Analog circuit fault diagnosis method based on GOA and SVM
By using a GOA-SVM-based method, feature vectors are extracted through wavelet packet decomposition and singular value decomposition, and SVM parameters are optimized. This solves the problem of low accuracy in analog circuit fault diagnosis and achieves efficient fault identification and diagnosis.
Patent Information
- Application Number
- CN202410181000.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-02-18
- Publication Date
- 2026-02-13
AI Technical Summary
Existing analog circuit fault diagnosis technologies suffer from low accuracy, particularly in feature extraction and pattern recognition, where it is difficult to effectively improve the efficiency of circuit fault diagnosis.
A fault diagnosis model is established by using a GOA-SVM-based method to obtain impulse response data from the output of an analog circuit, performing wavelet packet decomposition and singular value decomposition to extract feature vectors, and then using the GOA algorithm to optimize the penalty factor and kernel function parameters of the SVM.
It significantly improves the accuracy and efficiency of analog circuit fault diagnosis, ensuring the stability and safety of circuit operation.
Smart Images
Figure CN121524769A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of fault diagnosis technology, and more specifically, relates to a fault diagnosis method for analog circuits based on GOA-SVM. Background Technology
[0002] With the development of modern industry, highly integrated and complex analog circuits are widely used in various intelligent instruments and meters. This has significantly increased the complexity of circuits, leading to a higher probability of circuit failures. As the foundation of electronic systems, accurate diagnosis of component faults is crucial for ensuring the stable operation of electronic systems. Analog circuit faults are generally classified into structural faults and parameter variation faults. Due to the inherent nonlinearity and tolerance of circuits, the development of analog circuit fault diagnosis technology has been slow, hindering its effective application in daily life and production. Therefore, to improve the stability and safety of circuit systems, it is necessary to research practical and efficient analog circuit fault diagnosis methods to address these issues.
[0003] The key factors affecting the efficiency of analog circuit fault diagnosis are mainly feature extraction and pattern recognition. Feature extraction is the first problem to be solved in fault diagnosis and has a significant impact on classification efficiency. Feature extraction methods for analog circuit fault diagnosis mainly include wavelet packet transform, principal component analysis, and machine learning. Among these, wavelet packet transform is the most widely used. However, wavelet packets also have problems such as difficulty in selecting wavelet basis functions and the presence of a large amount of redundant information. Therefore, combining wavelet packet transform with singular value decomposition can effectively solve this problem. For pattern recognition, methods mainly include support vector machines (SVMs) and extreme learning machines (ELMs). SVMs are widely used in fault classification. However, in the diagnosis of analog circuits, it is difficult to find suitable penalty factors and kernel function parameters when using SVMs alone. An optimization algorithm can be used to find the optimal parameters to improve the classification performance of SVMs and improve the overall efficiency of analog circuit fault diagnosis. Summary of the Invention
[0004] The purpose of this invention is to solve the problem of low accuracy in existing analog circuit fault diagnosis processes, and to propose an analog circuit fault diagnosis method based on GOA-SVM.
[0005] To achieve the above objectives, according to one aspect of the present invention, a method for diagnosing analog circuit faults based on GOA-SVM is provided, comprising:
[0006] Simulations were performed on the analog circuit under different fault states of the experimental circuit, and the circuit output was used as a test point to obtain the circuit pulse response.
[0007] The circuit impulse response was analyzed using Monte Carlo simulation, and multiple sets of sample data were obtained for each circuit fault state.
[0008] Wavelet packet decomposition is performed on the sample data corresponding to each circuit fault state to obtain the corresponding wavelet coefficients. Singular value decomposition is then performed on the wavelet coefficients to obtain the singular values corresponding to each fault state. The singular values are used as feature vectors, and all feature vectors are combined to construct a sample set. The sample set is then randomly divided into a training set and a test set.
[0009] Based on the training set data, the GOA algorithm is used to find the optimal penalty factor and kernel function parameters of SVM, and a GOA-SVM fault diagnosis model is established.
[0010] The GOA-SVM fault diagnosis model is used to classify faults in the test set data and obtain the diagnostic results of analog circuit fault diagnosis.
[0011] In some optional implementations, the step of using the GOA algorithm to find the optimal penalty factor and kernel function parameters of the SVM based on the training set data, and establishing a GOA-SVM fault diagnosis model, includes:
[0012] Initialize the GOA algorithm parameters, including population size N and maximum number of iterations T. max Control parameter c max ,c min The spatial dimension dim and variable range are used to map the penalty factor and kernel parameters to be optimized to different dimensions of the locust's position.
[0013] The population positions are initialized, the fitness value corresponding to each locust position is calculated, and the optimal position of the current locust individual and the fitness value corresponding to the optimal position are calculated based on the obtained fitness value.
[0014] Calculate and update the current control parameter c of the GOA algorithm, calculate and update the current position of the locust, calculate the individual fitness value and update the optimal position of the locust and the fitness value corresponding to the optimal position;
[0015] Determine if the current number of iterations of the GOA algorithm has reached the maximum number of iterations T. max If so, the parameters corresponding to the different dimensions of the obtained optimal position are used as the penalty factor and kernel function parameters of the SVM; otherwise, return to the steps of initializing the population position, calculating the fitness value corresponding to the position of each locust, and calculating the optimal position of the current locust individual and the fitness value corresponding to the optimal position based on the obtained fitness value.
[0016] A GOA-SVM fault diagnosis model is established using the obtained optimal penalty factor and kernel function parameters.
[0017] In some alternative implementations, by Update the control parameter c, where c max and c min Let T represent the maximum and minimum values of the decreasing interval, respectively. max t represents the maximum number of iterations, and t represents the current number of iterations.
[0018] In some alternative implementations, by Update the current position of the locusts, where d represents the dimension of the variable, and i and j represent the individual locust numbers. This represents the position of the i-th locust in the d-th dimension, where d = (1, 2, ..., dim), c represents the control parameter, and ub d and lb d Let T represent the upper and lower bounds of the locust in d-dimensional space, respectively; d Indicates the optimal locust individual position; Let represent the Euclidean distance between the j-th locust and the i-th locust; s() represents the mutual influence function between the two locusts.
[0019] In some optional implementations, the step of performing wavelet packet decomposition on the sample data corresponding to each circuit fault state to obtain the corresponding wavelet coefficients, and performing singular value decomposition on the wavelet coefficients to obtain the singular values corresponding to each fault state, using the singular values as feature vectors, and combining all feature vectors to construct a sample set, including:
[0020] The circuit impulse response data collected at the output of the analog circuit is decomposed into three layers of wavelet packets using the Dmeyer wavelet basis function. The coefficients of each sample signal at each node of the third layer of wavelet packet are obtained. Then, singular value decomposition is performed on each node coefficient to obtain the singular values corresponding to each node coefficient and use them as feature vectors. The singular values of all fault sample signals are combined to form the feature vector sample set for analog circuit fault diagnosis.
[0021] In some alternative implementations, the control parameter c of the GOA algorithm is designed to decrease linearly, enabling the algorithm to have dynamic search capabilities and to balance the algorithm's local exploration and global exploration capabilities based on parameter updates.
[0022] In some alternative implementations, by Determine the interaction function s() between locust individuals, where f and l represent the attraction intensity parameter and attraction scale parameter, respectively, and r represents the region range.
[0023] In some alternative implementations, f and l take values of 0.5 and 1.5, respectively. When s(r) > 0, the range of r represents the attraction zone; when s(r) < 0, the range of r represents the repulsion zone; and when s(r) = 0, the range of r represents the comfort zone.
[0024] In some optional implementations, the process of establishing the GOA-SVM fault diagnosis model involves mapping the spatial dimension of the optimal individual location to a penalty factor and kernel function parameters, thereby optimizing the diagnosis model.
[0025] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.
[0026] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:
[0027] This invention proposes an analog circuit fault diagnosis method based on the Grasshopper Optimization Algorithm (GOA) and Support Vector Machines (SVM). Its purpose is to effectively improve the efficiency and accuracy of analog circuit fault diagnosis, thereby enhancing the stability and safety of circuit operation. The GOA algorithm is an open-source heuristic algorithm that simulates the migration and foraging behavior of grasshoppers in nature. This algorithm features few parameter adjustments, high search efficiency, and fast convergence speed. Furthermore, it can effectively balance the local and global search processes based on its adaptive mechanism, achieving an efficient parameter optimization process. Attached Figure Description
[0028] Figure 1 This is a flowchart of a method provided in an embodiment of the present invention;
[0029] Figure 2 This is a flowchart of a GOA algorithm provided in an embodiment of the present invention;
[0030] Figure 3 This is a circuit diagram of a bandpass filter circuit described in an experiment provided by an embodiment of the present invention;
[0031] Figure 4 This is a test result diagram obtained by using the GOA-SVM fault diagnosis model according to an embodiment of the present invention;
[0032] Figure 5 This is a fitness curve obtained by diagnosis using the method of the present invention, as provided in an embodiment of the present invention. Detailed Implementation
[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0034] Example 1
[0035] Combination Figure 1 and Figure 2 This embodiment describes a fault diagnosis method for analog circuits based on GOA-SVM, specifically including:
[0036] (1) The simulation circuit under different fault states of the experimental circuit was simulated using PSPICE software, and the circuit output terminal was used as the test point to obtain the circuit pulse response.
[0037] (2) Analyze the circuit impulse response using Monte Carlo and obtain multiple sets of sample data for each circuit fault state;
[0038] (3) Perform wavelet packet decomposition on the sample data corresponding to each circuit fault state to obtain the corresponding wavelet coefficients, and perform singular value decomposition on the wavelet coefficients to obtain the singular values corresponding to each fault state; use the singular values as feature vectors, and combine all feature vectors to construct a sample set; randomly divide the sample set into a training set and a test set;
[0039] (4) Based on the training set data, the GOA algorithm is used to find the optimal penalty factor and kernel function parameters of SVM, and a GOA-SVM fault diagnosis model is established.
[0040] (5) Use the GOA-SVM fault diagnosis model to classify the test set data for faults and obtain the diagnostic results of analog circuit fault diagnosis.
[0041] Further, the multiple sets of sample signals mentioned in step (3) are subjected to wavelet packet and singular value decomposition respectively, and the singular values obtained from the decomposition are used as feature vectors to construct a singular value feature vector sample set, which is constructed as follows:
[0042] (3.1) Use Dmeyer wavelet basis functions to perform three-level wavelet packet decomposition on the fault sample signal data collected at the output of the analog circuit to obtain the wavelet packet coefficients corresponding to each type of fault.
[0043] (3.2) Reconstruct and sort the third-layer nodes of the wavelet packet to obtain the wavelet packet node coefficients in ascending order of their spectrum;
[0044] (3.3) Perform singular value decomposition on the reconstructed wavelet packet coefficients to obtain the singular values corresponding to each node coefficient, and use the singular values as feature vectors to combine the singular values of all fault sample signals after decomposition into a feature vector sample set for analog circuit fault diagnosis.
[0045] (3.4) The sample set is randomly divided into a training set and a test set according to a certain ratio.
[0046] Furthermore, step (4) describes the process of using the GOA algorithm to find the optimal penalty factor and kernel function parameters of the SVM based on the training set data, and establishing the GOA-SVM fault diagnosis model as follows:
[0047] (4.1) Initialize the GOA algorithm parameters, including population size N and maximum number of iterations T. max Control parameter c max ,c min The spatial dimension dim and variable range are used to map the penalty factor and kernel parameters to be optimized to different dimensions of the locust's position.
[0048] (4.2) Initialize the population position, calculate the fitness value corresponding to the position of each locust, and calculate the optimal position and fitness value of the current locust individual based on the obtained fitness value;
[0049] (4.3) Calculate and update the current control parameter c of the GOA algorithm;
[0050] (4.4) Calculate and update the current position of the locusts;
[0051] (4.5) Calculate the individual fitness value and update the optimal position and fitness value of the locust;
[0052] (4.6) Determine whether the current iteration count of the GOA algorithm has reached the maximum iteration count T. max If yes, proceed to step (4.7); otherwise, return to step (4.2).
[0053] (4.7) Use the parameters corresponding to different dimensions of the optimal position obtained in step (4.5) as the penalty factor and kernel function parameters of SVM;
[0054] (4.8) Establish the GOA-SVM fault diagnosis model using the obtained optimal penalty factor and kernel function parameters.
[0055] In this embodiment, the update control parameter c in step (4.3) is:
[0056]
[0057] Among them, c max and c minLet T represent the maximum and minimum values of the decreasing interval, respectively. max This indicates the maximum number of iterations.
[0058] In this embodiment, the formula for updating the current position of the locust in step (4.4) is:
[0059]
[0060] Where d represents the dimension of the variable, and i and j represent the individual locust numbers. The position of the i-th locust in the d-th dimension is represented by d = (1, 2, ..., dim); c represents the control parameter; ub d and lb d Let T represent the upper and lower bounds of the locust in d-dimensional space, respectively; d Indicates the optimal locust individual position; represents the Euclidean distance between the j-th and i-th locusts; s() represents the mutual influence function between the two locusts.
[0061] In this embodiment, the inter-locust interaction function s() in the formula for updating the current position of the locusts in step (4.4) is specifically expressed as:
[0062]
[0063] Where f and l represent the attraction intensity parameter and attraction scale parameter, respectively, with values of 0.5 and 1.5. Furthermore, when s(r) > 0, the range of r represents the attraction zone; when s(r) < 0, the range of r represents the repulsion zone; and when s(r) = 0, the range of r represents the comfort zone.
[0064] In this embodiment, the process of establishing the GOA-SVM fault diagnosis model involves mapping the spatial dimension of the optimal individual location to a penalty factor and kernel function parameters, thereby optimizing the diagnosis model.
[0065] In this embodiment, the control parameter c of the GOA algorithm calculated in step (4.3) is generally designed to decrease linearly, which enables the algorithm to have dynamic search capabilities and can balance the local development and global exploration capabilities of the algorithm based on parameter updates.
[0066] Example 2
[0067] This invention selects the bandpass filter circuit in classic analog circuits as the research object. This circuit can pass frequency components within a certain frequency range and attenuate frequency components in other ranges to an extremely low level, which can ensure the validity of the sample.
[0068] Diagnostic examples for bandpass filter circuits, combined with Figure 3 The specific steps are as follows:
[0069] (1-1) The bandpass filter circuit was simulated using PSPICE software, and the circuit output terminal was selected as the test point.
[0070] (1-2) Circuit schematic diagram as follows Figure 3 As shown, the circuit includes resistive and capacitive components. Through sensitivity analysis of the circuit components, R1, R6, C1, and C2 were selected as faulty components. The tolerance range for both resistors and capacitors was set to 5%. A component is considered faulty when its parameter value deviates from its nominal value by 50%. Considering the normal nominal value state, nine different fault types were set, denoted by labels F0-F8, and their corresponding fault codes are shown in Table 1 below:
[0071] Table 1
[0072]
[0073]
[0074] (1-3) Perform 100 Monte Carlo analyses for each type of fault state and generate 100 sets of pulse response signals at the circuit output terminal respectively;
[0075] (2-1) Select the Dmeyer wavelet basis function to perform three-level wavelet packet decomposition on the fault sample signal data collected at the output of the analog circuit, and obtain the wavelet packet coefficients corresponding to each type of fault.
[0076] (2-2) Reconstruct and sort the third-layer nodes of the wavelet packet to obtain the wavelet packet node coefficients in ascending order of their spectrum;
[0077] (2-3) Perform singular value decomposition on the reconstructed wavelet packet coefficients to obtain the singular values corresponding to each node coefficient, and use the singular values as feature vectors to combine the singular values of all fault sample signals after decomposition into a feature vector sample set for analog circuit fault diagnosis.
[0078] (2-4) The sample set is randomly divided into training set and test set according to a certain ratio.
[0079] (3-1) Using the fitness value as the accuracy of fault diagnosis, the formula for the accuracy of fault diagnosis is: Where n is the number of accurately identified samples, and N is the number of samples in the test set;
[0080] (3-2) Based on the training set data, the GOA algorithm is used to find the optimal penalty factor and kernel function parameters of SVM, and a GOA-SVM fault diagnosis model is established. The specific process is as follows:
[0081] (3-2-1) Initialize the GOA algorithm parameters, including population size N and maximum number of iterations T. max Control parameter c max ,c min The spatial dimension dim and variable range are used to map the penalty factor and kernel parameters to be optimized to different dimensions of the locust's position.
[0082] (3-2-2) The population position is initialized, the fitness value corresponding to the position of each locust is calculated, and the optimal position and fitness value of the current locust individual are calculated based on the fitness value.
[0083] (3-2-3) Calculate and update the current control parameter c of the GOA algorithm;
[0084] (3-2-4) Calculate and update the current position of the locusts;
[0085] (3-2-5) Calculate the individual fitness value and update the optimal position and fitness value of the locust;
[0086] (3-2-6) Determine whether the current iteration count of the GOA algorithm has reached the maximum iteration count T. max If yes, then execute (3-2-7); otherwise, return to execute (3-2-2).
[0087] (3-2-7) Use the parameters corresponding to the different dimensions of the optimal position obtained in steps four and five as the penalty factor and kernel function parameters of SVM;
[0088] (3-2-8) A GOA-SVM fault diagnosis model is established using the obtained optimal penalty factor and kernel function parameters;
[0089] The formula for updating the control parameter c is:
[0090]
[0091] Among them, c max and c min Let T represent the maximum and minimum values of the decreasing interval, respectively. max Indicates the maximum number of iterations;
[0092] The formula for updating the locust's current position is:
[0093]
[0094] Where d represents the dimension of the variable, and i and j represent the individual locust numbers. The position of the i-th locust in the d-th dimension is represented by d = (1, 2, ..., dim); c represents the control parameter; ub d and lb dLet T represent the upper and lower bounds of the locust in d-dimensional space, respectively; d Indicates the optimal locust individual position; represents the Euclidean distance between the j-th and i-th locusts; s() represents the mutual influence function between the two locusts.
[0095] The function s(), which represents the interaction between individual locusts, is specifically expressed as follows:
[0096]
[0097] Where f and l represent the attraction intensity parameter and attraction scale parameter, respectively, with values of 0.5 and 1.5. Furthermore, when s(r) > 0, the range of r represents the attraction zone; when s(r) < 0, the range of r represents the repulsion zone; and when s(r) = 0, the range of r represents the comfort zone.
[0098] (4) The fault diagnosis model of the constructed GOA-SVM is used to classify the test set sample data for fault diagnosis, and the accuracy of the fault diagnosis of the analog circuit is obtained.
[0099] pass Figure 4 The results shown indicate that the method described in this invention is effective for... Figure 3 The bandpass filter circuit shown is used for fault diagnosis, and the diagnostic accuracy obtained is 99.7222%, which significantly improves the accuracy of fault identification. Through... Figure 5 As shown in the fitness curve during the diagnostic process, the optimal fitness value was reached after 12 iterations, and the diagnostic accuracy reached 99.7222%.
[0100] The present invention has the following advantages:
[0101] (1) In the data preprocessing process, the decomposition method combining wavelet packet transform and singular value can effectively solve the problem of wavelet packets having a large amount of redundant information, thereby improving the efficiency of the feature extraction process.
[0102] (2) The GOA algorithm has the characteristics of simple formula, few adjustment parameters, high search efficiency and fast convergence speed. It can also effectively balance the local search and global search process according to its own adaptive mechanism to achieve efficient parameter optimization process. Furthermore, the fault diagnosis of analog circuits through the GOA-SVM fault diagnosis model can significantly improve the diagnosis efficiency and accuracy.
[0103] It should be noted that, depending on the implementation needs, the various steps / components described in this application can be broken down into more steps / components, or two or more steps / components or parts of the operation of steps / components can be combined into new steps / components to achieve the purpose of this invention.
[0104] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for diagnosing analog circuit faults based on GOA-SVM, characterized in that, include: Simulations were performed on the analog circuit under different fault states of the experimental circuit, and the circuit output was used as a test point to obtain the circuit pulse response. The circuit impulse response was analyzed using Monte Carlo simulation, and multiple sets of sample data were obtained for each circuit fault state. Wavelet packet decomposition is performed on the sample data corresponding to each circuit fault state to obtain the corresponding wavelet coefficients. Singular value decomposition is then performed on the wavelet coefficients to obtain the singular values corresponding to each fault state. The singular values are used as feature vectors, and all feature vectors are combined to construct a sample set. The sample set is then randomly divided into a training set and a test set. Based on the training set data, the GOA algorithm is used to find the optimal penalty factor and kernel function parameters of SVM, and a GOA-SVM fault diagnosis model is established. The GOA-SVM fault diagnosis model is used to classify faults in the test set data and obtain the diagnostic results of analog circuit fault diagnosis.
2. The method according to claim 1, characterized in that, The process involves using the GOA algorithm based on the training set data to find the optimal penalty factor and kernel function parameters for the SVM, and establishing a GOA-SVM fault diagnosis model, including: Initialize the GOA algorithm parameters, including population size N and maximum number of iterations T. max Control parameter c max ,c min The spatial dimension dim and variable range are used to map the penalty factor and kernel parameters to be optimized to different dimensions of the locust's position. The population positions are initialized, the fitness value corresponding to each locust position is calculated, and the optimal position of the current locust individual and the fitness value corresponding to the optimal position are calculated based on the obtained fitness value. Calculate and update the current control parameter c of the GOA algorithm, calculate and update the current position of the locust, calculate the individual fitness value and update the optimal position of the locust and the fitness value corresponding to the optimal position; Determine if the current number of iterations of the GOA algorithm has reached the maximum number of iterations T. max If so, the parameters corresponding to the different dimensions of the obtained optimal position are used as the penalty factor and kernel function parameters of the SVM; otherwise, return to the steps of initializing the population position, calculating the fitness value corresponding to the position of each locust, and calculating the optimal position of the current locust individual and the fitness value corresponding to the optimal position based on the obtained fitness value. A GOA-SVM fault diagnosis model is established using the obtained optimal penalty factor and kernel function parameters.
3. The method according to claim 2, characterized in that, Depend on Update the control parameter c, where c max and c min Let T represent the maximum and minimum values of the decreasing interval, respectively. max t represents the maximum number of iterations, and t represents the current number of iterations.
4. The method according to claim 3, characterized in that, Depend on Update the current position of the locusts, where d represents the dimension of the variable, and i and j represent the individual locust numbers. This represents the position of the i-th locust in the d-th dimension, where d = (1, 2, ..., dim), c represents the control parameter, and ub d and lb d Let T represent the upper and lower bounds of the locust in d-dimensional space, respectively; d Indicates the optimal locust individual position; Let represent the Euclidean distance between the j-th locust and the i-th locust; s() represents the mutual influence function between the two locusts.
5. The method according to claim 1, characterized in that, The process involves performing wavelet packet decomposition on the sample data corresponding to each circuit fault state to obtain the corresponding wavelet coefficients, and then performing singular value decomposition on the wavelet coefficients to obtain the singular values corresponding to each fault state. These singular values are then used as feature vectors, and all feature vectors are combined to construct a sample set, including: The circuit impulse response data collected at the output of the analog circuit is decomposed into three layers of wavelet packets using the Dmeyer wavelet basis function. The coefficients of each sample signal at each node of the third layer of wavelet packet are obtained. Then, singular value decomposition is performed on each node coefficient to obtain the singular values corresponding to each node coefficient and use them as feature vectors. The singular values of all fault sample signals are combined to form the feature vector sample set for analog circuit fault diagnosis.
6. The method according to claim 3, characterized in that, The control parameter c of the GOA algorithm is designed to decrease linearly, which enables the algorithm to have dynamic search capabilities and can balance the local development and global exploration capabilities of the algorithm based on parameter updates.
7. The method according to claim 4, characterized in that, Depend on Determine the interaction function s() between locust individuals, where f and l represent the attraction intensity parameter and attraction scale parameter, respectively, and r represents the region range.
8. The method according to claim 7, characterized in that, f and l take values of 0.5 and 1.5 respectively. When s(r)>0, the range of r represents the attraction zone; when s(r)<0, the range of r represents the repulsion zone; when s(r)=0, the range of r represents the comfort zone.
9. The method according to claim 8, characterized in that, The process of establishing the GOA-SVM fault diagnosis model involves mapping the spatial dimension of the optimal individual location to a penalty factor and kernel function parameters, thereby optimizing the diagnosis model.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.