Asphalt mixture dynamic modulus prediction method and system based on prior knowledge fusion

Abstract

The invention provides an asphalt mixture dynamic modulus prediction method and system based on prior knowledge fusion. The method comprises the following steps: S1, acquiring the dynamic modulus and characteristic variables of the asphalt mixture; s2, optimizing the characteristic variables based on a priori knowledge model to obtain key characteristic variables, and constructing a dynamic modulus data set according to the key characteristic variables and the dynamic modulus of the asphalt mixture; s3, constructing an asphalt mixture dynamic modulus prediction model by adopting a machine learning method based on prior knowledge fusion, and performing model parameter searching training; s4, evaluating the prediction precision of the asphalt mixture dynamic modulus prediction model after parameter searching training, and preferably selecting the asphalt mixture dynamic modulus prediction model with the best fitting effect; and S5, inputting the key characteristic variables of the to-be-predicted asphalt mixture into the optimal asphalt mixture dynamic modulus prediction model to obtain the dynamic modulus of the to-be-predicted asphalt mixture. The technical problem that a traditional model is poor in comprehensive performance is solved.

Description

Technical Field

[0001] The present invention relates to the field of prediction of mechanical properties of asphalt mixtures, and in particular to a method and system for predicting the dynamic modulus of asphalt mixtures based on prior knowledge fusion. Background Art

[0002] Asphalt pavements in high-temperature areas are prone to rutting and other defects under repeated heavy loads. In order to obtain asphalt mixtures that meet the service requirements of roads in high-temperature areas, it is necessary to control the high-temperature performance of asphalt mixtures.

[0003] According to the US MEPDG mechanical-empirical design method, dynamic modulus is a key indicator for measuring the high-temperature performance of asphalt mixtures. However, the dynamic modulus test specimens for asphalt mixtures are complex to prepare and require significant testing time and cost. Therefore, empirical formulas or data-driven methods are often used to predict the dynamic modulus of asphalt mixtures. However, while empirical formulas are well-suited for various asphalt mixtures, their predictive accuracy is often limited. Furthermore, data-driven models, while capable of high accuracy, suffer from poor generalization, making them difficult to apply. Summary of the Invention

[0004] The purpose of the present invention is to propose a method and system for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion, which solves the technical problem of the lack of comprehensive performance of traditional models.

[0005] To achieve the above object, the technical solution of the present invention is as follows:

[0006] The present invention proposes a method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion, which includes the following steps:

[0007] S1. Obtain the dynamic modulus of asphalt mixture and its characteristic variables;

[0008] S2. Based on the prior knowledge model, the characteristic variables are optimized to obtain the key characteristic variables. The key characteristic variables are used as input and the dynamic modulus of asphalt mixture is used as output to construct a dynamic modulus data set.

[0009] S3. Use a machine learning method based on prior knowledge fusion to build an asphalt mixture dynamic modulus prediction model, and perform parameter training on the asphalt mixture dynamic modulus prediction model;

[0010] S4. Using data statistical analysis indicators to evaluate the prediction accuracy of the asphalt mixture dynamic modulus prediction model after parameter search and training, and selecting the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion with the best fitting effect;

[0011] S5. Input the key characteristic variables of the asphalt mixture to be predicted into the preferred asphalt mixture dynamic modulus prediction model to obtain the dynamic modulus of the asphalt mixture to be predicted.

[0012] Preferably, the dynamic modulus of the asphalt mixture and its characteristic variables are obtained from public databases, public literature or laboratory measured data.

[0013] Preferably, the characteristic variables include the cumulative sieve retention percentage (%) of 19mm sieve hole, the cumulative sieve retention percentage (%) of 9.5mm, the cumulative sieve retention percentage (%) of 4.75mm, the cumulative sieve retention percentage (%) of 0.075mm, the void ratio (%), the effective asphalt content (%), the mineral gap ratio, the asphalt saturation (%), the logarithm of asphalt viscosity (cps), the dynamic modulus loading frequency (Hz), the temperature (℃), the asphalt viscosity parameter A, the asphalt viscosity parameter VTS, the logarithm of the dynamic modulus of asphalt (psi), and the logarithm of the asphalt phase angle (°).

[0014] Preferably, the prior knowledge model includes a Witczak model, an improved Witczak model, a modified Bari-Witczak model, a Hirsch model, a modified Hirsch model, an AI-Khateeb model, a Global model, and a simplified Global model.

[0015] Preferably, the feature variables are optimized based on the prior knowledge model to obtain key feature variables; specifically, it includes: selecting the prior knowledge model according to the type of feature variables obtained in step S1, and setting the input parameters of the selected prior knowledge model as key feature variables to reduce the workload of feature selection in feature engineering.

[0016] Preferably, in step S3, the following steps are specifically included:

[0017] S31. Based on the selected prior knowledge model and the key characteristic variable values, calculate the prior prediction values ​​of the dynamic modulus of the asphalt mixture for all samples in the dynamic modulus data set;

[0018] S32. Subtract the actual dynamic modulus values ​​of all samples in the dynamic modulus data set from the prior predicted values ​​of the dynamic modulus of the asphalt mixture to obtain the dynamic modulus difference values ​​of the asphalt mixture for all samples; use the key characteristic variable values ​​as the actual input of the machine learning model, use the dynamic modulus difference values ​​of the asphalt mixture as the output of the machine learning model, and construct a data set based on the key characteristic variable values ​​and the corresponding dynamic modulus difference values ​​of the asphalt mixture;

[0019] S33, dividing the data set constructed in S32 into a training set and a test set, and normalizing the training set and the test set respectively; wherein the normalized training set is used for parameter training of the machine learning model, and the normalized test set is used for evaluation of the machine learning model;

[0020] S34. Use one or more machine learning algorithms to participate in parameter search and training of the machine learning model: use a genetic algorithm to search for parameters of the machine learning model, and train the machine learning model based on the optimal model hyperparameters obtained from the parameter search;

[0021] S35. Combining the machine learning model after parameter search training and the prior knowledge model, construct an asphalt mixture dynamic modulus prediction model based on prior knowledge fusion; the output of the constructed asphalt mixture dynamic modulus prediction model is the sum of the prediction results of the machine learning model and the prior knowledge model.

[0022] Preferably, the optimization target of the genetic algorithm is the minimum determination coefficient R of the k-fold cross validation of the training set. 2 score.

[0023] Preferably, the machine learning algorithm includes support vector machine regression, multi-layer perceptron, random forest, and gradient boosting tree.

[0024] Preferably, the data statistical analysis indicators include the coefficient of determination R2, mean square error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The closer R2 is to 1, the smaller the MSE, MAE, and MAPE are, the better the model fitting effect is.

[0025] The present invention also proposes a dynamic modulus prediction system for asphalt mixture based on prior knowledge fusion, including a processor, a memory and a computer program stored on the memory. When the processor executes the computer program, it specifically executes any step in the above-mentioned asphalt mixture dynamic modulus prediction method.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The present invention can quickly select key feature variables that play an important role in improving the performance of the dynamic modulus prediction model based on the prior knowledge of the empirical formula method, thereby reducing the workload of feature engineering in the machine learning method; further, the predicted value of the empirical formula prior knowledge model is introduced into the machine learning model prediction task, and the empirical formula method and the data-driven method are organically integrated to further improve the prediction accuracy of the dynamic modulus. Since the dynamic modulus prediction model relies on the prior knowledge model, it has better generalization and a wider range of application scenarios. At the same time, its data-driven core ensures the prediction accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Flowchart of the asphalt mixture dynamic modulus prediction method based on prior knowledge fusion of the present invention;

[0029] Figure 2 The structure of the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion and its calculation flow diagram of the present invention;

[0030] Figure 3 Schematic diagram of the prediction performance of the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion of the present invention. DETAILED DESCRIPTION

[0031] The following is combined with Figure 1-3 The specific implementation methods of the embodiments of the present invention are described in detail to demonstrate the implementation process of the technical effects of this application and enhance the understanding of this technical solution. It should be noted that the specific implementation methods described here are only used to illustrate and explain the embodiments of the present invention and are not used to limit the present invention.

[0032] Asphalt pavements in high-temperature areas are prone to rutting and other defects under repeated heavy loads. In order to obtain asphalt mixtures that meet the service requirements of roads in high-temperature areas, it is necessary to control the high-temperature performance of asphalt mixtures.

[0033] According to the US MEPDG mechanical-empirical design methodology, dynamic modulus is a key indicator of the high-temperature performance of asphalt mixtures. Currently, empirical formulas or data-driven methods are often used to predict the dynamic modulus of asphalt mixtures. However, while empirical formulas are well-suited for various asphalt mixtures, their predictive accuracy is often limited. While data-driven models can achieve high accuracy, their generalization capabilities are poor, hindering widespread application. Therefore, there is an urgent need for an efficient method for predicting the dynamic modulus of asphalt mixtures that combines generalizability with accuracy.

[0034] In order to solve the technical problems existing in the above-mentioned prior art, Figure 1 The present invention proposes a method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion, which includes the following steps:

[0035] S1. Obtain the dynamic modulus of asphalt mixture and its characteristic variables from public databases, public literature or laboratory measured data;

[0036] S2. Based on the prior knowledge model, the characteristic variables are optimized to obtain the key characteristic variables. The key characteristic variables are used as input and the dynamic modulus of asphalt mixture is used as output to construct a dynamic modulus data set.

[0037] S3. Use a machine learning method based on prior knowledge fusion to build an asphalt mixture dynamic modulus prediction model, and perform parameter training on the asphalt mixture dynamic modulus prediction model;

[0038] S4. Using data statistical analysis indicators to evaluate the prediction accuracy of the asphalt mixture dynamic modulus prediction model after parameter search and training, and selecting the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion with the best fitting effect;

[0039] S5. Input the key characteristic variables of the asphalt mixture to be predicted into the asphalt mixture dynamic modulus prediction model integrated with prior knowledge to obtain the dynamic modulus of the asphalt mixture to be predicted.

[0040] Reference Figure 2 A specific embodiment is shown:

[0041] S1. Data on the dynamic modulus and related characteristic variables of several asphalt mixtures were obtained from the public dataset NCHRP 9-19. The relevant characteristic variables in the public dataset include the cumulative sieve retention percentage (%) of 19 mm sieve aperture, the cumulative sieve retention percentage (%) of 9.5 mm sieve aperture, the cumulative sieve retention percentage (%) of 4.75 mm sieve aperture, the cumulative sieve retention percentage (%) of 0.075 mm sieve aperture, void fraction (%), effective asphalt content (%), logarithmic value of asphalt viscosity (cps), frequency (Hz), temperature (°C), asphalt viscosity parameter A, asphalt viscosity parameter VTS, logarithmic value of asphalt dynamic modulus (psi), and logarithmic value of asphalt phase angle (°).

[0042] S2. Select the improved Witczak model as the prior knowledge model. The formula of the prior knowledge model is as follows:

[0043]

[0044] Where: is the dynamic modulus of asphalt;

[0045] δ b is the asphalt phase angle;

[0046] p 0.075 The cumulative percentage of sieve residue with a 0.075 mm sieve hole (%);

[0047] p 4.75 The cumulative percentage of 4.75mm sieve residue (%);

[0048] p 9.5 The cumulative sieve residue percentage (%) is 9.5;

[0049] p 19 Cumulative sieve residue percentage of 19mm (%)

[0050] Va is the void ratio (%);

[0051] V beff is the effective asphalt content of asphalt mixture (%)

[0052] Based on the input of the improved Witczak prior knowledge model, the characteristic variables are optimized to obtain the key characteristic variables, which include the logarithmic value of the asphalt dynamic modulus lg(|G b * |)(psi), logarithm of asphalt phase angle lg(σ a )(°), 0.075mm cumulative sieve residue percentage p 0.075 (%), 4.75mm cumulative sieve residue percentage p 4.75 (%), 9.5mm cumulative sieve residue percentage p 9.5 (%), 19mm sieve hole cumulative sieve residue percentage p 19 (%), porosity V a (%), effective asphalt content V beff (%), frequency f m (Hz), temperature T (°C). Although frequency and temperature are not explicitly included in the improved Witczak prior knowledge model, they are actually included in the asphalt dynamic modulus and phase angle. Therefore, for ease of input, the frequency and temperature corresponding to the asphalt dynamic modulus and phase angle are explicitly used as key characteristic variables. Furthermore, a dynamic modulus dataset is constructed using the key characteristic variables as input and the logarithm of the asphalt mixture dynamic modulus lg|E*|(psi) as output.

[0053] The data in the collected dynamic modulus data set are calculated based on statistical parameters and are shown in Table 1 below:

[0054] Table 1. Statistical parameter calculation results of the collected dynamic modulus data of asphalt mixture and its key characteristic variables

[0055]

[0056]

[0057] This embodiment provides data on the dynamic modulus of a portion of representative asphalt mixtures through the improved Witczak model. In this embodiment, dynamic modulus and its related characteristic variable data can also be extracted from relevant published literature or laboratory measurement results, and other prior knowledge models can be used to select key characteristic variables as data sets. No limitation is made here.

[0058] The input of the prior knowledge model is usually based on the viscoelastic properties of asphalt, the proportion of aggregate, and the volumetric properties of the asphalt mixture. It can usually fully consider the characteristic variables that have a significant impact on the dynamic modulus of the asphalt mixture. The purpose of feature engineering is to select characteristic variables that have a significant impact on the target. Therefore, the integration of prior knowledge models can efficiently realize the selection of characteristic variables and significantly reduce the workload of engineers or road designers.

[0059] S3. Using a machine learning method based on prior knowledge fusion to construct an asphalt mixture dynamic modulus prediction model, and performing parameter training on the asphalt mixture dynamic modulus prediction model; in step S3, specifically including the following steps:

[0060] S31. Using the improved Witczak prior knowledge model, the prior prediction values ​​of the dynamic modulus of asphalt mixtures for all samples in the dynamic modulus database are calculated based on the values ​​of key characteristic variables;

[0061] S32. Subtract the actual dynamic modulus values ​​of all samples in the dynamic modulus data set from the prior predicted values ​​of the dynamic modulus of the asphalt mixture to obtain the delta values ​​of the dynamic modulus of the asphalt mixture for all samples. The actual input of the machine learning model is the key feature variable, and the output is the delta value of the dynamic modulus of the asphalt mixture;

[0062] S33. Divide the input and output of the machine learning method into a training set and a test set, where the training set and the test set account for 70-80% and 20-30% of the input and output, respectively, and then normalize the training set and the test set respectively;

[0063] S34. Select support vector machine regression, multi-layer perceptron, random forest, and gradient boosting tree models to participate in model parameter search and training. Only training set data can be used for parameter search and training. The genetic algorithm is used for parameter search. The optimization goal of the genetic algorithm is the minimum determination coefficient R of the training set k-fold cross validation. 2 Scoring; during training, the model hyperparameters are set to the optimal model hyperparameters obtained by parameter search, and the training set data is used to fit the above machine learning model.

[0064] S35. The prediction results of the machine learning model and the improved Witczak prior knowledge model are summed to obtain the output of the asphalt mixture dynamic modulus prediction model, and then the asphalt mixture dynamic modulus prediction model based on the prior knowledge fusion is constructed.

[0065] S4. Use statistical analysis indicators to evaluate the prediction accuracy of the asphalt mixture dynamic modulus prediction model after parameter search and training. The statistical analysis indicators include the determination coefficient R2, mean square error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The calculation formulas for each statistical analysis indicator are as follows:

[0066]

[0067] In the above formula: n is the number of samples, i is the number of input variables, lg(|E*|) predicted is the predicted value of the logarithmic modulus, lg(|E*|) measured is the predicted value of the logarithmic modulus.

[0068] In this embodiment, the statistical analysis index results of the dynamic modulus prediction performance of asphalt mixture using different machine learning methods based on the fusion of prior knowledge are shown in Table 2 below. Figure 3 This further reflects the prediction performance of the above models. Most of the data points of different models are located or distributed near the contour lines, which indicates that there is a good correlation between the predicted values ​​of the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion and the measured values. Furthermore, the asphalt mixture dynamic modulus prediction model composed of the prior knowledge model fusion gradient boosting tree model has the best performance.

[0069] Table 2. Statistical analysis of dynamic modulus prediction performance

[0070] MSE MAE MAPE <![CDATA[R 2 ]]> Support Vector Machine 0.02914 0.13157 0.02321 0.93786 Multilayer Perceptron 0.01474 0.09361 0.01651 0.96857 Random Forest 0.00833 0.06350 0.01153 0.98223 Gradient Boosting Trees 0.00566 0.05420 0.00963 0.98793

[0071] This embodiment establishes a prediction model for the dynamic modulus of asphalt mixtures by using a machine learning method based on the fusion of prior knowledge and data from a public dataset. This overcomes the shortcomings of traditional empirical formulas, which have limited prediction accuracy and poor generalization ability of data-driven models, and finds a method that is both generalizable and accurate and efficient for predicting the dynamic modulus of asphalt mixtures.

[0072] S5. Input the key characteristic variables of the asphalt mixture to be predicted into the asphalt mixture dynamic modulus prediction model based on prior knowledge to obtain the dynamic modulus of the asphalt mixture. The resulting dynamic modulus of the asphalt mixture is based on the input frequency and temperature, corresponding to the load frequency and service environment temperature of actual asphalt roads. By controlling the dynamic modulus of the asphalt mixture, a mixture design solution that meets the high-temperature service requirements of actual asphalt roads can be quickly obtained, thereby controlling the high-temperature performance of the asphalt mixture.

[0073] In summary, dynamic modulus is a key indicator for characterizing the high-temperature performance of asphalt mixture. In view of the shortcomings of existing technologies, the prior knowledge based on the empirical formula method is used to quickly select key feature variables that play an important role in improving the performance of the dynamic modulus prediction model, thereby reducing the workload of feature engineering in the machine learning method. Furthermore, the predicted value of the empirical formula prior knowledge model is introduced into the machine learning model prediction task, and the empirical formula method and the data-driven method are organically integrated to further improve the prediction accuracy of the dynamic modulus. Since the dynamic modulus prediction model relies on the prior knowledge model, it has better generalization and a wider range of application scenarios. At the same time, its data-driven core ensures the prediction accuracy of the model, guarantees the high-temperature performance of asphalt mixture, and provides a solid foundation for road design.

[0074] The above embodiments provide a detailed introduction to the present invention and use specific schemes to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, within the technical concept of the embodiments of the present invention, the technical solutions of the present invention can be arbitrarily simply combined, but as long as they do not violate the ideas of the embodiments of the present invention, they should also be regarded as being within the scope of protection of the present invention.

Claims

1. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion, characterized by: The following steps are involved: S1. Obtain the dynamic modulus of asphalt mixture and its characteristic variables; S2. Based on the prior knowledge model, the characteristic variables are optimized to obtain the key characteristic variables. The key characteristic variables are used as input and the dynamic modulus of asphalt mixture is used as output to construct a dynamic modulus data set. S3. Use a machine learning method based on prior knowledge fusion to build an asphalt mixture dynamic modulus prediction model, and perform parameter training on the asphalt mixture dynamic modulus prediction model; S4. Using data statistical analysis indicators to evaluate the prediction accuracy of the asphalt mixture dynamic modulus prediction model after parameter search and training, and selecting the asphalt mixture dynamic modulus prediction model based on prior knowledge fusion with the best fitting effect; S5. Input the key characteristic variables of the asphalt mixture to be predicted into the preferred asphalt mixture dynamic modulus prediction model to obtain the dynamic modulus of the asphalt mixture to be predicted.

2. The method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 1 is characterized in that: The dynamic modulus of the asphalt mixture and its characteristic variables are obtained from public databases, public documents or laboratory measured data.

3. The method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 1 is characterized in that: The characteristic variables include the cumulative sieve retention percentage of 19mm sieve hole, the cumulative sieve retention percentage of 9.5mm, the cumulative sieve retention percentage of 4.75mm, the cumulative sieve retention percentage of 0.075mm, the void ratio, the effective asphalt content, the mineral gap ratio, the asphalt saturation, the logarithm of the asphalt viscosity, the dynamic modulus loading frequency, the temperature, the asphalt viscosity parameter, the asphalt viscosity parameter, the logarithm of the dynamic modulus of the asphalt, and the logarithm of the asphalt phase angle.

4. The method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 1 is characterized in that: The prior knowledge models include the Witczak model, the improved Witczak model, the modified Bari-Witczak model, the Hirsch model, the modified Hirsch model, the AI-Khateeb model, the Global model, and the simplified Global model.

5. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 4, characterized in that: The method of optimizing the feature variables based on the prior knowledge model to obtain key feature variables specifically includes: selecting the prior knowledge model according to the type of feature variables obtained in step S1, and setting the input parameters of the selected prior knowledge model as key feature variables to reduce the workload of feature selection in feature engineering.

6. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 1, characterized in that: In step S3, the following steps are specifically included: S31. Based on the selected prior knowledge model and the key characteristic variable values, calculate the prior prediction values ​​of the dynamic modulus of the asphalt mixture for all samples in the dynamic modulus data set; S32. Subtract the actual dynamic modulus values ​​of all samples in the dynamic modulus data set from the prior predicted values ​​of the dynamic modulus of the asphalt mixture to obtain the dynamic modulus difference values ​​of the asphalt mixture for all samples; use the key characteristic variable values ​​as the actual input of the machine learning model, use the dynamic modulus difference values ​​of the asphalt mixture as the output of the machine learning model, and construct a data set based on the key characteristic variable values ​​and the corresponding dynamic modulus difference values ​​of the asphalt mixture; S33, dividing the data set constructed in S32 into a training set and a test set, and normalizing the training set and the test set respectively; wherein the normalized training set is used for parameter training of the machine learning model, and the normalized test set is used for evaluation of the machine learning model; S34. Using one or more machine learning algorithms to participate in parameter search and training of the machine learning model: using a genetic algorithm to search for parameters of the machine learning model, and training the machine learning model based on the optimal model hyperparameters obtained from the parameter search; S35. Combining the machine learning model after parameter search training and the prior knowledge model, construct an asphalt mixture dynamic modulus prediction model based on prior knowledge fusion; the output of the constructed asphalt mixture dynamic modulus prediction model is the sum of the prediction results of the machine learning model and the prior knowledge model.

7. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to patent claim 6, characterized in that: The optimization goal of the genetic algorithm is to minimize the coefficient of determination R of the k-fold cross validation of the training set. 2 score.

8. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 6, characterized in that: The machine learning algorithms include support vector machine regression, multi-layer perceptron, random forest, and gradient boosting tree.

9. A method for predicting the dynamic modulus of asphalt mixture based on prior knowledge fusion according to claim 1, characterized in that: The data statistical analysis indicators include the determination coefficient R2, mean square error MSE, mean absolute error MAE and mean absolute percentage error MAPE.

10. A dynamic modulus prediction system for asphalt mixture based on prior knowledge fusion, characterized in that: The method comprises a processor, a memory and a computer program stored in the memory. When the processor executes the computer program, the method specifically performs the steps in the asphalt mixture dynamic modulus prediction method according to any one of claims 1 to 9.

Images