和一个输出端,输入端输入正弦信号,输出端采样电压信号。
进一步,所述步骤(2)中,对采集的输出信号即电压信号应用db10小波包变换。
进一步,所述步骤(3)中,基于VVRKFA的故障诊断模型的构建过程为:
(3.a)确定核函数类型:
选用高斯核函数K(xi,xj)=exp(σ||xi-xj||2)为VVRKFA的核函数,用以建立VVRKFA的数学模型,σ为高斯核函数的宽度因子;
VVRKFA的数学模型如下:
Figure PCTCN2017114969-appb-000001
其中,
Figure PCTCN2017114969-appb-000002
是回归系数矩阵,目的是将特征内积空间映射到标签空间;N是故障类别数;m为训练样本集中训练样本的维数;
Figure PCTCN2017114969-appb-000003
是训练样本集降维后的降维样本数据集中样本的维数;
Figure PCTCN2017114969-appb-000004
是基向量;C是正则参数;
Figure PCTCN2017114969-appb-000005
是松弛变量;矩阵
Figure PCTCN2017114969-appb-000006
是训练样本集降维后的降维样本数据集
Figure PCTCN2017114969-appb-000007
n为训练样本集中所有训练样本总数;K(.,.)是核函数;xi为训练样本集中的训练样本;Yi为训练样本xi的类别标签向量;T为矩阵转置符号;J为目标函数名。[Θ b]=YPT[CI+PPT]-1,其中
Figure PCTCN2017114969-appb-000008
Figure PCTCN2017114969-appb-000009
为训练样本数据集;I为单位矩阵,
Figure PCTCN2017114969-appb-000010
是包含训练样本的m个类别标签向量的矩阵。
(3.b)应用量子粒子群算法优化VVRKFA数学模型,获取VVRKFA数学模型正则参数及核参数的最优值。
(3.c)以训练样本集中的训练样本xi为输入数据,将步骤(3.b)中获 取的最优正则参数及核参数构建如下的向量值映射函数:
Figure PCTCN2017114969-appb-000011
此处,
Figure PCTCN2017114969-appb-000012
为回归系数矩阵,
Figure PCTCN2017114969-appb-000013
为训练样本集降维后的降维样本数据集。K(.,.)是核函数,式(2)中,核函数K(.,.)为步骤(3.b)获取的最优核函数;C是正则参数,式(2)中,正则参数C为步骤(3.b)获取的最优核参数;
Figure PCTCN2017114969-appb-000014
是基向量,N为故障类别数;m是训练样本集中训练样本的维数;
Figure PCTCN2017114969-appb-000015
是训练样本集降维后的降维样本数据集中样本的维数;n为训练样本集中所有训练样本的总数。
(3.d)应用步骤(3.c)中构建的映射函数,建立VVRKFA的决策函数,VVRKFA的决策函数表示为:
Figure PCTCN2017114969-appb-000016
其中xt为测试样本集内的测试样本;
Figure PCTCN2017114969-appb-000017
是训练样本集中第j类故障中所有训练样本的中心点,dM表示马氏距离,xi为训练样本集中的训练样本,
Figure PCTCN2017114969-appb-000018
是类内协方差矩阵,n是训练样本集中训练样本总数;N为故障类别数;nj为训练样本集中第j类故障类中的训练样本个数;
Figure PCTCN2017114969-appb-000019
为训练样本xi在特征子空间的映射投影。
Figure PCTCN2017114969-appb-000020
为测试样本xt在特征子空间的映射投影。
决策函数的建立即为基于VVRKFA故障诊断模型的构建完成。
进一步,所述步骤(3.b)中,应用量子粒子群算法优化VVRKFA数学模型的正则参数与核参数,获取VVRKFA数学模型的最优正则参数及核参数,具体步骤为:
(3.b.1)初始化QPSO算法参数,包括种速度、位置、种群规模、迭代次数和寻优范围。此处每个粒子均为二维向量,第一维为VVRKFA数学模型 的正则参数,第二维为VVRKFA数学模型的核参数。
(3.b.2)计算粒子的适应度,获取全局最优个体和局部最优个体。
(3.b.3)更新粒子的速度及位置。
(3.b.4)重复步骤(3.b.2)和(3.b.3)直至达到最大迭代次数,输出最优参数结果(即VVRKFA数学模型的最优正则参数及核参数)。
进一步,所述步骤(4)中,向故障诊断模型输入测试样本数据集,对电路故障类别进行识别,得到测试样本集中每个测试样本所属故障类别,进而获得每类故障的诊断正确率,完成对被测试模拟电路的诊断。
本发明使用向量值正则核函数逼近(Vector-Value regularized kernel function approximation,VVRKFA)方法,在保持较高诊断准确率下,能显著减少时间及空间开销,明显优于传统的故障分类方法。
相比现有技术,本发明具有如下优点:
(1)首次提出将VVRKFA引入模拟电路的故障诊断中。相比于传统的ANN及SVM方法,VVRKFA具有更高的分类精度,更少的时间及空间开销。(2)提出应用QPSO算法优化VVRKFA的正则化参数和核参数。相比于传统的网格搜索法,QPSO算法可以获得最优的参数,显著提高VVRKFA的性能。
附图说明
图1为本发明基于向量值正则核函数逼近方法的模拟电路故障诊断方法的流程图。
图2为本发明选用的视频放大电路电路图。
图3为量子粒子群算法优化向量值正则核函数逼近方法参数的训练过程。
具体实施方式
以下结合附图和实施例对本发明作进一步详细说明。
参照图1,本发明由4个步骤构成:步骤1,提取被测模拟电路的时域响应信号。步骤2,对信号进行3层db10小波包分解,提取第3层节点,共8个节点的能量作为原始样本特征数据。并将原始样本特征数据平均划分为训练样本集及测试样本集两部分。步骤3,基于训练样本集,应用QPSO算法优化VVRKFA数学模型的参数,构建基于VVRKFA的故障诊断模型。步骤4,向构建的基于VVRKFA的故障诊断模型输入测试样本集,对电路故障类别进行识别。
步骤1中,获取被测模拟电路的时域响应信号,输入端通过幅值为5伏,频率为100Hz的正弦信号激励,输出端采样电压信号。
步骤2中,各节点的能量的计算方法如下:
小波包分析时将信号投影到一组互相正交的小波基函数张成的空间上,并将信号分解成高频和低频两部分,在下一层分解中同时对低频和高频部分进行分解,是一种更为精细的分析方法。
小波包函数μj,k(t)定义为:
μj,k(t)=2j/2μ(2jt-k);
其中:j∈Z为分解层数,k∈Z为频带数据点数;t为时间点。
对于一组离散信号x(t),小波包分解算法如下式:
Figure PCTCN2017114969-appb-000021
其中h(k-2t),g(k-2t)分别为相应的多尺度分析中低通滤波器系数和高通滤波器;
Figure PCTCN2017114969-appb-000022
是第j层第n个频带中第k个小波分解系数点;
Figure PCTCN2017114969-appb-000023
是第j+1层第2n个频带的小波分解序列;
Figure PCTCN2017114969-appb-000024
是第j+1层第2n+1个频带的小波分解序列;,k∈Z为频带数据点数;t为时间点。
Figure PCTCN2017114969-appb-000025
表示经小波包分解后节点(j,n)所对应的第k个系数,节点(j,n)表示第j层的第n个频带。τ为平移参数。
则小波包节点能量值为:
Figure PCTCN2017114969-appb-000026
N为第i个频带的长度;j是小波分解层数;k为各频带的序列点;
Figure PCTCN2017114969-appb-000027
是第j层第i个频带第k个小波分解系数。
VVRKFA的故障诊断模型建立步骤为:
(3.a)确定核函数类型:
选用高斯核函数K(xi,xj)=exp(σ||xi-xj||2)为VVRKFA的核函数,用以建立VVRKFA的数学模型,σ为高斯核函数的宽度因子;
VVRKFA的数学模型如下:
Figure PCTCN2017114969-appb-000028
其中,
Figure PCTCN2017114969-appb-000029
是回归系数矩阵,目的是将特征内积空间映射到标签空间;N是故障类别数;m为训练样本集中训练样本的维数;
Figure PCTCN2017114969-appb-000030
是训练样本集降维后的降维样本数据集中样本的维数;
Figure PCTCN2017114969-appb-000031
是基向量;C是正则参数;
Figure PCTCN2017114969-appb-000032
是松弛变量。VVRKFA的核函数中的元素xj取值为矩阵B,矩阵
Figure PCTCN2017114969-appb-000033
是训练样本集降维后的降维样本数据集
Figure PCTCN2017114969-appb-000034
n为训练样本集中所有训练样本总数;K(.,.)是核函数;xi为训练样本集中的训练样本;Yi为训练样本xi的类别标签向量;T为矩阵转置符号;J为目标函数名。[Θ b]=YPT[CI+PPT]-1,其中
Figure PCTCN2017114969-appb-000035
Figure PCTCN2017114969-appb-000036
为训练样本数据集; I为单位矩阵,
Figure PCTCN2017114969-appb-000037
是包含训练样本的m个类别标签向量的矩阵。
(3.b)应用量子粒子群算法优化VVRKFA数学模型,获取VVRKFA数学模型正则参数的最优值及核参数的最优值。
(3.c)以训练样本集中的训练样本xi为输入数据,将步骤(3.b)中获取的最优正则参数及最优核参数构建如下的向量值映射函数:
Figure PCTCN2017114969-appb-000038
此处,
Figure PCTCN2017114969-appb-000039
为回归系数矩阵,
Figure PCTCN2017114969-appb-000040
为训练样本集降维后的降维样本数据集。K(.,.)是核函数,式(2)中,核函数K(.,.)为步骤(3.b)获取的最优核函数;C是正则参数,式(2)中,正则参数C为步骤(3.b)获取的最优核参数;
Figure PCTCN2017114969-appb-000041
是基向量,N为故障类别数;m是训练样本集中训练样本的维数;
Figure PCTCN2017114969-appb-000042
是训练样本集降维后的降维样本数据集中样本的维数;n为训练样本集中所有训练样本的总数。
(3.d)应用步骤(3.c)中构建的映射函数,建立VVRKFA的决策函数,VVRKFA的决策函数表示为:
Figure PCTCN2017114969-appb-000043
其中xt为测试样本集内的测试样本;
Figure PCTCN2017114969-appb-000044
是训练样本集中第j类故障中所有训练样本的中心点,dM表示马氏距离,xi为训练样本集中的训练样本,
Figure PCTCN2017114969-appb-000045
是类内协方差矩阵,n是训练样本集中训练样本总数;N为故障类别数;nj为训练样本集中第j类故障类中的训练样本个数;
Figure PCTCN2017114969-appb-000046
为训练样本xi在特征子空间的映射投影。
Figure PCTCN2017114969-appb-000047
为测试样本xt在特征子空间的映射投影。
决策函数的建立即为基于VVRKFA故障诊断模型的构建完成。
应用量子粒子群算法优化VVRKFA数学模型的正则参数与核参数,获取 VVRKFA数学模型的最优正则参数及核参数,步骤为:
(3.b.1)初始化QPSO算法相关参数,包括迭代次数,寻优范围,种群规模,粒子位置和速度。此处每个粒子均为二维向量,第一维为VVRKFA数学模型的正则参数,第二维为VVRKFA数学模型的核参数。
(3.b.2)计算每个粒子的适应度,获取全局最优及局部最优个体。
(3.b.3)更新每个粒子的位置,更新表达式如下式所示:
(3.b.4)重复(3.b.2)和(3.b.3)直至迭代结束,输出结果。
QPSO算法中的粒子位置更新公式为:
Xi(t+1)=Pi'(t)±α|Mbesti(t+1)-Xi(t)|×In(1/u);
式中,
Figure PCTCN2017114969-appb-000048
P′i(t)=Pi(t-1)+(1-β)Pg(t-1),
Figure PCTCN2017114969-appb-000049
N为种群规模,Mbest为所有粒子的个体最优位置的平均点;ωmax是最大惯性权重;ωmin是最小惯性权重,Pj和Pg分别是粒子j个体最优位置和全局最优位置;X是粒子的位置;t为当前的迭代次数,α为压缩扩张因子;u和β是[0,1]之间均匀分布的随机数;P′i是粒子i更新后的位置;Pi是粒子i当前的位置。
所述步骤(4)中,向故障诊断模型输入测试样本数据集,对电路故障类别进行识别,得到测试样本集中每个测试样本所属故障类别,进而获得每类故障的诊断正确率,完成对被测试模拟电路的诊断。
下面以一个实例说明本发明所提出的基于VVRKFA模拟电路故障诊断方法的执行过程及性能。
图2所示为视频放大器电路,各元件的标称值及容差均于图上标出。以此电路为例展示本发明提出的故障诊断方法的整个流程,激励源是幅值为5V, 频率维100Hz的正弦信号,故障响应时域信号在电路输出端采样获得。选择R1、R2、R3、C4、R4、R5、R6、R8和Q1作为测试对象。表1、表2给出了各测试元件的故障码,标称值和故障值,其中↑和↓分别表示故障值高出和低于标称值。无故障情况同样视为一种故障,故障码为F0。为每种故障类别分别采样60个数据,分成2部分,前30个用于建立VVRKFA的故障诊断模型,后30个数据用于测试该故障诊断模型的性能。
表1 视频放大电路参数故障
故障码 故障类 故障值 故障码 故障类 故障值
F1 R2↑ 15kΩ F2 R2↓ 6kΩ
F3 R4↑ 100Ω F4 R4↓ 27Ω
F5 R6↑ 150Ω F6 R6↓
F7 R8↑ 1kΩ F8 R8↓ 50Ω
表2 视频放大电路灾难性故障
故障码 故障值 故障码 故障值
F9 R1开路 F10 R1短路
F11 R3短路 F12 C4开路
F13 R5开路 F14 基极开路
F15 基极-发射极短路 F16 集电极开路
QPSO算法中种群规模和迭代次数分别设置为10和100,最大惯性权重设为1,最小惯性权重为0.5。在仿真中,寻优得到的正则化参数和核宽度因子分别为1.0076×10-4和1.0095。QPSO优化VVRKFA的训练过程如图3。应用寻优得到的最优正则化参数和核宽度因子构建基于VVRKFA的故障诊断模型,输入测试数据进行故障识别。诊断结果如表3所示。经PSO算法参数选择后的GMKL-SVM 故障诊断模型正确地识别了所有的F0,F1,F3,F4,F7,F8,F11,F11,F12,F13,F14,F15和F14故障;错误地将2个F2故障识别为F5故障,错误地将1个F5故障识别为F3故障,错误地将1个F6故障识别为F0故障。可以认为,经QPSO算法优化正则化参数和和宽度因子后的VVRKFA的故障诊断模型,在故障诊断中取得了较好的诊断效果。经计算,模拟电路的故障整体正确诊断率可以达到98.82%。
表3 各故障类别诊断结果
  F0 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16
F0 30                                
F1   30                              
F2     28     2                      
F3       30                          
F4         30                        
F5       1   29                      
F6 1           29                    
F7               30                  
F8                 30                
F9                   30              
F10                     30            
F11                       30          
F12                         30        
F13                           30      
F14                             30    
F15                               30  
F16                                 30

Claims (4)

  1. 基于向量值正则核函数逼近的模拟电路故障诊断方法,其特征在于,包括以下步骤:
    (1)提取被测试模拟电路各节点的时域响应电压信号,即采集输出信号;
    (2)对采集的输出信号进行小波包分解,计算各节点的能量作为原始样本特征数据;并将原始样本特征数据平均划分为训练样本集及测试样本集两部分;
    (3)基于训练样本集,应用量子粒子群算法QPSO优化向量值正则核函数逼近VVRKFA数学模型的正则参数及核参数,构建基于VVRKFA的故障诊断模型;
    (4)向构建的基于VVRKFA的故障诊断模型输入测试样本集,对电路故障类别进行识别。
  2. 根据权利要求1所述的基于VVRKFA的模拟电路故障诊断方法,其特征在于,所述步骤(3)中,基于VVRKFA的故障诊断模型的构建过程为:
    (3.a)确定核函数类型:
    选用高斯核函数
    Figure PCTCN2017114969-appb-100001
    为VVRKFA的核函数,用以建立VVRKFA的数学模型,σ为高斯核函数的宽度因子;
    VVRKFA的数学模型如下:
    Figure PCTCN2017114969-appb-100002
    Figure PCTCN2017114969-appb-100003
    其中,
    Figure PCTCN2017114969-appb-100004
    是回归系数矩阵,目的是将特征内积空间映射到标签空 间;N是故障类别数;m为训练样本集中训练样本的维数;
    Figure PCTCN2017114969-appb-100005
    是训练样本集降维后的降维样本数据集中样本的维数;
    Figure PCTCN2017114969-appb-100006
    是基向量;C是正则参数;
    Figure PCTCN2017114969-appb-100007
    是松弛变量;矩阵
    Figure PCTCN2017114969-appb-100008
    是训练样本集降维后的降维样本数据集
    Figure PCTCN2017114969-appb-100009
    n为训练样本集中所有训练样本总数;K(.,.)是核函数;xi为训练样本集中的训练样本;Yi为训练样本xi的类别标签向量;T为矩阵转置符号;J为目标函数名;[Θ b]=YPT[CI+PPT]-1,其中
    Figure PCTCN2017114969-appb-100010
    Figure PCTCN2017114969-appb-100011
    为训练样本数据集;I为单位矩阵,
    Figure PCTCN2017114969-appb-100012
    是包含训练样本的m个类别标签向量的矩阵;
    (3.b)应用量子粒子群算法优化VVRKFA数学模型,获取VVRKFA数学模型正则参数及核参数的最优值;
    (3.c)以训练样本集中的训练样本xi为输入数据,将步骤(3.b)中获取的最优正则参数及核参数构建如下的向量值映射函数:
    Figure PCTCN2017114969-appb-100013
    此处,
    Figure PCTCN2017114969-appb-100014
    为回归系数矩阵,
    Figure PCTCN2017114969-appb-100015
    为训练样本集降维后的降维样本数据集;K(.,.)是核函数,式(2)中,核函数K(.,.)为步骤(3.b)获取的最优核函数;C是正则参数,式(2)中,正则参数C为步骤(3.b)获取的最优核参数;
    Figure PCTCN2017114969-appb-100016
    是基向量,N为故障类别数;m是训练样本集中训练样本的维数;
    Figure PCTCN2017114969-appb-100017
    是训练样本集降维后的降维样本数据集中样本的维数;n为训练样本集中所有训练样本的总数;
    (3.d)应用步骤(3.c)中构建的映射函数,建立VVRKFA的决策函数,VVRKFA的决策函数表示为:
    Figure PCTCN2017114969-appb-100018
    其中xt为测试样本集内的测试样本;
    Figure PCTCN2017114969-appb-100019
    是训练样本集 中第j类故障中所有训练样本的中心点,dM表示马氏距离,xi为训练样本集中的训练样本,
    Figure PCTCN2017114969-appb-100020
    是类内协方差矩阵,n是训练样本集中训练样本总数;N为故障类别数;nj为训练样本集中第j类故障类中的训练样本个数;
    Figure PCTCN2017114969-appb-100021
    为训练样本xi在特征子空间的映射投影;
    Figure PCTCN2017114969-appb-100022
    为测试样本xt在特征子空间的映射投影;
    决策函数的建立即为基于VVRKFA故障诊断模型的构建完成。
  3. 根据权利要求1或2所述的基于VVRKFA的模拟电路故障诊断方法,其特征在于,所述步骤(3.b)中,应用量子粒子群算法优化VVRKFA数学模型的正则参数与核参数,获取VVRKFA数学模型的最优正则参数及核参数,具体步骤为:
    (3.b.1)初始化QPSO算法参数,包括种速度、位置、种群规模、迭代次数和寻优范围;此处每个粒子均为二维向量,第一维为VVRKFA数学模型的正则参数,第二维为VVRKFA数学模型的核参数;
    (3.b.2)计算粒子的适应度,获取全局最优个体和局部最优个体;
    (3.b.3)更新粒子的速度及位置;
    (3.b.4)重复步骤(3.b.2)和(3.b.3)直至达到最大迭代次数,输出最优参数结果即VVRKFA数学模型的最优正则参数及核参数。
  4. 根据权利要求1或2所述的基于VVRKFA的模拟电路故障诊断方法,其特征在于,所述步骤(4)中,向故障诊断模型输入测试样本数据集,对电路故障类别进行识别,得到测试样本集中每个测试样本所属故障类别,进而获得每类故障的诊断正确率,完成对被测试模拟电路的诊断。
PCT/CN2017/114969 2017-11-09 2017-12-07 基于向量值正则核函数逼近的模拟电路故障诊断方法 WO2019090878A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/497,444 US11287470B2 (en) 2017-11-09 2017-12-07 Vector-valued regularized kernel function approximation based fault diagnosis method for analog circuit

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201711097329.4A CN107884706B (zh) 2017-11-09 2017-11-09 基于向量值正则核函数逼近的模拟电路故障诊断方法
CN201711097329.4 2017-11-09

Publications (1)

Publication Number Publication Date
WO2019090878A1 true WO2019090878A1 (zh) 2019-05-16

Family

ID=61779713

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/114969 WO2019090878A1 (zh) 2017-11-09 2017-12-07 基于向量值正则核函数逼近的模拟电路故障诊断方法

Country Status (3)

Country Link
US (1) US11287470B2 (zh)
CN (1) CN107884706B (zh)
WO (1) WO2019090878A1 (zh)

Cited By (26)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110321627A (zh) * 2019-06-28 2019-10-11 西北农林科技大学 融合叶片光合潜能的光合速率预测方法
CN110837851A (zh) * 2019-10-25 2020-02-25 西安交通大学 一种电静液作动器液压泵故障诊断方法
CN110969206A (zh) * 2019-11-29 2020-04-07 大连理工大学 基于层次划分的电路故障实时诊断与自修复方法
CN111507422A (zh) * 2020-04-24 2020-08-07 西安工程大学 基于cqfpa-wnn的变压器故障诊断方法
CN111610428A (zh) * 2020-04-26 2020-09-01 哈尔滨工业大学 一种基于响应混叠性度量小波包分解算法的参数优化方法
CN111898442A (zh) * 2020-06-29 2020-11-06 西北大学 一种基于多模态特征融合的人体动作识别方法及装置
CN112116017A (zh) * 2020-09-25 2020-12-22 西安电子科技大学 基于核保持的数据降维方法
CN112216352A (zh) * 2020-10-13 2021-01-12 山东建筑大学 基于tdksfa非线性贡献图的间歇过程故障辨识方法及系统
CN112241605A (zh) * 2019-07-17 2021-01-19 华北电力大学(保定) 声振信号构造cnn特征矩阵的断路器储能过程状态辨识方法
CN112308147A (zh) * 2020-11-02 2021-02-02 西安电子科技大学 基于多源域锚适配器集成迁移的旋转机械故障诊断方法
CN112381141A (zh) * 2020-11-13 2021-02-19 西安建筑科技大学 一种空调传感器故障检测方法及系统
CN112396113A (zh) * 2020-11-23 2021-02-23 国家电网公司东北分部 一种电力系统运行方式数据特征两阶段选择方法
CN112580568A (zh) * 2020-12-25 2021-03-30 江苏集萃复合材料装备研究所有限公司 基于高斯核均值局部线性嵌入的芯片缺陷检测方法
CN112668459A (zh) * 2020-12-25 2021-04-16 合肥工业大学 基于有监督对比学习和卷积神经网络的滚动轴承故障诊断方法
CN113283027A (zh) * 2021-05-20 2021-08-20 南京航空航天大学 一种基于知识图谱和图神经网络的机械故障诊断方法
CN113295702A (zh) * 2021-05-20 2021-08-24 国网山东省电力公司枣庄供电公司 电气设备故障诊断模型训练方法和电气设备故障诊断方法
CN113447813A (zh) * 2020-09-03 2021-09-28 鲁能集团有限公司 海上风力发电机组的故障诊断方法及设备
CN113610179A (zh) * 2021-08-17 2021-11-05 安徽容知日新科技股份有限公司 一种设备故障检测分类器训练方法、计算设备及存储介质
CN113657556A (zh) * 2021-09-23 2021-11-16 华北电力大学 基于多元统计分析的燃气轮机进口导叶系统故障诊断方法
CN114167729A (zh) * 2021-12-08 2022-03-11 北京工业大学 基于分布式二段自适应加权堆叠自动编码器的微小故障检测方法及系统
US11287470B2 (en) * 2017-11-09 2022-03-29 Hefei University Of Technology Vector-valued regularized kernel function approximation based fault diagnosis method for analog circuit
CN114325352A (zh) * 2022-01-04 2022-04-12 电子科技大学 一种基于经验小波变换的模拟滤波电路故障诊断方法
CN114330114A (zh) * 2021-12-20 2022-04-12 郑州信大先进技术研究院 基于量子支持向量机的铍青铜合金腐蚀速率预测方法
CN114330067A (zh) * 2021-12-30 2022-04-12 南昌大学 一种软基水闸有限元模型修正方法
CN114722923A (zh) * 2022-03-22 2022-07-08 西北工业大学 一种轻量型的机电设备故障诊断方法
CN114764538A (zh) * 2020-12-30 2022-07-19 河北云酷科技有限公司 一种设备声音信号模式识别模型

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107894564B (zh) * 2017-11-09 2020-02-18 合肥工业大学 一种基于交叉小波特征的模拟电路故障诊断方法
CN109766992B (zh) * 2018-12-06 2020-12-04 北京工业大学 基于深度学习的工控异常检测及攻击分类方法
CN110135121B (zh) * 2019-06-13 2023-04-18 中国人民解放军海军航空大学 一种基于拉格朗日-粒子群更新算法的故障诊断方法
CN111240279B (zh) * 2019-12-26 2021-04-06 浙江大学 一种面向工业不平衡数据的对抗增强故障分类方法
CN111239587A (zh) * 2020-01-20 2020-06-05 哈尔滨工业大学 一种基于frft和lle特征提取的模拟电路故障诊断方法
CN112257335B (zh) * 2020-10-10 2023-04-07 西南交通大学 一种结合pnn和svm的油浸式变压器的故障诊断方法
CN114609994B (zh) * 2022-02-24 2023-11-07 天津大学 基于多粒度正则化重平衡增量学习的故障诊断方法及装置
CN116305733B (zh) * 2022-11-29 2024-05-07 华东理工大学 一种基于全局和局部特征提取的质量相关故障检测方法
CN116070151B (zh) * 2023-03-17 2023-06-20 国网安徽省电力有限公司超高压分公司 基于广义回归神经网络特高压直流输电线路故障检测方法
CN116505526B (zh) * 2023-06-30 2023-08-22 中国电力科学研究院有限公司 基于电压受扰曲线多摆特征的电压稳定分析方法及系统
CN116545853B (zh) * 2023-07-04 2023-10-13 南京理工大学 基于量子粒子群的一体化网络多目标优化资源管理方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101221213A (zh) * 2008-01-25 2008-07-16 湖南大学 基于粒子群算法的模拟电路故障诊断神经网络方法
CN104198924A (zh) * 2014-09-11 2014-12-10 合肥工业大学 一种新颖的模拟电路早期故障诊断方法
CN105548862A (zh) * 2016-01-25 2016-05-04 合肥工业大学 一种基于广义多核支持向量机的模拟电路故障诊断方法
US20160267216A1 (en) * 2015-03-13 2016-09-15 Taiwan Semiconductor Manufacturing Company Limited Methods and systems for circuit fault diagnosis

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009152180A2 (en) * 2008-06-10 2009-12-17 D-Wave Systems Inc. Parameter learning system for solvers
US20100094784A1 (en) * 2008-10-13 2010-04-15 Microsoft Corporation Generalized kernel learning in support vector regression
US8838508B2 (en) * 2011-10-13 2014-09-16 Nec Laboratories America, Inc. Two-stage multiple kernel learning method
JP6206276B2 (ja) * 2014-03-19 2017-10-04 株式会社デンソー 自己診断機能を有する入力回路
CN106597260B (zh) * 2016-12-29 2020-04-03 合肥工业大学 基于连续小波分析和elm网络的模拟电路故障诊断方法
CN107884706B (zh) * 2017-11-09 2020-04-07 合肥工业大学 基于向量值正则核函数逼近的模拟电路故障诊断方法
WO2019144386A1 (zh) * 2018-01-26 2019-08-01 大连理工大学 一种航空发动机过渡态关键性能参数预测方法
CN108414923A (zh) * 2018-02-05 2018-08-17 武汉大学 一种基于深度置信网络特征提取的模拟电路故障诊断方法
US11486925B2 (en) * 2020-05-09 2022-11-01 Hefei University Of Technology Method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101221213A (zh) * 2008-01-25 2008-07-16 湖南大学 基于粒子群算法的模拟电路故障诊断神经网络方法
CN104198924A (zh) * 2014-09-11 2014-12-10 合肥工业大学 一种新颖的模拟电路早期故障诊断方法
US20160267216A1 (en) * 2015-03-13 2016-09-15 Taiwan Semiconductor Manufacturing Company Limited Methods and systems for circuit fault diagnosis
CN105548862A (zh) * 2016-01-25 2016-05-04 合肥工业大学 一种基于广义多核支持向量机的模拟电路故障诊断方法

Cited By (42)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11287470B2 (en) * 2017-11-09 2022-03-29 Hefei University Of Technology Vector-valued regularized kernel function approximation based fault diagnosis method for analog circuit
CN110321627A (zh) * 2019-06-28 2019-10-11 西北农林科技大学 融合叶片光合潜能的光合速率预测方法
CN110321627B (zh) * 2019-06-28 2022-10-14 西北农林科技大学 融合叶片光合潜能的光合速率预测方法
CN112241605A (zh) * 2019-07-17 2021-01-19 华北电力大学(保定) 声振信号构造cnn特征矩阵的断路器储能过程状态辨识方法
CN112241605B (zh) * 2019-07-17 2023-12-19 华北电力大学(保定) 声振信号构造cnn特征矩阵的断路器储能过程状态辨识方法
CN110837851A (zh) * 2019-10-25 2020-02-25 西安交通大学 一种电静液作动器液压泵故障诊断方法
CN110969206B (zh) * 2019-11-29 2023-06-06 大连理工大学 基于层次划分的电路故障实时诊断与自修复方法
CN110969206A (zh) * 2019-11-29 2020-04-07 大连理工大学 基于层次划分的电路故障实时诊断与自修复方法
CN111507422B (zh) * 2020-04-24 2023-04-07 西安工程大学 基于cqfpa-wnn的变压器故障诊断方法
CN111507422A (zh) * 2020-04-24 2020-08-07 西安工程大学 基于cqfpa-wnn的变压器故障诊断方法
CN111610428A (zh) * 2020-04-26 2020-09-01 哈尔滨工业大学 一种基于响应混叠性度量小波包分解算法的参数优化方法
CN111898442A (zh) * 2020-06-29 2020-11-06 西北大学 一种基于多模态特征融合的人体动作识别方法及装置
CN111898442B (zh) * 2020-06-29 2023-08-11 西北大学 一种基于多模态特征融合的人体动作识别方法及装置
CN113447813A (zh) * 2020-09-03 2021-09-28 鲁能集团有限公司 海上风力发电机组的故障诊断方法及设备
CN112116017B (zh) * 2020-09-25 2024-02-13 西安电子科技大学 基于核保持的图像数据降维方法
CN112116017A (zh) * 2020-09-25 2020-12-22 西安电子科技大学 基于核保持的数据降维方法
CN112216352A (zh) * 2020-10-13 2021-01-12 山东建筑大学 基于tdksfa非线性贡献图的间歇过程故障辨识方法及系统
CN112308147A (zh) * 2020-11-02 2021-02-02 西安电子科技大学 基于多源域锚适配器集成迁移的旋转机械故障诊断方法
CN112308147B (zh) * 2020-11-02 2024-02-09 西安电子科技大学 基于多源域锚适配器集成迁移的旋转机械故障诊断方法
CN112381141A (zh) * 2020-11-13 2021-02-19 西安建筑科技大学 一种空调传感器故障检测方法及系统
CN112396113A (zh) * 2020-11-23 2021-02-23 国家电网公司东北分部 一种电力系统运行方式数据特征两阶段选择方法
CN112580568A (zh) * 2020-12-25 2021-03-30 江苏集萃复合材料装备研究所有限公司 基于高斯核均值局部线性嵌入的芯片缺陷检测方法
CN112580568B (zh) * 2020-12-25 2024-03-12 江苏集萃复合材料装备研究所有限公司 基于高斯核均值局部线性嵌入的芯片缺陷检测方法
CN112668459A (zh) * 2020-12-25 2021-04-16 合肥工业大学 基于有监督对比学习和卷积神经网络的滚动轴承故障诊断方法
CN112668459B (zh) * 2020-12-25 2022-09-13 合肥工业大学 基于有监督对比学习和卷积神经网络的滚动轴承故障诊断方法
CN114764538B (zh) * 2020-12-30 2024-04-26 河北云酷科技有限公司 一种设备声音信号模式识别方法
CN114764538A (zh) * 2020-12-30 2022-07-19 河北云酷科技有限公司 一种设备声音信号模式识别模型
CN113295702A (zh) * 2021-05-20 2021-08-24 国网山东省电力公司枣庄供电公司 电气设备故障诊断模型训练方法和电气设备故障诊断方法
CN113283027A (zh) * 2021-05-20 2021-08-20 南京航空航天大学 一种基于知识图谱和图神经网络的机械故障诊断方法
CN113283027B (zh) * 2021-05-20 2024-04-02 南京航空航天大学 一种基于知识图谱和图神经网络的机械故障诊断方法
CN113610179A (zh) * 2021-08-17 2021-11-05 安徽容知日新科技股份有限公司 一种设备故障检测分类器训练方法、计算设备及存储介质
CN113657556B (zh) * 2021-09-23 2023-12-26 华北电力大学 基于多元统计分析的燃气轮机进口导叶系统故障诊断方法
CN113657556A (zh) * 2021-09-23 2021-11-16 华北电力大学 基于多元统计分析的燃气轮机进口导叶系统故障诊断方法
CN114167729A (zh) * 2021-12-08 2022-03-11 北京工业大学 基于分布式二段自适应加权堆叠自动编码器的微小故障检测方法及系统
CN114167729B (zh) * 2021-12-08 2024-05-28 北京工业大学 基于分布式二段自适应加权堆叠自动编码器的微小故障检测方法及系统
CN114330114A (zh) * 2021-12-20 2022-04-12 郑州信大先进技术研究院 基于量子支持向量机的铍青铜合金腐蚀速率预测方法
CN114330114B (zh) * 2021-12-20 2024-03-22 郑州信大先进技术研究院 基于量子支持向量机的铍青铜合金腐蚀速率预测方法
CN114330067A (zh) * 2021-12-30 2022-04-12 南昌大学 一种软基水闸有限元模型修正方法
CN114325352A (zh) * 2022-01-04 2022-04-12 电子科技大学 一种基于经验小波变换的模拟滤波电路故障诊断方法
CN114325352B (zh) * 2022-01-04 2023-04-18 电子科技大学 一种基于经验小波变换的模拟滤波电路故障诊断方法
CN114722923A (zh) * 2022-03-22 2022-07-08 西北工业大学 一种轻量型的机电设备故障诊断方法
CN114722923B (zh) * 2022-03-22 2024-02-27 西北工业大学 一种轻量型的机电设备故障诊断方法

Also Published As

Publication number Publication date
US20210293881A1 (en) 2021-09-23
US11287470B2 (en) 2022-03-29
CN107884706B (zh) 2020-04-07
CN107884706A (zh) 2018-04-06

Similar Documents

Publication Publication Date Title
WO2019090878A1 (zh) 基于向量值正则核函数逼近的模拟电路故障诊断方法
US20200271720A1 (en) Method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation
US11409933B2 (en) Method for diagnosing analog circuit fault based on cross wavelet features
WO2017128455A1 (zh) 一种基于广义多核支持向量机的模拟电路故障诊断方法
WO2018120283A1 (zh) 基于连续小波分析和elm网络的模拟电路故障诊断方法
Shimizu LiNGAM: Non-Gaussian methods for estimating causal structures
CN112200244B (zh) 一种基于递阶对抗训练的航天发动机异常智能检测方法
CN107830996B (zh) 一种飞行器舵面系统故障诊断方法
CN104793124A (zh) 基于小波变换和ica特征提取的开关电路故障诊断方法
CN113252322B (zh) 一种含噪背景下的断路器机械故障诊断方法及装置
Gao et al. A novel fault diagnostic method for analog circuits using frequency response features
CN112596016A (zh) 基于多个一维卷积神经网络集成的互感器故障诊断方法
CN106997373A (zh) 一种基于深度置信网络的链路预测方法
CN108805206A (zh) 一种用于模拟电路故障分类的改进型lssvm建立方法
CN113112039B (zh) 基于时频记忆递归神经网络的主动配电系统初期故障识别方法
Xie et al. Fault diagnosis of analog circuit based on high-order cumulants and information fusion
Alshabrawy et al. Underdetermined blind source separation based on fuzzy c-means and semi-nonnegative matrix factorization
Bakrania et al. Using dimensionality reduction and clustering techniques to classify space plasma regimes
CN109948267B (zh) 基于圆模型参数的线性模拟电路故障诊断方法
CN115019129B (zh) 基于时序成像与图像融合的双有源桥变换器故障诊断方法及系统
D’Souza et al. mSPD-NN: A Geometrically Aware Neural Framework for Biomarker Discovery from Functional Connectomics Manifolds
Cai et al. Application of IWO-SVM approach in fault diagnosis of analog circuits
Deng et al. Diagnosis of incipient faults in nonlinear analog circuits
CN113743363B (zh) 一种基于无人机系统小样本的遮挡目标识别方法
Pan et al. DC-CNN-Based Quantum State Estimation

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17931680

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17931680

Country of ref document: EP

Kind code of ref document: A1