Optimized SVM-based analog circuit fault diagnosis method and system, and storage medium

Abstract

The invention discloses an analog circuit fault diagnosis method and system for optimizing an SVM (Support Vector Machine) and a storage medium. An SVM algorithm is optimized by adopting a fused particle swarm algorithm and an improved grey wolf algorithm. Comprising the following steps: extracting time domain features and wavelet packet features of various fault sample signals of an analog circuit, and fusing the two features as a final fault feature sample set; dividing the fault feature sample set into a training set and a test set according to a certain proportion; optimizing a kernel function and a penalty factor of the SVM by using a fusion particle swarm algorithm and an improved grey wolf algorithm, and constructing a PSO-IGWO-SVM model; and training and testing the PSO-IGWO-SVM model by using the test set, and outputting a fault diagnosis result of the analog circuit. By combining the advantages of the two algorithms, the classification effect and diagnosis precision of the SVM are improved. Meanwhile, by introducing a nonlinear convergence factor and a proportional weight dynamic adjustment strategy, the search efficiency and convergence speed of the algorithm are further improved.

Description

Technical Field

[0001] The present invention relates to the field of electronic engineering, and in particular to an analog circuit fault diagnosis method, system and storage medium which integrate a particle swarm algorithm and an improved grey wolf algorithm to optimize SVM. Background Art

[0002] With the rapid development of electronic technology, analog circuits are widely used in various electronic devices and instruments. However, with the increasing number of electronic devices, the high density, high integration and complexity of analog circuits make fault diagnosis particularly difficult. According to research, 80% of the faults in electronic product circuits are analog circuit faults. Failures will inevitably affect the normal operation of electronic equipment and instruments, causing irreversible losses to production operations. In addition, as an important part of the electronics industry, analog integrated circuits will inevitably have loopholes that are difficult to find manually during the design, manufacturing and packaging process, resulting in serious losses in the process of using the circuit. However, due to the tolerance of analog circuit components such as resistors and capacitors, faults are difficult to diagnose. Therefore, seeking a method that can quickly and accurately realize the fault diagnosis and positioning of analog circuits is of great significance to improving the application safety of instruments and equipment and the quality of integrated circuits.

[0003] Traditional analog circuit fault diagnosis methods mainly rely on expert experience and simulation models, and have problems such as low diagnostic accuracy and low efficiency. Looking at traditional analog circuit fault diagnosis methods, there are mainly two categories: the first method is to simulate the analog circuit first, establish a fault dictionary, and then extract the specific data features of the measured circuit during measurement, using a method similar to "looking up a dictionary" to find the fault point of the analog circuit; the second method is based on the measured data information, using fault verification technology and parameter identification technology to analyze the circuit fault, which belongs to post-test simulation. However, the main problem of traditional diagnosis methods is that they are too dependent on expert experience, and the accuracy of diagnosis depends on the accuracy of simulation. In recent years, more and more research has been conducted to explore the use of artificial intelligence and machine learning methods to further improve the accuracy of analog circuit fault diagnosis. SVM, or support vector machine, has outstanding performance in processing small samples, nonlinear and high-dimensional recognition and classification, so it has been widely used in the field of fault diagnosis, but the SVM classification effect is greatly affected by the kernel function g and the penalty factor C. The parameters selected by the traditional grid search cross-validation method are not accurate enough, which affects the diagnostic effect of SVM.

[0004] To this end, scholars at home and abroad have introduced intelligent optimization algorithms to optimize SVM. In 2024, Chen Xiaomei et al. extracted the permutation entropy of IMF after adaptive variational mode decomposition and used Harris Eagle to optimize the support vector machine. This method has a certain effect on improving diagnostic accuracy. However, due to the limitations of the search strategy, the Harris Eagle algorithm sometimes falls into local optimality. In 2023, Wang Li et al. proposed a method of optimizing SVM by integrating particle swarm and particle filter algorithms, which greatly improved the diagnostic effect. In 2019, Xiong Kui et al. proposed an improved gray wolf algorithm to optimize SVM, which can effectively prevent the population from falling into local optimality, and the convergence speed has been greatly improved. The diagnostic effect improved by optimizing SVM is still the goal pursued in this field. Summary of the invention

[0005] In view of this, the present invention adopts a method combining the Grey Wolf Algorithm (GWO) and the Particle Swarm Optimization (PSO) to perform optimization and optimize the SVM to fill the blank of the application of this method in the field of analog circuit fault diagnosis.

[0006] To solve the above problems, the present invention proposes an analog circuit fault diagnosis method that integrates particle swarm algorithm and improved grey wolf algorithm to optimize SVM. The method comprises the following steps:

[0007] S1 extracts the time domain features and wavelet packet features of various fault sample signals of the analog circuit, and fuses the two features as the final fault feature sample set;

[0008] S2 divides the fault feature sample set into a training set and a test set according to a certain ratio;

[0009] S3 uses the fusion particle swarm algorithm and improved grey wolf algorithm to optimize the kernel function g and penalty factor C of SVM and construct the PSO-IGWO-SVM model;

[0010] S4 inputs the test set into the PSO-IGWO-SVM model, trains and tests the PSO-IGWO-SVM model, and outputs the fault diagnosis results of the simulation circuit.

[0011] This method combines the global search capability of the particle swarm algorithm and the local search capability of the gray wolf algorithm. By introducing a nonlinear convergence factor and a proportional weight dynamic adjustment strategy, the search efficiency and accuracy of the algorithm are improved, and the parameters of the SVM are optimized to improve the accuracy and efficiency of fault diagnosis.

[0012] The present invention integrates a particle swarm algorithm and an improved grey wolf algorithm to optimize the SVM analog circuit fault diagnosis method. The time domain characteristics include maximum value, minimum value, mean value, peak-to-peak value, root mean square, root square amplitude, standard deviation, kurtosis and skewness.

[0013] The present invention integrates the particle swarm algorithm and the improved gray wolf algorithm to optimize the SVM analog circuit fault diagnosis method, wherein the 3, the integration of the particle swarm algorithm and the improved gray wolf algorithm includes:

[0014] (1) Introducing nonlinear convergence factors to optimize the algorithm’s search capabilities;

[0015] (2) A dynamic adjustment strategy of the proportional weight of location updates is introduced to distinguish the impact of wolf level on location updates.

[0016] Preferably, the steps include:

[0017] 1) Use Logistic-Tent composite chaotic mapping to initialize the population;

[0018] 2) Initialize the nonlinear convergence factors A and C, and randomly generate the positions of α wolf, β wolf, and δ wolf;

[0019] 3) Calculate the fitness value, and define the three with the highest fitness values ​​as α wolf, β wolf, and δ wolf;

[0020] 4) Update the convergence factor A according to the nonlinear convergence factor update formula;

[0021] 5) Dynamically adjust the strategy to update the gray wolf's position according to the proportional weight of the position update;

[0022] 6) Determine whether the iteration termination condition is met. If so, output the optimal value; if not, continue iterating by updating the inertia weight, speed and position until the condition is met.

[0023] The present invention also provides an analog circuit fault diagnosis system based on the above method, which includes a data acquisition module, a feature extraction module, and a model training and fault diagnosis module, wherein: the data acquisition module is used to collect the original fault data of the analog circuit; the feature extraction module is used to extract the features of the analog circuit fault data, that is, the time domain features and the wavelet packet features, and fuse the two features; the model training and fault diagnosis module is used to optimize the parameters of the SVM by using the fusion particle swarm algorithm and the improved grey wolf algorithm, construct a training PSO-IGWO-SVM model, perform fault diagnosis on the test set data and output the fault diagnosis results.

[0024] And a computer-readable storage medium having a computer program stored thereon, the program being designed to execute the analog circuit fault diagnosis method based on optimized SVM.

[0025] Compared with the prior art, the analog circuit fault diagnosis method of the present invention which integrates the particle swarm algorithm and the improved grey wolf algorithm to optimize the SVM has at least the following beneficial effects:

[0026] 1. Improve diagnostic accuracy: By integrating the particle swarm algorithm and the improved grey wolf algorithm, the particle swarm algorithm (PSO) is combined with the improved grey wolf algorithm (IGWO) to optimize the selection of SVM parameters and improve the classification effect and diagnostic accuracy of the model.

[0027] 2. Improve search efficiency: Through algorithm fusion, the nonlinear convergence factor and proportional weight dynamic adjustment strategy are introduced to better balance the global search and local search capabilities of the algorithm, and improve the convergence speed and accuracy of the algorithm's search.

[0028] 3. Chaotic map initialization: Logistic-Tent composite chaotic map is used to initialize the population. This initialization method can generate more uniform random sequences in a short time, which helps the algorithm to escape from the local optimum and improve the global search capability.

[0029] 4. Enhanced robustness: Through algorithm fusion, Logistic-Tent composite chaotic mapping is used to initialize the population and generate uniformly distributed random sequences to enhance the robustness of the algorithm.

[0030] 5. Feature extraction and fusion: By extracting the time domain features and wavelet packet features of the original circuit fault data and fusing them as the final fault feature sample set, the feature dimension of fault diagnosis is increased, which helps to improve the accuracy of fault diagnosis.

[0031] 6. Wide application: The present invention solves the problems of inaccurate SVM parameter optimization and limited diagnostic accuracy in the prior art. It can be applied to analog circuit fault diagnosis of various electronic equipment and instruments, pioneering the application in the field of analog circuit fault diagnosis and has broad application prospects.

[0032] The analog circuit fault diagnosis method based on optimized SVM of the present invention is further described below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of fault diagnosis of analog circuit based on optimized SVM of the present invention;

[0034] Figure 2 is the variation graph of the nonlinear convergence factor a;

[0035] Figure 3 The figure is a flow chart of the PSO-IGWO algorithm in the analog circuit fault diagnosis method based on optimized SVM of the present invention. DETAILED DESCRIPTION

[0036] The present invention proposes an analog circuit fault diagnosis method that combines particle swarm optimization and improved gray wolf algorithm to optimize SVM. First, time domain features and wavelet packet features are extracted from sample signals of various faults, the two features are fused as the final fault feature sample set, and the sample set is divided into a training set and a test set according to a certain ratio; secondly, the optimal parameters of SVM are found by fusing particle swarm optimization and improved gray wolf algorithm, and a PSO-IGWO-SVM model is constructed; finally, the PSO-IGWO-SVM is used to diagnose and classify the test set data to obtain the fault diagnosis result.

[0037] The present invention optimizes SVM by integrating particle swarm algorithm and improved grey wolf algorithm to make it suitable for application in analog circuit fault diagnosis method. The analog circuit fault diagnosis method based on optimized SVM of the present invention mainly includes:

[0038] 1. Data collection and preprocessing

[0039] First, connect the analog circuit to be tested with the signal generator and power supply, and input the excitation signal to the circuit to be tested through the signal generator. The output end of the circuit to be tested is connected to the A / D converter, and the output analog voltage signal is converted into a digital quantity and input into the data acquisition card. The data in the acquisition card is transmitted to the PC through the serial port, and finally the circuit output voltage signal is stored in the PC to complete the collection of the original fault data.

[0040] 2. Feature extraction

[0041] The collected original fault data is subjected to feature extraction, including time domain features and wavelet packet features. The time domain features include 9 dimensions, including maximum value, minimum value, mean value, peak-to-peak value, root mean square value, root square amplitude, standard deviation, kurtosis, and skewness. The wavelet packet features use the db3 wavelet basis function to perform a three-layer wavelet packet decomposition on the output voltage signal, and extract the energy of each node in the third layer as a feature.

[0042] 3. Dataset Division

[0043] The extracted fault features are fused to form the final fault feature sample set. Then, the sample set is divided into a training set and a test set in a ratio of 6:4.

[0044] 4. Algorithm optimization and model construction

[0045] The method of integrating particle swarm optimization and improved grey wolf algorithm is used to optimize the kernel function g and penalty factor C of SVM. The specific steps are as follows:

[0046] Initialization: Logistic-Tent composite chaotic mapping is used to initialize the population and generate a uniformly distributed random sequence.

[0047] Nonlinear convergence factor: A nonlinear convergence factor is introduced to dynamically adjust the value of a according to the number of iterations to better balance the global search and local search capabilities.

[0048] Position update: Based on the position update formula of the gray wolf algorithm and the speed update concept of the particle swarm algorithm, the position update is performed. At the same time, a proportional weight dynamic adjustment strategy is introduced to distinguish the different effects of α wolf, β wolf, and δ wolf on position update.

[0049] Iterative optimization: Find the optimal parameters (g, C) of SVM through iterative calculation.

[0050] Construct the PSO-IGWO-SVM model and input the training set into the model for training.

[0051] 5. Troubleshooting

[0052] The test set is input into the trained PSO-IGWO-SVM model, and the fault diagnosis result is output. The effectiveness and accuracy of the present invention are verified by comparing the actual fault with the diagnosis result.

[0053] like Figure 1 FIG. 1 is a flow chart of a specific embodiment of the analog circuit fault diagnosis method based on optimized SVM of the present invention. In this specific embodiment, it includes:

[0054] Step 1: extract the time domain features of the original circuit fault data and the node energy features of the three-layer wavelet packet decomposition;

[0055] Step 2: Fault features are integrated and divided into training set and test set in a ratio of 6:4;

[0056] Step 3: Input the training set into the support vector machine for training. The fusion particle swarm algorithm and improved gray wolf algorithm are used to optimize the kernel function g and penalty factor C of SVM, combining the advantages of both particle swarm and gray wolf algorithms, and introducing nonlinear convergence factors and proportional weight dynamic adjustment strategies to find the optimal parameters.

[0057] Step 4: Input the test set into the trained PSO-IGWO-SVM and output the diagnosis results.

[0058] Furthermore, in the acquisition of the original data of the circuit in step 1, the circuit under test is connected to a signal generator and a power supply, the signal generator inputs an excitation signal into the circuit under test, the output end of the circuit under test is connected to an A / D converter, the output analog voltage signal is converted into a digital quantity and input into a data acquisition card, the data in the acquisition card is transmitted to the PC through a serial port, and finally the circuit output voltage signal is stored on the PC, thereby completing the acquisition of the original fault data.

[0059] Extract time domain features: Extract time domain features of 9 dimensions, including maximum value, minimum value, mean value, peak-to-peak value, root mean square, root square amplitude, standard deviation, kurtosis, and skewness.

[0060] Extract wavelet packet features: Extract wavelet packet features, use the db3 wavelet basis function with better frequency band decomposition effect to perform three-layer wavelet packet decomposition on the output voltage signal, and extract the energy of each node in the third layer. The specific steps are as follows:

[0061] Step 1: Calculate the energy of each node after three-layer wavelet packet decomposition:

[0062]

[0063] where d 3j (k) represents the jth node after the signal is decomposed by three layers of wavelet packets.

[0064] Step 2: Calculate the total energy:

[0065]

[0066] Step 3: Calculate the percentage of total energy for each node:

[0067]

[0068] Step 4: Construct a feature vector P that can characterize the waveform energy of the fault state:

[0069] P=[P 30 , P 31 ,…,P 37 ]

[0070] Further, in step 3, the gray wolf algorithm is improved, and the improvement strategy of the gray wolf algorithm is as follows:

[0071] (1) Introducing nonlinear convergence factor:

[0072] In the gray wolf algorithm, when |A|≤1, the wolf pack narrows the search range and conducts concentrated attacks on the prey, that is, local search; when |A|>1, the wolf pack will expand the search range and conduct global search. Therefore, the value of A determines the global and local search capabilities of the population. From A=2ar1-a, it can be seen that the value of A is controlled by the value of a. a decreases linearly from 2 to 0 with the increase of the number of iterations, but the change of the gray wolf population is not linear during the convergence process of the algorithm. The linearly decreasing convergence factor cannot fully reflect the actual optimization search process. a decreases at the same rate during the iteration process, and it is impossible to take into account the global search and local search well. Therefore, a nonlinear decreasing convergence factor is designed. In the early stage of the iteration, a decreases at a smaller rate, keeping A at a larger value, which is conducive to the algorithm's large-scale search; in the later stage, a decreases faster, so that A has a longer time to maintain a smaller value, ensuring that the algorithm performs a more refined search. The Sigmoid function is used to introduce nonlinear factors. The expression of the Sigmoid function is:

[0073]

[0074] The calculation formula of nonlinear convergence factor is:

[0075]

[0076] a start and a end denote the starting value and final value of a, which are 2 and 0 respectively, and T is the maximum number of iterations. When T = 200, the change of a is as follows: Figure 2 As shown in the figure, it can be seen that at the beginning of the iteration, a decreases slowly, then decreases rapidly in the middle of the evolution, and then decreases gently at the end. In this way, the optimization search process can be better controlled, and the global and local search capabilities can be effectively coordinated.

[0077] (2) Dynamic adjustment strategy of the proportional weight of location update:

[0078] The position update of the ω wolf in the GWO algorithm is based on the distance between the ω wolf and the α wolf, β wolf, and δ wolf. However, the algorithm does not reflect the different levels of the α, β, and δ wolves. This average effect affects the convergence speed of the population optimization. Therefore, the proportional weight is introduced to distinguish the different degrees of influence of the positions of the α, β, and δ wolves on the position update of the ω wolf. The proportional weights w1, w2, and w3 of the ω wolf position update are as follows:

[0079]

[0080] The final position update formula of ω wolf is:

[0081]

[0082] ε is a constant, with a value of 10-5 , x1, x2, and x3 are the updated positions of α, β, and δ wolves respectively.

[0083] Furthermore, in step 3, the particle swarm algorithm and the improved grey wolf algorithm are integrated to further balance the global search and local search capabilities of the algorithm, and the PSO algorithm and the improved GWO algorithm are combined. The new position update formula is as follows:

[0084] v(t+1)=wv(t)+c1r1[x1-x(t)]+c2r2[x2-x(t)]+c3r3[x3-x(t)] (3)

[0085] x(t+1)=x(t)+v(t+1) (4)

[0086] The steps of the fusion algorithm are as follows, and the flow chart is as follows Figure 3 As shown:

[0087] Step 1: Use the Logistic-Tent composite chaotic map to initialize the population. Single chaotic map often leads to uneven distribution of solution space due to its own shortcomings. Therefore, the present invention proposes to fuse the Logistic chaotic map and the Tent chaotic map to generate the Logistic-Tent composite chaotic map. This chaotic system combines the complex dynamic characteristics of the Logistic system with the fast iteration speed and strong autocorrelation of the Tent chaotic system, so that the system can generate more and more uniform random sequences in a short time compared to a single chaotic system. It is defined as:

[0088]

[0089] Where: x i+1 is the generated composite chaotic sequence, x i ∈[0,1]; r is the control parameter, r∈(0,4); mod is the modulo operation.

[0090] Step 2: Initialize the nonlinear convergence factor of the fundamental formula (1), and initialize A and C, and randomly generate the positions of α, β, and δ wolves;

[0091] Step 3, calculate the fitness value, define the three with the highest fitness values ​​as α, β, and δ wolves, and update the selection of α, β, and δ wolves;

[0092] Step 4: Update the convergence factor according to formula (1) and update the gray wolf position according to formula (2)

[0093] Step 5: Determine whether the iteration termination condition is met. If so, output the optimal value. Otherwise, jump back to step 3 by updating the inertia weight, speed and position, reach the maximum number of iterations, stop the loop, and output the optimal value.

[0094] In summary, the algorithm fusion proposed in the present invention is an analog circuit fault diagnosis method that integrates particle swarm algorithm and improved gray wolf algorithm to optimize SVM. The core idea is that the gray wolf algorithm wolves, wolves and wolves guide the population to update the position, which makes up for the defect of the single optimal individual guiding the population to iteratively optimize the particle swarm algorithm; at the same time, the gray wolf algorithm optimizes in an encircling manner to approach the optimal value, which can effectively avoid the problem of premature convergence of the particle swarm algorithm; on the other hand, the particle swarm defines the particle swarm speed, and introduces the concept of population speed update on the basis of the gray wolf algorithm, which is conducive to improving the convergence speed. Therefore, the fusion of particle swarm algorithm and gray wolf algorithm can effectively avoid the defects of particle swarm being prone to local optimality and premature convergence, and can also improve the problem of slow convergence of gray wolf algorithm. In addition, the present invention improves the algorithm initialization strategy, introduces a nonlinear convergence factor to balance the global and local search capabilities, proposes a dynamic adjustment strategy for the proportional weight of position update, distinguishes the influence of wolf levels on position update to different degrees, and improves the original gray wolf algorithm. The optimal solution position obtained by integrating the particle swarm algorithm and the improved gray wolf algorithm is the hyperparameter (g, C) of the support vector machine. Applying this method to find the optimal hyperparameters improves the diagnostic accuracy of the SVM model.

[0095] By combining the advantages of the two algorithms, the classification effect and diagnostic accuracy of SVM are improved. At the same time, by introducing a nonlinear convergence factor and a proportional weight dynamic adjustment strategy, the search efficiency and convergence speed of the algorithm are further improved. The embodiments of the present invention have verified its effectiveness and accuracy, and have broad application prospects and promotion value.

[0096] The embodiments described above are merely descriptions of preferred implementation modes of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for fault diagnosis of analog circuits based on optimized SVM, characterized in that: include: S1 extracts the time domain features and wavelet packet features of various fault sample signals of the analog circuit, and fuses the two features as the final fault feature sample set; S2 divides the fault feature sample set into a training set and a test set according to a certain ratio; S3 uses the fusion particle swarm algorithm and improved grey wolf algorithm to optimize the kernel function g and penalty factor C of SVM and construct the PSO-IGWO-SVM model; S4 inputs the test set into the PSO-IGWO-SVM model, trains and tests the PSO-IGWO-SVM model, and outputs the fault diagnosis results of the simulation circuit.

2. The analog circuit fault diagnosis method based on optimized SVM according to claim 1 is characterized in that: The time domain features include maximum value, minimum value, mean value, peak-to-peak value, root mean square, root square amplitude, standard deviation, kurtosis and skewness.

3. The analog circuit fault diagnosis method based on optimized SVM according to claim 1 is characterized in that: The fusion particle swarm algorithm and improved grey wolf algorithm include: (1) Introducing nonlinear convergence factors to optimize the algorithm’s search capabilities; (2) A dynamic adjustment strategy of the proportional weight of location updates is introduced to distinguish the impact of wolf level on location updates.

4. The analog circuit fault diagnosis method based on optimized SVM according to claim 1 is characterized in that: The wavelet packet feature is obtained by performing three-layer wavelet packet decomposition on the output voltage signal through the db3 wavelet basis function, and extracting the energy feature of each node in the third layer.

5. The analog circuit fault diagnosis method based on optimized SVM according to claim 1 or 2, characterized in that: The fusion of particle swarm algorithm and improved grey wolf algorithm comprises the following steps: 1) Use Logistic-Tent composite chaotic mapping to initialize the population; 2) Initialize the nonlinear convergence factors A and C, and randomly generate the positions of α wolf, β wolf, and δ wolf; 3) Calculate the fitness value, and define the three with the highest fitness values ​​as α wolf, β wolf, and δ wolf; 4) Update the convergence factor A according to the nonlinear convergence factor update formula; 5) Dynamically adjust the strategy to update the gray wolf's position according to the proportional weight of the position update; 6) Determine whether the iteration termination condition is met. If so, output the optimal value; if not, continue iterating by updating the inertia weight, speed and position until the condition is met.

6. The method according to claim 5, characterized in that The update of the nonlinear convergence factor A adopts the Sigmoid function to introduce nonlinear factors, so that A decreases at a small rate in the early stage of iteration, decreases faster in the middle stage of iteration, and decreases gently at the end of iteration, so as to better balance the global search and local search capabilities.

7. The method according to claim 5, characterized in that The dynamic adjustment strategy of the proportion weight of the position update distinguishes the influence of the different levels of the α wolf, the β wolf and the δ wolf in the wolf pack on the position update, so as to improve the convergence speed of the algorithm.

8. The method according to claim 1, characterized in that: The kernel function g and penalty factor C of the SVM model are determined by the optimal solution obtained by fusing the particle swarm algorithm and the improved grey wolf algorithm to improve the classification effect and diagnosis accuracy of the SVM model.

9. An analog circuit fault diagnosis system based on the method according to any one of claims 1 to 7, characterized in that It includes data acquisition module, feature extraction module, model training and fault diagnosis module, among which: The data acquisition module is used to collect the original fault data of the analog circuit; The feature extraction module is used to extract the features of the simulated circuit fault data, namely, the time domain features and the wavelet packet features, and fuse the two features; The model training and fault diagnosis module is used to optimize the parameters of SVM by fusion particle swarm algorithm and improved grey wolf algorithm, build and train PSO-IGWO-SVM model, perform fault diagnosis on test set data and output fault diagnosis results.

10. A computer-readable storage medium having a computer program stored thereon, wherein the program is designed to execute the analog circuit fault diagnosis method based on optimized SVM as claimed in any one of claims 1 to 8.

Images