CN105866664A - Intelligent fault diagnosis method for analog circuit based on amplitude frequency features - Google Patents

Abstract

The invention provides an intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics. The method for intelligent fault diagnosis of analog circuits based on amplitude-frequency characteristics includes the following steps: a. Applying an excitation signal to the circuit to be diagnosed, and collecting the amplitude-frequency response information output by the circuit to be diagnosed as fault feature information; b. Analyze and process the amplitude-frequency response information, and obtain the optimized sample input parameters of the classifier according to the amplitude-frequency response information; c, input the optimized sample input parameters into the corresponding classifier to perform fault classification and identification, and obtain final diagnosis. The beneficial effect of the present invention is that: the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics can automatically diagnose analog circuit faults, and significantly improve the accuracy and effect of analog circuit fault diagnosis.

Description

An Intelligent Fault Diagnosis Method for Analog Circuits Based on Amplitude-Frequency Characteristics

technical field

The invention belongs to the technical field of integrated circuit testing, and in particular relates to an intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics, which can improve the fault diagnosis rate and efficiency of analog circuits and can diagnose more fault types.

Background technique

With the development of microelectronics technology, large-scale digital-analog hybrid integrated circuits are widely used in all walks of life. Research on how to accurately and quickly diagnose faults in analog circuits has become an urgent problem in practical engineering; The characteristics of linearity, tolerance of components, and diversity of fault types lead to the slow development of analog circuit fault diagnosis technology. Traditional fault diagnosis theories and methods are difficult to achieve the expected results in actual engineering.

Analog circuits are electronic circuits that process analog signals, that is, continuous signals in both time and amplitude, which exhibit many complex characteristics in fault diagnosis. Although many experts and scholars at home and abroad have carried out a lot of research on analog circuit fault diagnosis, and many analog circuit fault diagnosis methods have emerged, it is precisely because of the characteristics of analog circuit that the existing analog circuit fault diagnosis methods are not perfect. Based on the analysis and summary of the existing domestic and foreign analog circuit fault diagnosis research, it is found that there are still some problems in the existing analog circuit fault diagnosis methods:

(1) The accuracy of analog circuit fault diagnosis needs to be further improved. Due to the presence of tolerances in analog circuit components, the characteristics of some failure modes are very similar, thereby increasing the ambiguity and uncertainty of failure classification. Existing methods usually combine these indistinguishable failure modes into one type of failure treatment or only diagnose the diagnosable components in the circuit through the testability analysis of the circuit. Therefore, these methods cannot fully realize the detection of all failure modes. correct diagnosis.

(2) The determination of the analog circuit fault model is unreasonable, the fault setting interval is too large, and there is no unified standard, resulting in imprecise diagnosis of analog circuit faults. According to the fault degree of the analog circuit, there are soft faults and hard faults. At present, for hard faults, the open circuit or short circuit of the circuit is usually simulated by connecting a resistor in series or in parallel. The branch simulates an open-circuit fault of the component. It is generally believed that the principle of soft faults caused by component parameter values deviating from its tolerance range is that the component parameters are greater than or less than 10 times their nominal values. The existing method has great randomness in setting the fault mode, and it can have a better diagnostic effect on soft faults that deviate greatly from the nominal value, but if the soft fault is set close to the nominal value, it will be greatly reduced. Accuracy of analog circuit fault diagnosis.

(3) Non-linear circuits widely exist in analog circuits, coupled with the tolerance of components and other reasons, the output response of analog circuits changes dynamically within a certain range, making some fault characteristics of the circuit show very similar characteristics, and only There are small differences locally, which are very difficult to distinguish, which makes fault diagnosis difficult. Therefore, the feature extraction of analog circuit faults is particularly important, and it is one of the keys to obtain good fault diagnosis results. Most of the current feature extraction methods are based on signal processing methods to extract features from the local signals of analog circuits, but the effect is not satisfactory, similar faults are still difficult to distinguish, and the misdiagnosis rate of fuzzy faults is still high.

(4) It is difficult to obtain analog circuit fault samples. Analog circuit fault diagnosis is a typical small-sample pattern recognition problem. The lack of fault samples has become one of the important reasons that seriously affect the development of analog circuit fault diagnosis technology. Existing fault classification methods do not work well for small sample problems.

Therefore, it is necessary to provide an intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics that can improve the fault diagnosis rate and efficiency of analog circuits and can diagnose more fault types.

Contents of the invention

The purpose of the present invention is to provide an intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics, which can improve the fault diagnosis rate and efficiency of analog circuits and can diagnose more fault types.

The technical scheme of the present invention is as follows: a method for intelligent fault diagnosis of analog circuits based on amplitude-frequency characteristics comprises the following steps:

a. Apply an excitation signal to the circuit to be diagnosed, and collect the amplitude-frequency response information output by the circuit to be diagnosed as fault characteristic information;

b. Analyzing and processing the amplitude-frequency response information, and obtaining optimized sample input parameters of the classifier according to the amplitude-frequency response information;

c. Inputting the optimized sample input parameters into the corresponding classifier to perform fault classification and identification, and obtain a final diagnosis result.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, in step a, the excitation signal is a sinusoidal voltage signal with an amplitude of 1V.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, the step b specifically includes the following steps:

b1. Using Principal Component Analysis (PCA) to perform feature extraction and dimensionality reduction processing on the amplitude-frequency response information;

b2. Perform normalization processing on the amplitude-frequency response information after feature extraction and dimensionality reduction processing;

b3. Optimizing the normalized amplitude-frequency response information by using particle swarm optimization (PSO), so as to obtain the optimized sample input parameters of the classifier.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, in step b2, a plurality of attributes are selected according to the contribution rate to form fault features, and the fault features are normalized.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, the normalization process is to convert all data into numbers between [0,1], and the normalization The data normalization function used in the processing is as follows:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}$

Among them, x _max is the maximum number in the data sequence; x _min is the minimum number in the sequence, and x _k and x′ _k are the values before and after normalization respectively.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, the step b3 specifically includes the following steps:

Determine the fitness function, and the accuracy rate under the cross-validation of the training samples is used as the fitness function value in the particle swarm optimization algorithm;

Particle swarm initialization, random generation of particle swarms of moderate size in the solution space, individual particle swarms represent the parameters of the classifier;

Calculate the particle fitness value, set the penalty parameter C of the classifier and the kernel function parameter σ, input the sample set into the classifier for training, obtain the recognition rate of the test sample, and calculate the particle fitness according to the evaluation function of the classification performance of the classifier;

Determine the individual extremum P _best and the global extremum g _best , compare the updated particle fitness value with the fitness value corresponding to the individual extremum P _best , if it is better, update the individual extremum P _best , otherwise keep the original value; compare the updated individual extremum P _best of each particle with the global extremum g _best , if superior, update the global extremum g _best , otherwise retain the original value;

Update the position and velocity information of the particles;

Judging whether the termination condition is met, that is, whether the maximum number of iterations or other termination conditions set are met, and if so, output the classifier parameters, which are the optimal parameters, and the algorithm ends; otherwise, return to the step of calculating the particle fitness value until the termination condition is met , output the optimal classifier parameter values.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, the classifier adopts a support vector machine (SVM).

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, the optimized sample input parameters of the classifier include penalty parameter C and kernel function parameter σ corresponding to the support vector machine.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, in step c, multiple classifiers are constructed according to a one-to-many combination method.

In the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics provided by the embodiment of the present invention, before step a, further includes step c: determining the fault mode of the circuit according to the circuit to be diagnosed.

The beneficial effect of the present invention is that: in the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics, the amplitude-frequency response of the analog circuit is collected as fault characteristics, and then principal component analysis is used for fault feature extraction and dimension reduction processing, thereby effectively The redundancy and interference components in the fault signature can be reduced as much as possible. Moreover, the fault classifier adopts a support vector machine, and uses a particle swarm optimization algorithm to obtain the kernel function parameters and penalty parameters of the support vector machine, thereby improving the accuracy of analog circuit fault diagnosis. Therefore, the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics can automatically diagnose analog circuit faults, and significantly improve the accuracy and effect of analog circuit fault diagnosis.

Description of drawings

Fig. 1 is a schematic flow chart of an analog circuit intelligent fault diagnosis method based on amplitude-frequency characteristics provided by an embodiment of the present invention;

Fig. 2 is a block flow diagram of an analog circuit intelligent fault diagnosis method based on amplitude-frequency characteristics provided by an embodiment of the present invention;

Fig. 3 is the schematic diagram of the circuit structure of the double quadratic high-pass filter of four operational amplifiers;

Fig. 4 is a flow chart of step S3 of the analog circuit intelligent fault diagnosis method based on amplitude-frequency characteristics shown in Fig. 2;

FIG. 5 is a schematic flow diagram of the particle swarm optimization support vector machine in step S3 shown in FIG. 4 .

detailed description

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Unless the context clearly states otherwise, the number of elements and components in the present invention can exist in a single form or in multiple forms, and the present invention is not limited thereto. Although the steps in the present invention are arranged with labels, they are not used to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a certain step requires other steps as a basis, the relative order of the steps can be adjusted. It can be understood that the term "and/or" used herein refers to and covers any and all possible combinations of one or more of the associated listed items.

Please refer to Fig. 1 and Fig. 2 at the same time. Fig. 1 is a schematic flowchart of an analog circuit intelligent fault diagnosis method based on amplitude-frequency characteristics provided by an embodiment of the present invention, and Fig. 2 is an analog circuit based on amplitude-frequency characteristics provided by an embodiment of the present invention Flow chart of intelligent fault diagnosis method. The intelligent fault diagnosis method 100 for analog circuits based on amplitude-frequency characteristics provided by the present invention includes the following steps:

S1. Determine the failure mode of the circuit according to the circuit to be diagnosed.

Specifically, according to the type and performance of the electronic components of the circuit to be diagnosed, the possible failure mode of the circuit to be diagnosed can be determined.

For example, taking the quadratic high-pass filter with four operational amplifiers as shown in Figure 3 as an example, the nominal values of the components in the circuit are shown in Figure 3, where the resistors and capacitors each have a tolerance of 5%. When the component deviates from its nominal value by ±50%, the circuit has a soft fault, that is, the value of the component when the soft fault occurs should be in the interval [50%X, 95%X)∪(105%X, 150%X] (X is the nominal value of the component). Table 1 shows the normal value, soft fault value and corresponding fault category of the quadratic high-pass filter of the four operational amplifiers. Among them, Indicates that the component has an oversized fault, Indicates that the component has a small fault, and there are 13 fault modes including the non-fault state.

Table 1 Failure mode setting table

S2. Apply an excitation signal to the circuit to be diagnosed, and collect amplitude-frequency response information output by the circuit to be diagnosed as fault characteristic information.

Specifically, a sinusoidal voltage signal with an amplitude of 1V is applied to the input terminal of the circuit to be diagnosed as an excitation signal, and a parameter analysis (AC Sweep) is performed, and amplitude-frequency response signals are respectively collected at the output terminal of the circuit to be diagnosed as a fault of the circuit feature. Moreover, set the start frequency to 1kHz, the stop frequency to 25kHz, and 60 sampling points, and conduct Monte Carlo analysis 50 times on all failure modes, and obtain 50 samples with 60 attributes for each failure mode.

S3. Analyze and process the amplitude-frequency response information, and obtain optimized sample input parameters of the classifier according to the amplitude-frequency response information.

Specifically, please refer to FIG. 4 , which is a flowchart of step S3 of the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics shown in FIG. 2 . Described step S3 comprises the following steps:

S31. Using a principal component analysis method to perform feature extraction and dimensionality reduction processing on the amplitude-frequency response information;

S32. Perform normalization processing on the amplitude-frequency response information after feature extraction and dimensionality reduction processing;

S33. Using the particle swarm optimization algorithm to optimize the normalized amplitude-frequency response information, so as to obtain the optimized sample input parameters of the classifier.

In step S31 and step S32, because the sample of the amplitude-frequency response information contains a large number of redundant components, which will affect the final diagnosis result and performance, it is necessary to use the principal component analysis method for feature extraction and reduction of the obtained samples. dimension processing. Preferably, in the process of using the principal component analysis method, a plurality of attributes are selected according to the contribution rate to form the fault feature, and normalization processing is performed on the fault feature. For example, the first seven attributes with the largest contribution rate are selected to form fault features, and the fault feature values are normalized.

Moreover, the normalization process is to convert all the data into numbers between [0,1], the purpose of which is to cancel the magnitude difference between the data of each dimension, and avoid the recognition effect being poor due to the large difference in the magnitude of the input and output data. good.

In this embodiment, the data normalization function used in the normalization processing is as follows:

${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}$

Among them, x _max is the maximum number in the data sequence; x _min is the minimum number in the sequence, and x _k and x′ _k are the values before and after normalization respectively.

For example, if the biquadratic high-pass filter with four operational amplifiers shown in FIG. 3 is taken as an example, 50 samples with 7 attributes can be obtained according to step S31.

In step S33, the fault features processed in step S32 are divided into training samples and test samples, and the training samples are input into the support vector machine for training, and the kernel function parameters of the support vector machine are optimized by the particle swarm optimization algorithm and penalty parameters.

For example, the 50 samples obtained in step S31 are divided into two parts: the first 30 samples are used as training samples, and the last 20 samples are used as testing samples. Since there are 13 failure modes, 390 training samples and 260 testing samples are finally obtained. The samples are input into the support vector machine for training, and the kernel function parameters and penalty parameters of the support vector machine are optimized by the particle swarm optimization algorithm. The initial population generated by the particle swarm optimization algorithm is used as the parameter of the support vector machine to input the support vector machine model for training and testing, and iterative optimization is performed to generate the next generation of population parameters by updating the particle position and velocity information until the termination condition set in the particle swarm optimization algorithm is met. , and finally get the optimal penalty parameter C and kernel function parameter σ.

Specifically, please refer to FIG. 5 , which is a schematic flow chart of the particle swarm optimization support vector machine in step S3 shown in FIG. 4 . The step S33 specifically includes the following steps:

Determine the fitness function, and the accuracy rate under the cross-validation of the training samples is used as the fitness function value in the particle swarm optimization algorithm;

Particle swarm initialization, random generation of particle swarms of moderate size in the solution space, individual particle swarms represent the parameters of the classifier;

Calculate the particle fitness value, set the penalty parameter C of the classifier and the kernel function parameter σ, input the sample set into the classifier for training, obtain the recognition rate of the test sample, and calculate the particle fitness according to the evaluation function of the classification performance of the classifier;

Determine the individual extremum P _best and the global extremum g _best , compare the updated particle fitness value with the fitness value corresponding to the individual extremum P _best , if it is better, update the individual extremum P _best , otherwise keep the original value; compare the updated individual extremum P _best of each particle with the global extremum g _best , if superior, update the global extremum g _best , otherwise retain the original value;

Update the position and velocity information of the particles;

Judging whether the termination condition is met, that is, whether the maximum number of iterations or other termination conditions set are met, and if so, output the classifier parameters, which are the optimal parameters, and the algorithm ends; otherwise, return to the step of calculating the particle fitness value until the termination condition is met , output the optimal classifier parameter values.

S4. Input the optimized sample input parameters into the corresponding classifier to perform fault classification and identification, and obtain a final diagnosis result.

Specifically, in step S4, the classifier adopts a support vector machine, so the optimized sample input parameters of the classifier include a penalty parameter C and a kernel function parameter σ corresponding to the support vector machine.

Moreover, since the parameters of the support vector machine have a great influence on the final classification result, only by finding the optimal parameters can the fault diagnosis accuracy be improved. Therefore, by using the particle swarm optimization algorithm to obtain the penalty parameter C and the kernel function parameter of the support vector machine, the accuracy of fault diagnosis of the analog circuit can be improved.

In this embodiment, fault diagnosis of analog circuits is a multi-classification problem, and multi-classifiers need to be constructed. The present invention adopts a one-to-many combination method to construct multiple classifiers. For example, for K classification, first construct classifiers, and then input the samples into the model to vote on the results, and the one with the most votes is the final classification result.

As another example, taking the quadratic high-pass filter with four op amps as shown in Figure 3 as an example, as shown in Table 2, only 4 samples of Type 2 faults were diagnosed incorrectly, and the fault diagnosis rate reached 98.5%.

Table 2 Fault diagnosis results

Compared with the prior art, in the intelligent fault diagnosis method 100 for analog circuits based on amplitude-frequency characteristics provided by the present invention, the amplitude-frequency response of analog circuits is collected as fault features, and then principal component analysis is used to extract fault features and reduce dimensionality. Thereby, redundancy and interference components in fault signatures can be effectively reduced. Moreover, the fault classifier adopts a support vector machine, and uses a particle swarm optimization algorithm to obtain the kernel function parameters and penalty parameters of the support vector machine, thereby improving the accuracy of analog circuit fault diagnosis. Therefore, the analog circuit intelligent fault diagnosis method 100 based on amplitude-frequency characteristics can automatically diagnose analog circuit faults, and significantly improve the accuracy and effect of analog circuit fault diagnosis.

It will be apparent to those skilled in the art that the invention is not limited to the details of the above-described exemplary embodiments, but that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Accordingly, the embodiments should be regarded in all points of view as exemplary and not restrictive, the scope of the invention being defined by the appended claims rather than the foregoing description, and it is therefore intended that the scope of the invention be defined by the appended claims rather than by the foregoing description. All changes within the meaning and range of equivalents of the elements are embraced in the present invention. Any reference sign in a claim should not be construed as limiting the claim concerned.

In addition, it should be understood that although this specification is described according to implementation modes, not each implementation mode only includes an independent technical solution, and this description in the specification is only for clarity, and those skilled in the art should take the specification as a whole , the technical solutions in the various embodiments can also be properly combined to form other implementations that can be understood by those skilled in the art.

Claims (10)

1. an analog circuit intelligent fault diagnosis method based on amplitude-frequency characteristics, is characterized in that: comprise the steps: a. Apply an excitation signal to the circuit to be diagnosed, and collect the amplitude-frequency response information output by the circuit to be diagnosed as fault characteristic information; b. Analyzing and processing the amplitude-frequency response information, and obtaining optimized sample input parameters of the classifier according to the amplitude-frequency response information; c. Inputting the optimized sample input parameters into the corresponding classifier to perform fault classification and identification, and obtain a final diagnosis result. 2. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 1, characterized in that: in step a, the excitation signal is a sinusoidal voltage signal with an amplitude of 1V. 3. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 1, characterized in that: said step b specifically comprises the following steps: b1, performing feature extraction and dimension reduction processing on the amplitude-frequency response information by using a principal component analysis method; b2. Perform normalization processing on the amplitude-frequency response information after feature extraction and dimensionality reduction processing; b3. Using the particle swarm optimization algorithm to optimize the normalized amplitude-frequency response information, so as to obtain the optimized sample input parameters of the classifier. 4. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 3, characterized in that: in step b2, a plurality of attributes are selected according to the contribution rate to form fault characteristics, and the fault characteristics are classified One treatment. 5. the intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 4, characterized in that: said normalization process is to convert all data into numbers between [0,1], and , the data normalization function used in the normalization process is as follows: ${\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}{{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; a</span>}\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; min</span>}}}$ Among them, x _max is the maximum number in the data sequence; x _min is the minimum number in the sequence, and x _k and x′ _k are the values before and after normalization respectively. 6. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 3, characterized in that: said step b3 specifically comprises the following steps: Determine the fitness function, and the accuracy rate under the cross-validation of the training samples is used as the fitness function value in the particle swarm optimization algorithm; Particle swarm initialization, random generation of particle swarms of moderate size in the solution space, individual particle swarms represent the parameters of the classifier; Calculate the particle fitness value, set the penalty parameter C of the classifier and the kernel function parameter σ, input the sample set into the classifier for training, obtain the recognition rate of the test sample, and calculate the particle fitness according to the evaluation function of the classification performance of the classifier; Determine the individual extremum P _best and the global extremum g _best , compare the updated particle fitness value with the fitness value corresponding to the individual extremum P _best , if it is better, update the individual extremum P _best , otherwise keep the original value; compare the updated individual extremum P _best of each particle with the global extremum g _best , if superior, update the global extremum g _best , otherwise retain the original value; Update the position and velocity information of the particles; Judging whether the termination condition is met, that is, whether the maximum number of iterations or other termination conditions set are met, and if so, output the classifier parameters, which are the optimal parameters, and the algorithm ends; otherwise, return to the step of calculating the particle fitness value until the termination condition is met , output the optimal classifier parameter values. 7. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 1, characterized in that: the classifier adopts a support vector machine. 8. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency features according to claim 7, wherein: the optimized sample input parameters of the classifier include corresponding penalty parameters C and kernel Function parameter σ. 9. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 7, characterized in that: in step c, multiple classifiers are constructed according to one-to-many combination. 10. The intelligent fault diagnosis method for analog circuits based on amplitude-frequency characteristics according to claim 1, characterized in that: before step a, further comprising step c: determining the fault mode of the circuit according to the circuit to be diagnosed.