Feature selection method and system for circuit intelligent test
Through the time-frequency domain expansion and combination increment method, the problem of difficult to take into account the effect of feature selection and calculation speed in circuit intelligent testing is solved, and the number of effective features in the feature set after dimensionality reduction is improved, achieving more efficient feature selection.
Patent Information
- Application Number
- CN202410919642.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-10
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-07-10
AI Technical Summary
The time domain signal feature selection method in the existing circuit intelligent test is difficult to ensure the number of effective features in the feature set after dimensionality reduction, and it is difficult to take into account the feature selection effect and calculation speed. The packaging method takes a long time and the filter method is difficult to reliably select the optimal feature combination. .
The dimensionality reduction method based on time-frequency domain expansion is adopted, and the dimensionality reduction is reduced by discrete wavelet analysis and principal component analysis, combined with the combined increment method, to ensure the number of effective features in the feature set after dimensionality reduction, and to select in a short time Output the optimal feature combination.
The number of effective features in the feature set after dimensionality reduction is increased, taking into account the feature selection effect and calculation speed, and more efficient circuit intelligent testing is achieved.
Smart Images

Figure CN118897975A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of circuit testing, and more specifically, relates to a feature selection method and system for circuit intelligent testing. Background Art
[0002] Circuit testing technology is a key link in ensuring the quality of electronic products and is widely used in all aspects of electronic system / integrated circuit production. By analyzing the electrical signals of each output port of the circuit, various circuit performance indicators are obtained. Only products with all indicators within the qualified range will be delivered to customers.
[0003] Traditional circuit testing requires expensive testing equipment, complex testing processes, and lengthy testing time. With the development of machine learning technology, machine learning-based circuit intelligent testing technology has attracted people's attention. Machine learning-based circuit intelligent testing is used to replace traditional circuit testing methods to achieve the purpose of reducing circuit testing costs and reducing circuit testing time.
[0004] Selecting features from the circuit time domain response signal is a key step in circuit intelligent testing. Existing time domain signal feature selection methods mainly include two steps: dimensionality reduction and selection. First, the original signal is reduced in dimension to reduce the feature dimension to avoid problems such as data sparsity or model overfitting when training the machine learning model. Then, the optimal feature combination is selected from the original feature set after dimensionality reduction for indicator prediction. There are two main selection methods: packaging and filtering. The packaging method selects the optimal feature combination by testing the effect of all possible combinations of the original feature set for machine learning model training; the filtering method does not involve model training, and selects the feature combination based on the size of the correlation by calculating the correlation between each feature in the original feature set and the indicator to be tested.
[0005] At present, there are the following shortcomings in the feature selection of time domain signals. On the one hand, in the dimension reduction process, it is difficult to ensure the number of effective features in the feature set after dimension reduction (effective features refer to features that are highly correlated with the indicators to be predicted). On the other hand, it is difficult to balance the selection effect and calculation speed in the selection process. The packaging method can reliably select the optimal feature combination, but it is necessary to train the machine learning model for all possible combinations separately, which will consume a lot of time; the filtering method selects features based on correlation. Although it is fast, it is difficult to reliably select the optimal feature combination (two low-correlation features may have high correlation after combination). Therefore, it is urgent to propose a feature selection method that can increase the number of effective features in the feature set after dimension reduction and balance the feature selection effect and calculation speed. Summary of the invention
[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a feature selection method and system for intelligent circuit testing, so as to increase the number of effective features in the feature set after dimensionality reduction and take into account both the feature selection effect and the calculation speed.
[0007] To achieve the above-mentioned invention purpose, the feature selection method for intelligent circuit testing of the present invention is characterized by including the following steps:
[0008] (1) Perform dimensionality reduction on the circuit time-domain response signal based on time-frequency domain expansion
[0009] First, perform discrete wavelet analysis on the circuit time-domain response signal to obtain wavelet approximation coefficients and the first n levels of wavelet detail coefficients, and then use principal component analysis to process them respectively. Retain the first m dimensions of the wavelet approximation coefficients and the first n levels of wavelet detail coefficients after principal component analysis to obtain m×(n + 1) features after dimensionality reduction;
[0010] (2) Sort the features after dimensionality reduction based on correlation to obtain the original feature set;
[0011] For the m×(n + 1) features after dimensionality reduction, calculate the correlation coefficient between each feature and the index to be measured respectively, and sort the m×(n + 1) features from high to low based on the correlation coefficient to obtain the original feature set S0: S0 = {f1, f2, …, f m×(n+1)}, where f i , i = 1, 2, …, m×(n + 1) represents the i-th feature sorted from high to low based on the correlation coefficient;
[0012] (3) Use the combined incremental method to select the optimal feature set
[0013] For each feature f i in the original feature set S0, perform feature selection. Add features to the feature set S1 one by one from high to low according to the correlation coefficient, that is, select feature f1 to train and test the machine learning model, and obtain the test mean square error, that is, the loss is loss1. Select features f1 and f2 to train and test the machine learning model, and obtain the mean square error as loss2. If the mean square error loss2 < loss1, then continue to add features. Use features f1, f2, and f3 to train and test the machine learning model, and obtain the mean square error as loss3. If the mean square error loss3 < loss2, then continue to add features, and so on, gradually expanding the feature set S1 until loss k+1 ≮loss k , at this time the feature set is S1: S1 = {f1, f2, …, f k};
[0014] The remaining features f k+2 , f k+3 , …, fm×(n+1) Combine them with feature set S1 one by one, train and test the machine learning model, and select the combination with the smallest mean square error as feature set S2. Combine the features in the original feature set S0 except the features in feature set S2 in pairs, and combine them with feature set S2 one by one to train and test the machine learning model. Select the combination with the smallest mean square error as the optimal feature set S2. optimal , complete the feature selection of circuit intelligent testing;
[0015] Among them, the feature label value of the training and testing machine learning model is the correlation coefficient between the feature and the indicator to be tested.
[0016] The feature selection system for circuit intelligent testing of the present invention is characterized by comprising: a dimensionality reduction sorting module and a combination selection module, wherein the circuit time domain response signal first passes through the dimensionality reduction sorting module to obtain an original feature set S0, and the original feature set S0 is input into the combination selection module to obtain an optimal feature set S optimal ;
[0017] The dimension reduction and sorting module comprises a discrete wavelet analysis unit, a feature dimension reduction unit, a correlation calculation unit, and a feature sorting unit. The discrete wavelet analysis unit performs discrete wavelet analysis on the input circuit time domain response signal, and outputs the wavelet approximation coefficients and the first n-level wavelet detail coefficients obtained after the analysis. The feature dimension reduction unit performs principal component analysis on the wavelet approximation coefficients and the first n-level wavelet detail coefficients respectively, and processes them respectively using principal component analysis, retaining the first m dimensions of the wavelet approximation coefficients and the first n-level wavelet detail coefficients after the principal component analysis, and obtaining m×(n+1) features. The correlation calculation unit calculates the correlation coefficient of each feature and the index to be measured for the m×(n+1) features after dimension reduction, and the feature sorting unit sorts the m×(n+1) features from high to low in terms of correlation coefficients to obtain an original feature set S0.
[0018] The combination selection module includes a selection unit, an expansion unit, a training unit, a comparison unit, and a difference unit. The selection unit selects features from the original feature set S0 according to the correlation coefficient from high to low. The expansion unit combines the selected features with the feature set stored in the previous cycle and outputs the combined feature set S1. The training unit sends the features in the feature set S1 to the machine learning model for training and testing, and calculates the mean square error. The comparison unit compares the current mean square error with the mean square error of the previous cycle. If it is less than, the next cycle is carried out. The selection unit selects features from the original feature set S0 according to the correlation coefficient from high to low, combines, trains and tests, and compares and judges. In this way, features are continuously added, and by analogy, the feature set S1 is gradually expanded until the loss is reached. k+1 ≮loss k , at this time the feature set is S1: S1 = {f1, f2, ..., f k};
[0019] The difference unit calculates the difference between the original feature set S0 and the feature set S1 and outputs it. Then the combination unit combines the features in the difference between the original feature set S0 and the feature set S1, that is, the remaining features f k+2 ,f k+3 ,…,f m×(n+1) The features in the original feature set S0 and the feature set S2 are combined one by one with the feature set S1, and then the training unit uses the features in the combination to train and test the machine learning model. The comparison unit selects the combination with the smallest mean square error as the feature set S2 according to the obtained mean square error. The difference unit calculates the difference between the original feature set S0 and the feature set S2 and outputs it. Then the combination unit combines the features in the difference between the original feature set S0 and the feature set S2, that is, the residual features, in pairs, and combines them with the feature set S2 one by one. Then the training unit uses the features in the combination to train and test the machine learning model. The comparison unit selects the combination with the smallest mean square error as the optimal feature set S according to the obtained mean square error. optimal , complete the feature selection for circuit intelligent testing.
[0020] The object of the invention of the present invention is achieved in this way:
[0021] The feature selection method and system for circuit intelligent testing of the present invention first transforms the circuit time domain response signal using discrete wavelet analysis to expand the time-frequency domain information of the circuit time domain response signal, and then reduces the dimension of the wavelet approximation coefficients and the first n-level wavelet detail coefficients after discrete wavelet analysis to ensure that the feature set after dimension reduction has a sufficient number of effective features. The present invention uses a combined incremental method to select the optimal feature set, combines the advantages of the filtering method and the packaging method, and takes into account the effect of feature selection and the calculation speed.
[0022] Compared with the prior art, the present invention has the following advantages:
[0023] Compared with the existing dimensionality reduction methods, the dimensionality reduction method based on time-frequency domain expansion designed by the present invention has more effective feature numbers in the feature set after dimensionality reduction.
[0024] Compared with the packaging method, the combined incremental method designed by the present invention is more efficient and can select the optimal feature set in a shorter computing time. Compared with the filtering method, the combined incremental method designed by the present invention can select the optimal feature set more reliably.
[0025] Compared with the existing feature selection system, the feature selection system designed by the present invention can improve the effective combination while taking into account the effect and calculation speed of feature selection. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of a specific implementation method of the feature selection method for circuit intelligent testing of the present invention;
[0027] Figure 2 yes Figure 1 The schematic diagram of the specific processing process of steps S1 and S2 is shown;
[0028] Figure 3 yes Figure 1 The schematic diagram of the specific processing process of step S3 is shown;
[0029] Figure 4 It is a schematic diagram of the principle of a specific implementation of the feature selection system for circuit intelligent testing of the present invention. DETAILED DESCRIPTION
[0030] The specific implementation of the present invention is described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.
[0031] Figure 1 It is a flow chart of a specific implementation of the feature selection method for circuit intelligent testing of the present invention.
[0032] In this embodiment, if Figure 1 As shown, the feature selection method for circuit intelligent testing of the present invention comprises the following steps:
[0033] Step S1: Perform dimensionality reduction on the circuit time domain response signal based on time-frequency domain expansion
[0034] like Figure 2 As shown, first, discrete wavelet analysis is performed on the circuit time domain response signal to obtain the wavelet approximation coefficients and the first n-level wavelet detail coefficients, and then principal component analysis is used to process them separately, retaining the first m dimensions of the wavelet approximation coefficients and the first n-level wavelet detail coefficients after principal component analysis, and obtaining m×(n+1) features after dimensionality reduction.
[0035] In this embodiment, Haar wavelet in the Daubechies wavelet family is used as a wavelet basis function to perform discrete wavelet analysis, and the decomposition level of the wavelet detail coefficient is set to 3 levels, that is, n=3.
[0036] In this embodiment, principal component analysis is used to process the wavelet approximation coefficients and the first n-level wavelet detail coefficients respectively, and the first five dimensions after principal component analysis are retained, that is, m=5.
[0037] Step S2: Sort the reduced features based on relevance to obtain the original feature set
[0038] like Figure 2As shown in the figure, for the m×(n + 1) features after dimensionality reduction, the correlation coefficient between each feature and the index to be measured is calculated respectively. The correlation coefficients of all features form a correlation coefficient matrix. The m×(n + 1) features are sorted based on the correlation coefficient from high to low, and the original feature set S0 is obtained: S0 = {f1, f2, …, f m×(n+1)}, where f i , i = 1, 2, …, m×(n + 1) represents the i-th feature sorted from high to low based on the correlation coefficient.
[0039] In this embodiment, the correlation coefficient is the Pearson correlation coefficient, which is used as a correlation measurement index.
[0040] Step S3: Select the optimal feature set using the combined incremental method
[0041] The present invention uses the K-fold method as the machine learning model training / testing method, and uses the combined incremental method to select based on the average value of the mean square loss of each fold test as the test mean square error, and selects the optimal combination, that is, the optimal feature set:
[0042] As Figure 3 shown, for each feature f i in the original feature set S0, feature selection is performed. Features are added one by one to the feature set S1 in descending order of the correlation coefficient, that is, select feature f1 to train and test the machine learning model, and the test mean square error is loss1. Select features f1 and f2 to train and test the machine learning model, and the mean square error is loss2. If the mean square error loss2 < loss1, then continue to add features. Use features f1, f2, and f3 to train and test the machine learning model, and the mean square error is loss3. If the mean square error loss3 < loss2, then continue to add features, and so on, gradually expanding the feature set S1 until loss k+1 ≮loss k , at this time the feature set is S1: S1 = {f1, f2, …, f k};
[0043] Combine the remaining features f k+2 , f k+3 , …, f m×(n+1) with the feature set S1 one by one, train and test the machine learning model, and select the combination with the smallest mean square error among the obtained mean square errors as the feature set S2. Combine the features other than the features in the feature set S2 in the original feature set S0 in pairs, and combine and train and test the machine learning model with the feature set S2 pair by pair, and select the combination with the smallest mean square error among the obtained mean square errors as the optimal feature set S optim , completing the feature selection for circuit intelligent testing;
[0044] Among them, the feature label value of the training and testing machine learning model is the correlation coefficient between the feature and the indicator to be tested.
[0045] In this embodiment, a neural network is used as a machine learning model, which includes a fully connected layer as an input layer, 5 fully connected layers as hidden layers, and 1 fully connected layer as an output layer.
[0046] Figure 4 It is a schematic diagram of the principle of a specific implementation of the feature selection system for circuit intelligent testing of the present invention.
[0047] In this embodiment, if Figure 4 As shown, the feature selection system for circuit intelligent testing of the present invention includes two parts: a dimensionality reduction sorting module 1 and a combination selection module 2. The circuit time domain response signal first passes through the dimensionality reduction sorting module 1 to obtain an original feature set S0, and the original feature set S0 is input into the combination selection module 2 to obtain the optimal feature set S optimal .
[0048] like Figure 4 As shown, the dimension reduction and sorting module includes a discrete wavelet analysis unit 101, a feature dimension reduction unit 102, a correlation calculation unit 103, and a feature sorting unit 104. The discrete wavelet analysis unit 101 performs discrete wavelet analysis on the input circuit time domain response signal, and outputs the wavelet approximation coefficients and the first n wavelet detail coefficients obtained after the analysis. The feature dimension reduction unit 102 performs principal component analysis on the wavelet approximation coefficients and the first n wavelet detail coefficients simultaneously, and uses principal component analysis to process them respectively, retaining the first m dimensions of the wavelet approximation coefficients and the first n wavelet detail coefficients after principal component analysis, and obtaining m×(n+1) features. The correlation calculation unit 103 calculates the correlation coefficient of each feature and the index to be measured for the m×(n+1) features after dimension reduction, and the feature sorting unit sorts the m×(n+1) features from high to low in terms of correlation coefficients to obtain the original feature set S0.
[0049] The combination selection module 2 includes a selection unit 201, an expansion unit 202, a training unit 203, a comparison unit 204, a difference unit 205 and a combination unit 206. The selection unit 201 selects features from the original feature set S0 in order from high to low according to the correlation coefficient. The expansion unit 202 combines the selected features with the feature set stored in the previous cycle, and outputs the combined feature set S1. The training unit 203 sends the features in the feature set S1 to the machine learning model for training and testing, and calculates the mean square error. The comparison unit 204 compares the current mean square error with the mean square error of the previous cycle. If it is less than, the next cycle is carried out. The selection unit 201 selects features from the original feature set S0 in order from high to low according to the correlation coefficient, combines, trains and tests, and compares and judges. In this way, features continue to be added, and so on, and the feature set S1 is gradually expanded until the loss is reached.k+1 ≮loss k , at this time the feature set is S1: S1 = {f1, f2, ..., f k}.
[0050] The difference unit 205 obtains the difference between the original feature set S0 and the feature set S1 and outputs it. Then, the combination unit 206 combines the features in the difference between the original feature set S0 and the feature set S1, i.e., the remaining features f k+2 ,f k+3 ,…,f m×(n+1) The features in the original feature set S0 and the feature set S2 are combined one by one with the feature set S1, that is, all the combinations are traversed, and then the training unit 203 uses the features in the combination to train and test the machine learning model. The comparison unit 204 selects the combination with the smallest mean square error as the feature set S2 according to the obtained mean square error. The difference unit 205 calculates the difference between the original feature set S0 and the feature set S2 and outputs it. Then the combination unit 206 combines the features in the difference between the original feature set S0 and the feature set S2, that is, the residual features, in pairs, and combines them with the feature set S2 one by one, that is, all the combinations are traversed, and then the training unit 203 uses the features in the combination to train and test the machine learning model. The comparison unit 204 selects the combination with the smallest mean square error as the optimal feature set S according to the obtained mean square error. optimal , complete the feature selection for circuit intelligent testing.
[0051] Although the above describes the illustrative specific embodiments of the present invention to facilitate those skilled in the art to understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations using the concept of the present invention are protected.
Claims
1. A feature selection method for circuit intelligent testing, characterized in that: The following steps are involved: (1) Perform dimensionality reduction on the circuit time domain response signal based on time-frequency domain expansion Firstly, discrete wavelet analysis is performed on the circuit time domain response signal to obtain the wavelet approximation coefficient and the first n-level wavelet detail coefficients, and then principal component analysis is used to process them respectively, retaining the first m dimensions of the wavelet approximation coefficient and the first n-level wavelet detail coefficient after principal component analysis, and obtaining m×(n+1) features after dimensionality reduction; (2) Sort the reduced-dimensional features based on relevance to obtain the original feature set; For the m×(n+1) features after dimensionality reduction, the correlation coefficient between each feature and the index to be measured is calculated respectively, and the m×(n+1) features are sorted from high to low based on the correlation coefficient to obtain the original feature set S0: S0 = {f1, f2, ..., f m×(n+1) }, where f i , i=1,2,...,m×(n+1) represents the i-th feature sorted from high to low based on the correlation coefficient; (3) Use the combination incremental method to select the optimal feature set For each feature f in the original feature set S0 i Feature selection is performed. The features are added one by one to the feature set S1 in descending order of the correlation coefficient. That is, the feature f1 is selected to train and test the machine learning model, and the test mean squared error, i.e., the loss, is obtained as loss1. The features f1 and f2 are selected to train and test the machine learning model, and the mean squared error is obtained as loss2. If the mean squared error loss2 < loss1, then more features are added. The features f1, f2, and f3 are used to train and test the machine learning model, and the mean squared error is obtained as loss3. If the mean squared error loss3 < loss2, then more features are added, and so on, gradually expanding the feature set S1 until At this time, the feature set is S1: S1 = {f1, f2,..., f k}; The remaining features f k+2 , f k+3 , ..., f m×(n+1) Combine them with feature set S1 one by one, train and test the machine learning model, and select the combination with the smallest mean square error as feature set S2. Combine the features in the original feature set S0 except the features in feature set S2 in pairs, and combine them with feature set S2 one by one to train and test the machine learning model. Select the combination with the smallest mean square error as the optimal feature set S2. optim , complete the feature selection of circuit intelligent testing; Among them, the feature label value of the training and testing machine learning model is the correlation coefficient between the feature and the indicator to be tested.
2. The feature selection method for circuit intelligent testing according to claim 1, characterized in that: The correlation coefficient is the Pearson correlation coefficient.
3. The feature selection method for circuit intelligent testing according to claim 1, characterized in that: A neural network is used as the machine learning model, which includes a fully connected layer as the input layer, 5 fully connected layers as hidden layers, and 1 fully connected layer as the output layer.
4. A feature selection system for circuit intelligent testing, characterized in that: include: Dimensionality reduction sorting module and combination selection module. The circuit time domain response signal first passes through the dimension reduction sorting module to obtain the original feature set S0. The original feature set S0 is input into the combination selection module to obtain the optimal feature set S optimal ; The dimension reduction and sorting module comprises a discrete wavelet analysis unit, a feature dimension reduction unit, a correlation calculation unit, and a feature sorting unit. The discrete wavelet analysis unit performs discrete wavelet analysis on the input circuit time domain response signal, and outputs the wavelet approximation coefficients and the first n-level wavelet detail coefficients obtained after the analysis. The feature dimension reduction unit performs principal component analysis on the wavelet approximation coefficients and the first n-level wavelet detail coefficients respectively, and processes them respectively using principal component analysis, retaining the first m dimensions of the wavelet approximation coefficients and the first n-level wavelet detail coefficients after the principal component analysis, and obtaining m×(n+1) features. The correlation calculation unit calculates the correlation coefficient of each feature and the index to be measured for the m×(n+1) features after dimension reduction, and the feature sorting unit sorts the m×(n+1) features from high to low in terms of correlation coefficients to obtain an original feature set S0. The combination selection module includes a selection unit, an expansion unit, a training unit, a comparison unit, and a difference unit. The selection unit selects features from the original feature set S0 according to the correlation coefficient from high to low. The expansion unit combines the selected features with the feature set stored in the previous cycle, and outputs the combined feature set S1. The training unit sends the features in the feature set S1 to the machine learning model for training and testing, and calculates the mean square error. The comparison unit compares the current mean square error with the mean square error of the previous cycle. If it is less than, the next cycle is carried out. The selection unit selects features from the original feature set S0 according to the correlation coefficient from high to low, combines, trains and tests, and compares and judges. In this way, features are continuously added, and by analogy, the feature set S1 is gradually expanded until At this time, the feature set is S1: S1 = {f1, f2, ..., f k }; The difference unit calculates the difference between the original feature set S0 and the feature set S1 and outputs it. Then the combination unit combines the features in the difference between the original feature set S0 and the feature set S1, that is, the remaining features f k+2 , f k+3 , ..., f m×(n+1) The features in the original feature set S0 and the feature set S2 are combined one by one with the feature set S1, and then the training unit uses the features in the combination to train and test the machine learning model. The comparison unit selects the combination with the smallest mean square error as the feature set S2 according to the obtained mean square error. The difference unit calculates the difference between the original feature set S0 and the feature set S2 and outputs it. Then the combination unit combines the features in the difference between the original feature set S0 and the feature set S2, that is, the residual features, in pairs, and combines them with the feature set S2 one by one. Then the training unit uses the features in the combination to train and test the machine learning model. The comparison unit selects the combination with the smallest mean square error as the optimal feature set S according to the obtained mean square error. optim , complete the feature selection for circuit intelligent testing.
Citation Information
Patent Citations
Escalator synchronization rate intelligence test instrument and test method
CN101187601A
Online intelligent fault diagnosis method of power electronic circuit based on FRFT and IFSVC
CN101614786A
Automatic driving test case generation method and device, electronic equipment and storage medium
CN114815605A
Image feature dimension reduction selection method based on feature inter-class distance and machine learning
CN115359283A
Multi-dimensional layered hardware Trojan horse evaluation method for IP soft core
CN116578978A
