Bituminous mixture microstructure parameter visual screening method and system based on collinearity quantification technology

Abstract

The invention discloses an asphalt mixture microstructure parameter visual screening method and system based on a collinearity quantification technology, and relates to the field of road engineering material structure performance evaluation. The method aims at solving the problems that according to an existing asphalt mixture structure parameter screening method, key mechanism information is lost in a screening result, and consequently the follow-up engineering model is poor in universality and low in calculation efficiency. The method comprises the following steps: obtaining a correlation coefficient between mesostructure parameters of the asphalt mixture; obtaining structure parameter correlation frequencies, and iteratively eliminating the microstructure parameters of the to-be-screened asphalt mixture corresponding to the highest value of the structure parameter correlation frequencies until all the structure parameter correlation frequencies are smaller than a preset frequency threshold; and obtaining a variance expansion factor of the residual to-be-screened asphalt mixture microstructure parameters, and obtaining an asphalt mixture microstructure parameter screening result by using the variance expansion factor of the residual to-be-screened asphalt mixture microstructure parameters.

Description

Technical Field

[0001] This invention relates to the field of structural performance evaluation of road engineering materials, and in particular to a method and system for visually screening the microstructural parameters of asphalt mixtures based on collinearity quantification technology. Background Technology

[0002] As a typical heterogeneous multiphase composite material, the performance of asphalt mixtures is jointly determined by microstructural factors such as the morphological characteristics of aggregate particles, their spatial structure combination, and their bonding state with asphalt binder. These structural factors exhibit significant nonlinear correlations and high coupling characteristics at different scales, making the mapping relationship between macroscopic performance and microstructure extremely complex. Traditional structural parameter selection often relies on empirical judgment or univariate statistical analysis, which fails to fully reveal the redundancy and potential collinearity between parameters, resulting in large model fitting errors, poor stability, and weak engineering applicability. Against this backdrop, there is an urgent need for a scientific, systematic, and visualized parameter selection method that can accurately extract the structural features that play a dominant role in macroscopic performance from multidimensional microstructural indicators, providing a solid foundation for structure-performance relationship modeling and mechanism explanation.

[0003] In the field of road engineering and pavement performance research, establishing predictive models between structural parameters and performance indicators has become a crucial step. However, the practicality of these models depends not only on algorithm complexity but also, and more importantly, on the representativeness and independence of the input parameters. Faced with high-dimensional structural parameter systems, efficiently identifying key indicators with significant influencing mechanisms and low risk of collinearity becomes the core issue for improving the accuracy and stability of prediction equations.

[0004] Currently, the following methods are mainly used for screening structural parameters: (1) Selecting parameters based on experience or subjective judgment. Although this method is simple, it is highly dependent on expert knowledge, lacks objective standards, and is difficult to replicate. (2) Using dimensionality reduction methods such as principal component analysis (PCA) and minimum redundancy maximum correlation (mRMR) to reconstruct or select features. Compared with subjective selection of parameters based on experience, this method does not require expert knowledge, the evaluation standard is more objective, and it helps to compress feature dimensions. However, blind dimensionality reduction will cause the screening results to lose key mechanism information, resulting in poor generality of subsequent engineering models and significant deviations between calculation results and engineering reality. Summary of the Invention

[0005] The purpose of this invention is to address the problem that existing methods for screening structural parameters of asphalt mixtures lose key mechanistic information in the screening results, leading to poor universality of subsequent engineering models and significant deviations between calculation results and actual engineering conditions. Therefore, this invention proposes a visualization screening method and system for microstructural parameters of asphalt mixtures based on collinearity quantification technology.

[0006] A method for visually screening microstructural parameters of asphalt mixtures based on collinearity quantification technology, specifically as follows:

[0007] Step 1: Obtain the microstructure parameters of the asphalt mixture and the correlation coefficients between the microstructure parameters of the asphalt mixture;

[0008] Step 2: Obtain the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture. Compare the correlation frequency of structural parameters with the preset frequency threshold. If the correlation frequency of all structural parameters is less than the preset frequency threshold, proceed to Step 5; otherwise, obtain the highest correlation frequency of structural parameters. If there is only one highest correlation frequency, proceed to Step 4; if there are multiple highest correlation frequencies, proceed to Step 3.

[0009] Step 3: Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to Step 2;

[0010] Step 4: Remove the asphalt mixture microstructure parameters corresponding to the highest correlation frequency of structural parameters, and return to Step 2;

[0011] Step 5: Obtain the variance expansion factor of the remaining asphalt mixture microstructure parameters, and use the variance expansion factor of the remaining asphalt mixture microstructure parameters to obtain the screening results of the asphalt mixture microstructure parameters.

[0012] Furthermore, the acquisition of the microstructural parameters of the asphalt mixture in step one, and the obtaining of the correlation coefficients between the microstructural parameters of the asphalt mixture, specifically involves:

[0013] Step 1: Obtain the microstructure parameters of the asphalt mixture and the Pearson correlation coefficients between the microstructure parameters of the asphalt mixture, and construct the correlation coefficient matrix.

[0014] The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar.

[0015] The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure.

[0016] The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0017] The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure;

[0018] The elements in the correlation coefficient matrix are the Pearson correlation coefficients between two microstructure parameters;

[0019] Steps 1 and 2: Visualize the correlation coefficient matrix according to the absolute value of the correlation coefficient.

[0020] Furthermore, the step two, which involves obtaining the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture, specifically involves:

[0021] First, set the threshold for the Pearson correlation coefficient;

[0022] Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. .

[0023] Furthermore, the variance expansion factor for obtaining the microstructural parameters of the remaining asphalt mixture in step five is specifically as follows:

[0024]

[0025] in, It is the first The variance inflation factor of each residual microstructure parameter The coefficient of determination is obtained by regression analysis with the i-th residual microstructure parameter as the dependent variable and the remaining residual microstructure parameters as independent variables.

[0026] The regression analysis is a multiple linear regression analysis.

[0027] Furthermore, the step five, which involves obtaining the screening results of the microstructure parameters of the asphalt mixture using the variance expansion factor of the remaining asphalt mixture microstructure parameters, specifically involves:

[0028] By comparing the variance expansion factor of each remaining asphalt mixture microstructure parameter with the VIF threshold, the remaining asphalt mixture microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of asphalt mixture microstructure parameters.

[0029] A visualization screening system for microstructure parameters of asphalt mixtures based on collinearity quantification technology includes: a correlation coefficient acquisition module, a microstructure parameter pre-screening module, and a microstructure parameter final screening module.

[0030] The correlation coefficient acquisition module is used to acquire the correlation coefficients between the microstructure parameters of asphalt mixtures.

[0031] The microstructure parameter pre-screening module is used to obtain preliminary screening results of asphalt mixture microstructure parameters based on the correlation coefficients between the microstructure parameters of asphalt mixtures.

[0032] The final screening module for microstructure parameters uses the preliminary screening results of the microstructure parameters of asphalt mixtures to obtain the final screening results of the microstructure parameters of asphalt mixtures.

[0033] Furthermore, the correlation coefficient acquisition module specifically comprises:

[0034] A1. Obtain the microstructure parameters of asphalt mixtures and the Pearson correlation coefficients between the microstructure parameters of asphalt mixtures, and construct the correlation coefficient matrix.

[0035] The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar.

[0036] The skeleton structure parameters include: the fractal dimension of the aggregate skeleton structure, the width of the multifractal spectrum of the aggregate skeleton structure, the height difference of the multifractal spectrum of the aggregate skeleton structure, and the peak height of the aggregate skeleton structure.

[0037] The void structure parameters include: void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0038] The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure.

[0039] The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0040] The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure;

[0041] The elements in the correlation coefficient matrix are the Pearson correlation coefficients between two microstructure parameters;

[0042] A2. Visualize the correlation coefficient matrix using a heatmap or color-coded matrix format according to the absolute value of the correlation coefficient.

[0043] Furthermore, the microstructural parameter pre-screening module specifically comprises:

[0044] B1. Obtain the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture. Compare the correlation frequency of structural parameters with a preset frequency threshold. If the correlation frequency of all structural parameters is less than the preset frequency threshold, then the current microstructural parameter is used as the preliminary screening result of the microstructural parameters of the asphalt mixture, and the final screening module of the microstructural parameters is executed; otherwise, obtain the highest value of the correlation frequency of the structural parameters. If there is only one highest value of the correlation frequency of the structural parameters, then execute B3; if there are multiple highest values ​​of the correlation frequency of the structural parameters, then execute B2.

[0045] B2. Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to B1;

[0046] B3. Remove the asphalt mixture microstructure parameters corresponding to the highest frequency of correlation of structural parameters, and return to B1.

[0047] Furthermore, the correlation frequency of structural parameters obtained from the correlation coefficients between the microstructural parameters of the asphalt mixture in B1 is specifically as follows:

[0048] First, set the threshold for the Pearson correlation coefficient;

[0049] Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. .

[0050] Furthermore, the final screening module for the microstructure parameters specifically comprises:

[0051] C1. Obtain the variance inflation factor for each microstructure parameter in the preliminary screening results, specifically:

[0052]

[0053] in, It is the first in the preliminary screening results of detailed structural parameters. A detailed structural parameter, It is the first in the preliminary screening results of detailed structural parameters. The coefficients of determination are obtained by regression analysis with one microstructural parameter as the dependent variable and the remaining microstructural parameters as independent variables.

[0054] The regression analysis is a multiple linear regression analysis;

[0055] C2. Obtain the screening results of asphalt mixture microstructure parameters by using the variance expansion factor of each microstructure parameter in the preliminary screening results. Specifically:

[0056] By comparing the variance expansion factor of each microstructure parameter in the preliminary screening results with the VIF threshold, the microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of the microstructure parameters of asphalt mixture.

[0057] The beneficial effects of this invention are as follows:

[0058] This invention combines correlation analysis and visualization techniques to propose a method for identifying the correlation of microstructural parameters based on the Pearson correlation coefficient matrix. By setting correlation frequency and coefficient thresholds, a parameter elimination mechanism is constructed to improve the systematicness and accuracy of the screening process. Furthermore, a variance inflation factor is introduced to assess the risk of collinearity, thus establishing a final screening system for microstructural parameters. This invention utilizes collinearity quantification technology for the highly efficient and visual screening and identification of microstructural parameters in asphalt mixtures. It can systematically reveal the redundant relationships and degree of collinearity among microstructural parameters and intuitively extract key indicators that significantly affect performance, thereby optimizing the modeling input system and improving the simplicity and stability of the analytical equations. This invention helps to overcome the bottleneck problem of "high dimensionality and low efficiency" in structure-performance relationship modeling, avoids the loss of key mechanistic information, improves the universality of subsequent engineering models, and reduces the deviation between subsequent calculation results and engineering reality. Attached Figure Description

[0059] Figure 1 The Pearson correlation matrix is ​​the microstructural characteristic parameter related to fractals and multifractals.

[0060] Figure 2 The result is the variance inflation factor calculated for the remaining parameters. Detailed Implementation

[0061] Specific Implementation Method 1: The specific process of this implementation method for visually screening asphalt mixture microstructure parameters based on collinearity quantification technology is as follows:

[0062] Step 1: Obtain the microstructural parameters of the asphalt mixture and the correlation coefficients between these parameters. Visualize the correlation coefficients between the microstructural parameters of the asphalt mixture. Specifically:

[0063] Step 1: Obtain the microstructure parameters of the asphalt mixture and the Pearson correlation coefficients between the microstructure parameters of the asphalt mixture, and construct the correlation coefficient matrix.

[0064] The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar.

[0065] The skeleton structure parameters include: the fractal dimension of the aggregate skeleton structure, the width of the multifractal spectrum of the aggregate skeleton structure, the height difference of the multifractal spectrum of the aggregate skeleton structure, and the peak height of the aggregate skeleton structure.

[0066] The void structure parameters include: void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0067] The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure.

[0068] The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0069] The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure;

[0070] The correlation coefficient matrix contains the Pearson correlation coefficient values ​​between any one of the selected microstructure parameters and all other microstructure parameters.

[0071] Step 1 and Step 2: Visualize the correlation coefficient matrix using a heatmap or color-coded matrix based on the absolute value of the correlation coefficient, analyze the visualization results, and preliminarily identify the parameter combinations with strong linear relationships.

[0072] Specifically, the visualization of the correlation coefficient matrix is ​​as follows: the larger the absolute value of the correlation coefficient, the darker the color, such as... Figure 1 As shown;

[0073] Step 2: Obtain the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture. Compare the correlation frequency of structural parameters with the preset frequency threshold. If the correlation frequency of all structural parameters is less than the preset frequency threshold, proceed to Step 5; otherwise, obtain the highest correlation frequency of structural parameters. If the highest correlation frequency of structural parameters is one, proceed to Step 4; otherwise, proceed to Step 3.

[0074] The process of obtaining the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of asphalt mixtures is specifically as follows:

[0075] First, set the threshold for the Pearson correlation coefficient;

[0076] Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. ;

[0077] The preset frequency threshold is a natural number. When it is zero, it means that the filter does not allow strong correlations (i.e., the Pearson correlation coefficient exceeds the threshold).

[0078] Step 3: Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to Step 2;

[0079] Step 4: Remove the asphalt mixture microstructure parameters corresponding to the highest correlation frequency of structural parameters, and return to Step 2;

[0080] Step 5: Obtain the variance expansion factor of the remaining asphalt mixture microstructure parameters. Use the variance expansion factor of the remaining asphalt mixture microstructure parameters to obtain the screening results of the asphalt mixture microstructure parameters, specifically:

[0081] Step 51: Obtain the variance expansion factor of the remaining asphalt mixture's microstructure parameters, specifically:

[0082]

[0083] in, It is the first The variance inflation factor of each residual microstructure parameter The coefficient of determination is obtained by regression analysis with the i-th residual microstructure parameter as the dependent variable and the remaining residual microstructure parameters as independent variables.

[0084] The regression analysis is a multiple linear regression analysis, which does not consider the interaction effects between parameters or feature engineering.

[0085] Step 5.2: Obtain the screening results of the microstructure parameters of the asphalt mixture using the variance expansion factor of the remaining asphalt mixture microstructure parameters, specifically as follows:

[0086] By comparing the variance expansion factor of each remaining asphalt mixture microstructure parameter with the VIF threshold, the remaining asphalt mixture microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of asphalt mixture microstructure parameters.

[0087] The VIF threshold is usually set to 10, but a more stringent threshold is 5.

[0088] This step is based on the collinearity quantization method to obtain the microstructural parameters of asphalt mixtures;

[0089] Specific Implementation Method 2: A visualization screening system for microstructure parameters of asphalt mixtures based on collinearity quantification technology includes: a correlation coefficient acquisition module, a microstructure parameter pre-screening module, and a microstructure parameter final screening module;

[0090] The correlation coefficient acquisition module is used to obtain the correlation coefficients between the microstructural parameters of asphalt mixtures, specifically:

[0091] A1. Obtain the microstructure parameters of asphalt mixtures and the Pearson correlation coefficients between the microstructure parameters of asphalt mixtures, and construct the correlation coefficient matrix.

[0092] The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar.

[0093] The skeleton structure parameters include: the fractal dimension of the aggregate skeleton structure, the width of the multifractal spectrum of the aggregate skeleton structure, the height difference of the multifractal spectrum of the aggregate skeleton structure, and the peak height of the aggregate skeleton structure.

[0094] The void structure parameters include: void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0095] The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure.

[0096] The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference;

[0097] The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure;

[0098] The elements in the correlation coefficient matrix are the Pearson correlation coefficients between two microstructure parameters;

[0099] A2. Visualize the correlation coefficient matrix using a heatmap or color-coded matrix format according to the absolute value of the correlation coefficient.

[0100] The microstructure parameter pre-screening module is used to obtain preliminary screening results of asphalt mixture microstructure parameters based on the correlation coefficients between them. Specifically:

[0101] B1. Obtain the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture. Compare the correlation frequency of structural parameters with a preset frequency threshold. If the correlation frequency of all structural parameters is less than the preset frequency threshold, then the current microstructural parameter is used as the preliminary screening result of the microstructural parameters of the asphalt mixture, and the final screening module of the microstructural parameters is executed; otherwise, obtain the highest value of the correlation frequency of the structural parameters. If there is only one highest value of the correlation frequency of the structural parameters, then execute B3; if there are multiple highest values ​​of the correlation frequency of the structural parameters, then execute B2.

[0102] The process of obtaining the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of asphalt mixtures is specifically as follows:

[0103] First, set the threshold for the Pearson correlation coefficient;

[0104] Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. .

[0105] B2. Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to B1;

[0106] B3. Remove the asphalt mixture microstructure parameters corresponding to the highest frequency of correlation of structural parameters, and return to B1;

[0107] The final screening module for microstructure parameters uses the preliminary screening results of the microstructure parameters of asphalt mixtures to obtain the final screening results of the microstructure parameters of asphalt mixtures, specifically:

[0108] C1. Obtain the variance inflation factor for each microstructure parameter in the preliminary screening results, specifically:

[0109]

[0110] in, It is the first in the preliminary screening results of detailed structural parameters. A detailed structural parameter, It is the first in the preliminary screening results of detailed structural parameters. The coefficients of determination are obtained by regression analysis with one microstructural parameter as the dependent variable and the remaining microstructural parameters as independent variables.

[0111] The regression analysis is a multiple linear regression analysis;

[0112] C2. Obtain the screening results of asphalt mixture microstructure parameters by using the variance expansion factor of each microstructure parameter in the preliminary screening results. Specifically:

[0113] By comparing the variance expansion factor of each microstructure parameter in the preliminary screening results with the VIF threshold, the microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of the microstructure parameters of asphalt mixture.

[0114] This invention combines correlation analysis and visualization techniques to propose a method for identifying the correlation of structural parameters based on the Pearson correlation coefficient matrix. By setting correlation frequency and coefficient thresholds, a parameter elimination mechanism is constructed to improve the systematicness and accuracy of the screening process. Furthermore, a variance inflation factor is introduced to assess the risk of collinearity, thus establishing a final screening system for macroscopic structural parameters. The visualization screening method and system for macroscopic structural parameters of asphalt mixtures proposed in this invention, based on collinearity quantification technology, offers high identification efficiency and strong visibility. It can systematically reveal the redundant relationships and degree of collinearity among structural parameters and intuitively extract key indicators that significantly affect performance, thereby optimizing the modeling input system and improving the simplicity and stability of the analytical equations.

[0115] Example: To verify the beneficial effects of the present invention, the following experiments were conducted:

[0116] For asphalt mixture AC-16, with a correlation coefficient threshold of 0.5, the Pearson correlation matrix of the fractal and multifractal correlation microstructural characteristic parameters is as follows: Figure 1 As shown in Table 1, the calculation results of the microstructural analysis parameters and the correlation frequencies and correlation coefficients are as follows:

[0117] Table 1

[0118] serial number parameter Parameter Description Correlation frequency f Correlation coefficient and SUM 1# D_a Fractal dimension of aggregate skeleton structure <![CDATA[ 2 ]]> <![CDATA[ 2.308 ]]> 2# D_v Fractal dimension of void structure 0 0.801 3# Δα_a The multifractal width of the aggregate skeleton structure 2 2.208 4# Δα_v Void structure multifractal spectrum width 1 0.787 5# Δf(α)_a The aggregate skeleton structure has a multifractal spectrum height difference. 1 1.843 6# Δf(α)_v Void structure multifractal spectrum height difference 1 0.842 7# <![CDATA[f max _a]]> Peak height of aggregate skeleton structure 1 2.090

[0119] Step 1: Select the correlation frequency threshold as 0. Based on the correlation frequency, correlation coefficient and calculation results, delete parameter D_a, as shown in Table 2.

[0120] Table 2

[0121] serial number parameter f SUM 1# D_a - - 2# D_v 0 0.575 3# Δα_a <![CDATA[ 1 ]]> <![CDATA[ 1.532 ]]> 4# Δα_v 1 0.780 5# Δf(α)_a 1 1.381 6# Δf(α)_v 1 0.801 7# <![CDATA[f max _a]]> 0 1.195

[0122] Step 2: Based on the recalculation results of the correlation frequency and correlation coefficient, delete the parameter Δα_a, as shown in Table 3.

[0123] Table 3

[0124] serial number parameter f SUM 1# D_a - - 2# D_v 0 0.365 3# Δα_a - - 4# Δα_v 1 0.751 5# Δf(α)_a 0 0.593 6# Δf(α)_v <![CDATA[ 1 ]]> <![CDATA[ 0.788 ]]> 7# <![CDATA[f max _a]]> 0 0.703

[0125] Step 3: Based on the recalculation results of the correlation frequency and correlation coefficient, delete the parameter Δf(α)_v, as shown in Table 4;

[0126] Table 4

[0127] serial number parameter f SUM 1# D_a - - 2# D_v 0 0.325 3# Δα_a - - 4# Δα_v 0 0.034 5# Δf(α)_a 0 0.592 6# Δf(α)_v - - 7# <![CDATA[f max _a]]> 0 0.674

[0128] Step 4: Set the variance inflation factor threshold to 5 and check whether the variance inflation factors of the remaining parameters meet the requirements. The results show that all three parameters meet the requirements. Figure 2 As shown, when the variance inflation factor threshold is set to 10, all four parameters meet the requirements.

[0129] The calculation method of this invention is simple, which improves the versatility of subsequent engineering models. At the same time, this invention has a small amount of calculation and high calculation efficiency.

Claims

1. A method for visually screening microstructural parameters of asphalt mixtures based on collinearity quantification technology, characterized in that... The specific process of the method is as follows: Step 1: Obtain the microstructure parameters of the asphalt mixture and the correlation coefficients between the microstructure parameters of the asphalt mixture; Step 2: Obtain the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture, and compare the correlation frequency of structural parameters with the preset frequency threshold. If the correlation frequency of all structural parameters is less than the preset frequency threshold, proceed to Step 5. Otherwise, obtain the highest frequency value of the structural parameter correlation. If there is only one highest frequency value of the structural parameter correlation, proceed to step four; if there are multiple highest frequency values ​​of the structural parameter correlation, proceed to step three. Step 3: Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to Step 2; Step 4: Remove the asphalt mixture microstructure parameters corresponding to the highest correlation frequency of structural parameters, and return to Step 2; Step 5: Obtain the variance expansion factor of the remaining asphalt mixture microstructure parameters, and use the variance expansion factor of the remaining asphalt mixture microstructure parameters to obtain the screening results of the asphalt mixture microstructure parameters.

2. The method for visually screening the microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 1, characterized in that: The step one, obtaining the microstructure parameters of the asphalt mixture and the correlation coefficients between these parameters, specifically involves: Step 1: Obtain the microstructure parameters of the asphalt mixture and the Pearson correlation coefficients between the microstructure parameters of the asphalt mixture, and construct the correlation coefficient matrix. The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar. The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure. The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference; The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure; The elements in the correlation coefficient matrix are the Pearson correlation coefficients between two microstructure parameters; Steps 1 and 2: Visualize the correlation coefficient matrix according to the absolute value of the correlation coefficient.

3. The method for visually screening the microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 2, characterized in that: The step two, obtaining the correlation frequency of structural parameters based on the correlation coefficients between the microstructural parameters of the asphalt mixture, specifically involves: First, set the threshold for the Pearson correlation coefficient; Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. .

4. The method for visually screening microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 3, characterized in that: The variance expansion factor for obtaining the microstructure parameters of the remaining asphalt mixture in step five is specifically as follows: in, It is the first The variance inflation factor of each residual microstructure parameter The coefficient of determination is obtained by regression analysis with the i-th residual microstructure parameter as the dependent variable and the remaining residual microstructure parameters as independent variables. The regression analysis is a multiple linear regression analysis.

5. The method for visually screening microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 4, characterized in that: The step five, which involves using the variance expansion factor of the remaining asphalt mixture microstructure parameters to obtain the screening results of the asphalt mixture microstructure parameters, specifically involves: By comparing the variance expansion factor of each remaining asphalt mixture microstructure parameter with the VIF threshold, the remaining asphalt mixture microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of asphalt mixture microstructure parameters.

6. A visualization and screening system for microstructural parameters of asphalt mixtures based on collinearity quantification technology, characterized in that: The system includes: a correlation coefficient acquisition module, a microstructure parameter pre-screening module, and a microstructure parameter final screening module; The correlation coefficient acquisition module is used to acquire the correlation coefficients between the microstructure parameters of asphalt mixtures. The microstructure parameter pre-screening module is used to obtain preliminary screening results of asphalt mixture microstructure parameters based on the correlation coefficients between the microstructure parameters of asphalt mixtures. The final screening module for microstructure parameters uses the preliminary screening results of the microstructure parameters of asphalt mixtures to obtain the final screening results of the microstructure parameters of asphalt mixtures.

7. The visualization and screening system for microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 6, characterized in that: The correlation coefficient acquisition module specifically includes: A1. Obtain the microstructure parameters of asphalt mixtures and the Pearson correlation coefficients between the microstructure parameters of asphalt mixtures, and construct the correlation coefficient matrix. The microstructural parameters are structural parameters in three aspects: skeleton, voids, and mortar. The skeleton structure parameters include: the fractal dimension of the aggregate skeleton structure, the width of the multifractal spectrum of the aggregate skeleton structure, the height difference of the multifractal spectrum of the aggregate skeleton structure, and the peak height of the aggregate skeleton structure. The void structure parameters include: void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference; The skeleton structure parameters include: particle contact characteristics, particle distribution characteristics, particle morphology parameters, fractal dimension of aggregate skeleton structure, multifractal spectrum width of aggregate skeleton structure, multifractal spectrum height difference of aggregate skeleton structure, and peak height of aggregate skeleton structure. The void structure parameters include: void distribution characteristics, porosity, void shape characteristics, void structure fractal dimension, void structure multifractal spectrum width, and void structure multifractal spectrum height difference; The parameters of mortar structure include: mortar thickness and uniformity, dispersion index, topological parameters of mortar structure, fractal dimension of mortar structure, multifractal spectrum width of mortar structure, and multifractal spectrum height difference of mortar structure; The elements in the correlation coefficient matrix are the Pearson correlation coefficients between two microstructure parameters; A2. Visualize the correlation coefficient matrix using a heatmap or color-coded matrix format according to the absolute value of the correlation coefficient.

8. The visualization and screening system for microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 7, characterized in that: The microstructure parameter pre-screening module specifically includes: B1. Obtain the correlation frequency of structural parameters based on the correlation coefficient between the microstructure parameters of asphalt mixture, compare the correlation frequency of structural parameters with the preset frequency threshold, and if the correlation frequency of all structural parameters is less than the preset frequency threshold, then take the current microstructure parameter as the preliminary screening result of the microstructure parameter of asphalt mixture, and execute the final screening module of microstructure parameter. Otherwise, obtain the highest frequency value of the structural parameter correlation. If there is only one highest frequency value of the structural parameter correlation, execute B3; if there are multiple highest frequency values ​​of the structural parameter correlation, execute B2. B2. Obtain the sum of the absolute values ​​of the correlation coefficients between the microstructural parameter corresponding to the highest correlation frequency value of each structural parameter and other microstructural parameters, remove the microstructural parameter corresponding to the maximum value of the absolute value of the correlation coefficient, and return to B1; B3. Remove the asphalt mixture microstructure parameters corresponding to the highest frequency of correlation of structural parameters, and return to B1.

9. The visualization and screening system for microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 8, characterized in that: Specifically, in B1, the correlation frequency of structural parameters is obtained based on the correlation coefficient between the microstructural parameters of asphalt mixtures. First, set the threshold for the Pearson correlation coefficient; Then, among the Pearson correlation coefficients of the asphalt mixture microstructure parameter A and each microstructure parameter, the number 'a' of Pearson correlation coefficients greater than the Pearson correlation coefficient threshold is obtained, and 'a' is taken as the structural parameter correlation frequency of the asphalt mixture microstructure parameter A. .

10. The visualization and screening system for microstructural parameters of asphalt mixtures based on collinearity quantification technology according to claim 9, characterized in that: The final screening module for the microstructure parameters specifically includes: C1. Obtain the variance inflation factor for each microstructure parameter in the preliminary screening results, specifically: in, It is the first in the preliminary screening results of detailed structural parameters. A detailed structural parameter, It is the first in the preliminary screening results of detailed structural parameters. The coefficients of determination are obtained by regression analysis with one microstructural parameter as the dependent variable and the remaining microstructural parameters as independent variables. The regression analysis is a multiple linear regression analysis; C2. Obtain the screening results of asphalt mixture microstructure parameters by using the variance expansion factor of each microstructure parameter in the preliminary screening results. Specifically: By comparing the variance expansion factor of each microstructure parameter in the preliminary screening results with the VIF threshold, the microstructure parameters corresponding to variance expansion factors greater than the VIF threshold are deleted, thus obtaining the screening results of the microstructure parameters of asphalt mixture.

Images