Binder-based asphalt pavement repair effect evaluation method

Abstract

The invention discloses an asphalt pavement repair effect evaluation method based on a binder, which comprises the following steps of: acquiring chemical composition, molecular structure and surface characteristic data of an old pavement material through multi-dimensional characterization technologies such as spectrum, electron microscope and nuclear magnetic resonance, and establishing a characteristic database containing parameters such as carbonyl index and softening point increment; and analyzing a matching rule in historical data by a machine learning algorithm, and predicting the chemical compatibility and the physical matching degree of the new binder and the old material. And quantitatively detecting the chemical bonding density, adsorption work and roughness of the interface layer by adopting equipment such as an atomic force microscope and an X-ray energy spectrum, and verifying an interface action mechanism by combining molecular simulation. Macromechanical properties are obtained through a pull-out test and a fatigue test, and a nonlinear correlation model of microscopic parameters, the pull strength and the shear strength is established through regression analysis and machine learning.

Description

Technical Field

[0001] This application relates to the field of civil engineering technology, and in particular to a method for evaluating the repair effect of asphalt pavement based on binder. Background Technology

[0002] Asphalt pavement accounts for over 90% of my country's highway network. During its long service life, it inevitably develops defects such as cracks, potholes, and loosening due to repeated traffic loads, temperature cycles, and moisture erosion. As a key interface material connecting old asphalt and new repair materials, the performance of the binder directly determines the bonding quality between the repair layer and the original pavement, thus affecting the service life and performance of the repaired pavement.

[0003] Existing methods for evaluating the repair effectiveness of adhesives have significant shortcomings: First, the understanding of the bonding mechanism remains superficial, focusing only on macroscopic mechanical indicators such as pull-out strength, lacking quantitative characterization of microscopic mechanisms such as interfacial chemical bond density, physical adsorption energy, and mechanical interlocking degree, resulting in a lack of theoretical basis for adhesive selection; second, the microstructure is disconnected from macroscopic performance, and interfacial nanoscale characteristics cannot effectively predict long-term repair effects; third, the evaluation cycle is as long as several years and there is a lack of accelerated evaluation methods; finally, the level of intelligence is low, relying on human experience for judgment, and data from different projects are isolated and scattered, making it impossible to achieve knowledge sharing and intelligent decision-making.

[0004] The aforementioned shortcomings lead to a series of engineering problems: improper selection of adhesives causes interface peeling shortly after repair, resulting in a rework rate as high as 15% to 30%, causing material waste and traffic disruption losses; the lack of scientific prediction models makes the design of repair schemes highly arbitrary, and the actual service life of some projects is less than 50% of the design value; the excessively long evaluation cycle delays the timing of maintenance decisions, allowing small defects to develop into large defects, and increasing maintenance costs exponentially; traditional manual evaluation is highly subjective and inconsistent, with different evaluators reaching differences of more than 20% in their conclusions; difficulties in knowledge accumulation hinder technological progress, similar problems recur, and the overall technical level of the industry improves slowly. Summary of the Invention

[0005] The main purpose of this application is to provide a method for evaluating the repair effect of asphalt pavement based on binders, in order to solve the problem that the understanding of the bonding mechanism proposed in the background technology is only superficial, focusing only on macroscopic mechanical indicators such as pull-out strength, and lacking quantitative characterization of microscopic mechanisms such as interfacial chemical bond density, physical adsorption energy, and mechanical interlocking degree, which leads to a lack of theoretical basis for binder selection.

[0006] To achieve the above objectives, this application provides the following technical solution: A method for evaluating the repair effect of asphalt pavement based on binders, the specific steps of which are as follows: S1. Fourier transform infrared spectroscopy and scanning electron microscopy were used to characterize the old pavement materials, nuclear magnetic resonance was used to analyze the molecular structure of the binder, and a material feature database was established using machine learning algorithms to predict the compatibility between the binder and the pavement materials. S2. The thickness of the interface layer was measured using atomic force microscopy, the chemical bonding density was quantitatively detected by X-ray photoelectron spectroscopy, the physical adsorption work was calculated based on the contact angle test, the mechanical interlocking degree was evaluated by three-dimensional laser scanning, and the interface interaction was verified by molecular dynamics simulation. S3. Extract microscopic parameters including interfacial bonding density, adsorption work, roughness index and penetration depth, conduct pull-out tests and fatigue tests to obtain macroscopic performance data, and use multivariate nonlinear regression and machine learning algorithms to establish a correlation prediction model between microscopic parameters and macroscopic performance. S4. Install fiber optic grating sensors in the repair test section for in-situ monitoring, design multi-factor coupled tests of temperature cycling, water immersion and dynamic load, take samples regularly to characterize interface degradation features, and establish a life prediction model based on the accelerated aging coefficient. S5. Determine the optimization objective based on the correlation model, use genetic algorithm or particle swarm algorithm to perform multi-objective optimization of the formula, and verify the optimization effect through microscopic characterization and macroscopic performance testing. S6. Use portable testing equipment on site to quickly determine the microscopic parameters of the interface, conduct pull-out tests and non-destructive testing, establish a multi-dimensional evaluation index system that includes microscopic, macroscopic and durability, and use a prediction model to output a comprehensive score. S7. Establish standardized data collection templates, build a cloud-based knowledge base platform for repair effects, optimize prediction models through big data mining and analysis, and develop an intelligent decision support system.

[0007] Preferably, step S1 is performed in the following manner: S1.1 The wavenumber of old asphalt at 1700 cm⁻¹ was determined using a Fourier transform infrared spectrometer. -1 The carbonyl index was calculated by measuring the absorption peak intensity, the softening point temperature was determined by differential scanning calorimetry to assess the degree of aging, the surface morphology of the aggregate was observed by scanning electron microscopy to obtain the roughness value, the mineral composition was analyzed by X-ray diffraction, the chemical shift and content of the active functional groups of the binder were determined by nuclear magnetic resonance spectroscopy, and the number-average molecular weight and polydispersity index were determined by gel permeation chromatography. A multidimensional characteristic parameter set including carbonyl index, softening point increment, surface roughness, mineral composition, functional group content and molecular weight parameters was established. S1.2 Construct a material feature database containing characteristic data and corresponding bonding performance indicators of pavement materials with different aging degrees, aggregates of different mineral types, and different types of binders. Extract carbonyl index, softening point increment, surface free energy, roughness, functional group concentration, and molecular weight as input parameters. Use random forest algorithm or neural network algorithm to establish a prediction model. Use chemical compatibility index and physical matching degree as output variables. Optimize the model through k-fold cross-validation. Input the parameters of the material to be evaluated and output the matching score value. When the score value is greater than or equal to the preset threshold, the matching is judged to be good.

[0008] Preferably, step S2 is performed as follows: S2.1. The interface between the binder, old asphalt, and aggregate was scanned using an atomic force microscope to obtain the thickness distribution of the interface layer. The interface elements and chemical bonding type were analyzed using X-ray photoelectron spectroscopy. The interface chemical bonding density was calculated by peak fitting. The contact angle of the droplets on the old asphalt and aggregate surfaces was measured using a contact angle meter. The adhesion work was calculated based on the Young-Dupré equation, and the interface free energy was calculated using the Owens-Wendt method. The three-dimensional morphology of the aggregate surface was measured using a three-dimensional laser scanner. The surface roughness parameters Ra, Rz, and surface area ratio were extracted, and the correlation between roughness and mechanical interlocking strength was established. S2.2 Based on the experimental data in S2.1, molecular simulation software was used to construct models of binder molecules, aged asphalt, and aggregate surfaces. Periodic boundary conditions were set for molecular dynamics simulation. The COMPASS force field was used for energy minimization and equilibrium simulation. The interfacial binding energy and interaction energy were calculated, the interfacial stress distribution was analyzed, the contributions of hydrogen bonds and van der Waals forces were identified, and the interfacial layer thickness, number of bonding sites, and adsorption energy were calculated. The reliability of the experimental results was confirmed when the relative error between the simulation results and the experimental data was less than 15%.

[0009] Preferably, step S3 is performed as follows: S3.1 Extract the interfacial chemical bonding density, physical adsorption work, surface roughness index and penetration depth from step S2, prepare a composite specimen containing an old asphalt base layer, an adhesive interface layer and a repair material layer, conduct a pull-out test to determine the pull-out strength, conduct a shear test to determine the shear strength, conduct a dynamic fatigue test to record the number of cycles of failure, repeat the test for no less than 5 specimens in each group, calculate the average value and standard deviation of each performance index, and establish a sample dataset containing 4 micro parameters as input variables and 3 macro performance indexes as output variables; S3.2. Key micro-parameters were screened using grey relational analysis, and a prediction model Y=f(X1,X2,X3,X4) was established using multivariate nonlinear regression. The regression coefficients were fitted using the least squares method. When the coefficient of determination R... 2The model is considered effective when it is greater than or equal to 0.85. The model is then optimized using support vector machine, random forest, or neural network algorithms. The training and test sets are divided into a 7:3 or 8:2 ratio. The hyperparameters are optimized using k-fold cross-validation. The mean absolute percentage error (MAPE) is calculated. When the MAPE is less than or equal to 10%, the final association prediction model is established.

[0010] Preferably, step S4 is performed as follows: S4.1 Fiber Bragg grating strain sensors and temperature sensors are embedded in the adhesive interface layer of the repair test section at intervals of 2m to 5m. The strain measurement range is -3000με to +3000με, and the temperature measurement range is -40℃ to +80℃. Soil moisture sensors and triaxial accelerometers are also installed. Data acquisition frequency is 1 hour to 24 hours. A temperature cycling test is designed, with a high-temperature stage of 60℃ to 80℃ for 4 to 8 hours and a low-temperature stage of -20℃ to -10℃ for 4 to 8 hours, for a total of at least 30 cycles. A water immersion test is designed, with immersion for 24 to 72 hours followed by freeze-thaw cycles. The freezing temperature is -18℃ to -15℃ for 12 hours, and the thawing temperature is 20℃ to 25℃ for 12 hours, for a total of at least 15 cycles. A dynamic load test is designed, with a load amplitude of 0.5MPa to 0.9MPa and a frequency of 10Hz to 20Hz, for a total of 10 load cycles. 4 Next to 10 6 Next, the above experiments were combined into a multi-factor coupled accelerated experimental scheme according to the time series. S4.2. At least three specimens shall be sampled at the 0%, 25%, 50%, 75%, and 100% time points of the accelerated test, as well as at failure. The interfacial chemical bonding density, adsorption work, roughness, and penetration depth shall be determined using the method in step S2. The attenuation rate shall be calculated. Microcracks shall be observed using scanning electron microscopy to calculate the crack density. The porosity growth rate shall be determined using mercury intrusion porosimetry. Pull-out and shear tests shall be performed to calculate the strength retention rate. An accelerated aging model and a fitted apparent activation energy Ea shall be established based on the Arrhenius equation. The degradation kinetic model is as follows: t_real=t_test×exp[Ea / R×(1 / T_real-1 / T_test)]; P(t) = P0 × exp(-k × t^n); Where t_real is the time under actual service conditions, t_test is the time under accelerated test conditions, Ea is the apparent activation energy, R is the molar gas constant, T_real is the absolute temperature of the actual service environment, T_test is the absolute temperature of the accelerated test, P(t) is the performance parameter value at time t, P0 is the performance parameter value at the initial time, k is the degradation rate constant, and n is the degradation index. Failure is determined when the bond density decay rate is ≥50%, the strength retention rate is ≤60%, the crack density is ≥20mm / cm², or the porosity growth rate is ≥30%. The failure time t_failure is calculated, and the actual service life is calculated in combination with the acceleration factor. The specific method is as follows: L_service=t_failure×exp[Ea / R×(1 / T_real-1 / T_test)]; Where L_service is the actual service life and t_failure is the failure time under accelerated testing conditions.

[0011] Preferably, step S5 is performed as follows: S5.1. Based on the correlation prediction model established in step S3, determine the target values ​​of pull-out strength, shear strength, and fatigue life. Solve the target values ​​of micro-parameters in reverse. Determine the design variables of the binder formulation, including the content of base asphalt, the amount of modifier, the amount of emulsifier, and the amount of filler. Set the value range and constraints of each component. Establish a multi-objective optimization function and solve it using a genetic algorithm or a particle swarm optimization algorithm. For the genetic algorithm, set the population size to 50 to 200, the crossover probability to 0.6 to 0.9, and the mutation probability to 0.01 to 0.1. For the particle swarm optimization algorithm, set the number of particles to 30 to 100 and the learning factor to 1.5 to 2.5. Obtain the Pareto optimal solution set and select the optimal formulation using the ideal point method or the weighted method. S5.2 Prepare binder samples according to the optimal formula. Heat the base asphalt to 160℃ to 180℃, and shear it at high speed at 3000rpm to 5000rpm for 30 to 60 minutes. The emulsification temperature is 120℃ to 140℃, and the emulsification speed is 8000rpm to 12000rpm. Prepare no less than 5 sets of parallel samples. Use the method in step S2 to determine the optimized micro parameters, and use the method in step S3.1 to determine the optimized macro properties. Calculate the improvement rate, input the measured micro parameters into the correlation model to calculate the predicted value, and compare it with the measured value. When the pull-out strength improvement rate is ≥20%, the fatigue life improvement rate is ≥30%, and the cost increase rate is ≤15%, the optimization is deemed effective.

[0012] Preferably, step S6 is performed as follows: S6.1. Use a portable Fourier transform infrared spectrometer to scan the repaired section of the road surface after repair, with a wavenumber range of 400 cm⁻¹. -1 Up to 4000cm -1 The carbonyl index was determined, and the interfacial adhesion state was observed using a portable digital microscope at magnifications of 50x to 200x. The penetration depth of the adhesive was measured, and the arithmetic mean roughness Ra value was determined using a portable surface roughness meter, calculated per 100 μm. 2Pull-out tests were conducted at no fewer than three test points. A portable pull-out tester was used to determine the pull-out strength at a loading rate of 0.5 MPa / s to 1.0 MPa / s. A deflectometer was used to determine the rebound deflection value and calculate the deflection recovery rate. Ground penetrating radar was used to detect interlayer continuity at an antenna frequency of 1 GHz to 2 GHz. An infrared thermal imager was used to detect the temperature distribution. A permeability coefficient meter was used to test the water tightness. A pendulum friction coefficient meter was used to determine the pendulum value BPN. The microscopic parameters were input into the correlation prediction model in step S3 and the life prediction model in step S4 to calculate the predicted service life. S6.2 Establish a three-level evaluation index system encompassing microscopic, macroscopic, and durability levels. Set the qualification standards as follows: carbonyl index <0.8, penetration depth ≥1.0mm, roughness Ra ≥0.5mm, tensile strength ≥1.5MPa, deflection recovery rate ≥80%, defect rate <5%, water permeability coefficient <120ml / min, pendulum value BPN ≥45, and predicted lifespan ≥5 years. Use the analytic hierarchy process (AHP) or entropy weighting method to determine the weight coefficients for each level of index. Normalize each index to convert it into a score value from 0 to 100. Calculate the comprehensive score using the weighted summation method, or use the machine learning prediction model from step S3 to output the comprehensive score and performance level. Based on the comprehensive score, classify the levels as follows: 90-100 points (Excellent), 80-90 points (Good), 60-80 points (Medium), and below 60 points (Poor). Generate an evaluation report containing test data, scoring results, level determination, and improvement suggestions.

[0013] Preferably, step S7 is performed as follows: S7.1 Establish a standardized data collection template, which includes fields for basic project information, disease characteristics, material parameters, construction technology, microscopic characterization data, macroscopic performance data, environmental monitoring data, and usage effect data. Define the data type, value range, and required field identifiers for each field. Construct a cloud-based knowledge base platform using a B / S architecture, a microservice architecture for the backend, and a hybrid storage architecture for the database, including relational databases, object storage, and time-series databases. Set up a multi-level user permission management mechanism, use TLS encryption protocol and AES algorithm to ensure data security, establish a daily automatic backup mechanism, develop a data entry interface that supports manual input and batch import, set up data verification rules and review processes, and develop multi-dimensional query, statistical analysis, and GIS map visualization functions. S7.2. Descriptive statistical analysis is performed on historical data according to climate zone and traffic level to identify the optimal adhesive type and construction process parameters. Association rule mining is performed using the Apriori algorithm and FP-Growth algorithm, with a minimum support of 5% to 10% and a minimum confidence of 60% to 80%. K-means or hierarchical clustering algorithms are used to cluster failure cases to identify typical failure modes. Based on new data, the prediction models in steps S3 and S5 are updated periodically using incremental learning methods. An intelligent decision support system is developed. The system takes project information, material information, environmental conditions and performance requirements as input, calls the knowledge base and prediction model to recommend adhesive formulations, predicts repair effects and service life, and generates a decision report that includes recommended solutions, construction suggestions, risk warnings and cost estimates.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. Chemical composition, molecular structure, and surface properties of old pavement materials were acquired using multi-dimensional characterization techniques such as spectroscopy, electron microscopy, and nuclear magnetic resonance (NMR), establishing a feature database including parameters such as carbonyl index and softening point increment. Machine learning algorithms were used to analyze matching patterns in historical data to predict the chemical compatibility and physical matching degree between the new binder and the old material. Subsequently, atomic force microscopy and X-ray energy dispersive spectroscopy were used to quantitatively detect the chemical bonding density, adsorption work, and roughness of the interface layer, combined with molecular simulations to verify the interface interaction mechanism. Macroscopic mechanical properties were obtained through pull-out and fatigue tests, and nonlinear correlation models between microscopic parameters and pull-out strength and shear strength were established using regression analysis and machine learning.

[0015] 2. By establishing a multi-parameter correlation model, the macroscopic mechanical properties of repair materials can be rapidly evaluated based on microscopic test data, providing a theoretical basis for adhesive selection and formulation optimization. Simultaneously, the model validation mechanism ensures the reliability of the prediction results, avoiding the subjectivity and inconsistency of traditional experience-based judgments, and significantly shortening the evaluation cycle of repair effectiveness. Attached Figure Description

[0016] Figure 1 This is a diagram illustrating the method steps of this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0018] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of this application are only used to explain the relative positional relationships and movements between components in a specific orientation (as shown in the figures). If the specific orientation changes, the directional indications also change accordingly. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0019] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] Example 1: Please refer to Figure 1 A method for evaluating the repair effect of asphalt pavement based on binders, the specific steps of which are as follows: S1. Fourier transform infrared spectroscopy and scanning electron microscopy were used to characterize the old pavement materials, nuclear magnetic resonance was used to analyze the molecular structure of the binder, and a material feature database was established using machine learning algorithms to predict the compatibility between the binder and the pavement materials. S2. The thickness of the interface layer was measured using atomic force microscopy, the chemical bonding density was quantitatively detected by X-ray photoelectron spectroscopy, the physical adsorption work was calculated based on the contact angle test, the mechanical interlocking degree was evaluated by three-dimensional laser scanning, and the interface interaction was verified by molecular dynamics simulation. S3. Extract microscopic parameters including interfacial bonding density, adsorption work, roughness index and penetration depth, conduct pull-out tests and fatigue tests to obtain macroscopic performance data, and use multivariate nonlinear regression and machine learning algorithms to establish a correlation prediction model between microscopic parameters and macroscopic performance. S4. Install fiber optic grating sensors in the repair test section for in-situ monitoring, design multi-factor coupled tests of temperature cycling, water immersion and dynamic load, take samples regularly to characterize interface degradation features, and establish a life prediction model based on the accelerated aging coefficient. S5. Determine the optimization objective based on the correlation model, use genetic algorithm or particle swarm algorithm to perform multi-objective optimization of the formula, and verify the optimization effect through microscopic characterization and macroscopic performance testing. S6. Use portable testing equipment on site to quickly determine the microscopic parameters of the interface, conduct pull-out tests and non-destructive testing, establish a multi-dimensional evaluation index system that includes microscopic, macroscopic and durability, and use a prediction model to output a comprehensive score. S7. Establish standardized data collection templates, build a cloud-based knowledge base platform for repair effects, optimize prediction models through big data mining and analysis, and develop an intelligent decision support system.

[0021] In this embodiment: In the prior art, the evaluation of asphalt pavement repair effectiveness mainly relies on macroscopic mechanical indicators and human experience, lacking in-depth analysis of the microscopic mechanisms of the interface. Traditional methods cannot quantitatively characterize key parameters such as chemical bond density and physical adsorption work, resulting in a lack of scientific basis for binder selection. Existing evaluation systems suffer from problems such as a disconnect between microscopic and macroscopic performance, long evaluation cycles, and low levels of intelligence, making it difficult to accurately predict the long-term service performance of the repair layer and easily leading to engineering problems such as interface delamination and high rework rates.

[0022] To address the aforementioned issues, the inventors recognized the need to establish a correlation model between microscopic parameters and macroscopic performance, and to develop an intelligent evaluation system. Analysis revealed that interfacial adhesion quality is influenced by a combination of factors, including chemical bonding, physical adsorption, and mechanical interlocking, making it impossible for a single indicator to comprehensively reflect the repair effect. Therefore, they proposed integrating multiple characterization methods to obtain multi-scale data, employing machine learning algorithms to establish a cross-scale correlation model, and combining in-situ monitoring and accelerated testing to construct a durability evaluation system. Ultimately, this resulted in a standardized data platform to support intelligent decision-making.

[0023] Therefore, this application proposes a method for evaluating the repair effect of asphalt pavement based on binders. The specific steps are as follows: Fourier transform infrared spectroscopy and scanning electron microscopy are used to characterize the old pavement material; nuclear magnetic resonance analysis is used to analyze the molecular structure of the binder; a material characteristic database is established using machine learning algorithms to predict the compatibility between the binder and the pavement material; atomic force microscopy is used to measure the interfacial layer thickness; X-ray photoelectron spectroscopy is used to quantitatively detect the chemical bonding density; physical adsorption work is calculated based on contact angle testing; three-dimensional laser scanning is used to evaluate the mechanical interlocking degree; and molecular dynamics simulation is used to verify the interfacial interaction; microscopic parameters of the interface are extracted, and pull-out and fatigue tests are conducted to obtain macroscopic performance data. Multiple nonlinear regression and machine learning algorithms were used to establish a correlation prediction model between microscopic parameters and macroscopic performance. Fiber optic grating sensors were embedded in the repair test section for in-situ monitoring. Multi-factor coupling tests were designed, and interface degradation characteristics were characterized by periodic sampling. A lifetime prediction model was established based on the accelerated aging coefficient. Optimization objectives were determined according to the correlation model, and intelligent algorithms were used for multi-objective optimization of the formulation. The optimization effect was verified through microscopic characterization and macroscopic performance testing. Portable testing equipment was used on-site to quickly measure interface parameters, establish a multi-dimensional evaluation index system, and output a comprehensive score using the prediction model. A cloud-based repair effect knowledge base platform was constructed, and an intelligent decision support system was developed through big data mining to optimize the model.

[0024] Fourier transform infrared spectroscopy and scanning electron microscopy characterization refer to the analysis of the chemical composition and physical structure of old pavement materials by examining the intensity of spectral absorption peaks and their microscopic morphology. For example, using a 1700 cm⁻¹ spectral density... -1 Carbonyl absorption peak intensity is used to assess the degree of asphalt aging, and scanning electron microscopy (SEM) images can quantify aggregate surface roughness. Nuclear magnetic resonance (NMR) analysis is used to determine the types and contents of active functional groups in binder molecules, such as identifying carboxyl or epoxy groups through chemical shifts. Machine learning algorithms, such as random forests or neural networks, can be used to predict the compatibility of different binders with old materials by training a material feature database.

[0025] Atomic force microscopy (AFM) measures interfacial layer thickness by obtaining nanometer-resolution three-dimensional morphology data through probe scanning, for example, to detect the penetration depth of binders on aggregate surfaces. X-ray photoelectron spectroscopy (XPS) calculates chemical bonding density by analyzing changes in the chemical states of interfacial elements, for example, quantifying the number of sulfur bonds formed by changes in the peak area of ​​sulfur. Contact angle testing calculates surface free energy and adhesion work based on droplet morphology, for example, using water contact angle data combined with the Young-Dupré equation to calculate physical adsorption strength. Three-dimensional laser scanning extracts roughness parameters by acquiring three-dimensional surface coordinate data, for example, assessing mechanical interlocking effects through Ra values. Molecular dynamics simulations use the COMPASS force field to construct molecular models, for example, verifying the reliability of experimental data by calculating binding energy.

[0026] This method first acquires data on the chemical composition, molecular structure, and surface properties of old pavement materials using multi-dimensional characterization techniques such as spectroscopy, electron microscopy, and nuclear magnetic resonance, establishing a feature database including parameters such as carbonyl index and softening point increment. Machine learning algorithms analyze matching patterns in historical data to predict the chemical compatibility and physical matching degree between the new binder and the old material. Subsequently, atomic force microscopy and X-ray energy dispersive spectroscopy are used to quantitatively detect the chemical bonding density, adsorption work, and roughness of the interface layer, combined with molecular simulation to verify the interface interaction mechanism. Macroscopic mechanical properties are obtained through pull-out and fatigue tests, and nonlinear correlation models between microscopic parameters and pull-out strength and shear strength are established using regression analysis and machine learning.

[0027] Fiber optic sensors were deployed in the repair section to monitor parameters such as temperature and strain in real time. Accelerated aging tests were designed to simulate complex environments such as temperature cycling and water erosion. Interface degradation indicators were detected through periodic sampling, and a life prediction equation was established. Based on the model, the optimization objective was derived in reverse, and a genetic algorithm was used to adjust the proportion of adhesive formulation components. Samples were prepared to verify the optimization effect. On-site testing used portable spectrometers, roughness testers, and other equipment to quickly acquire microscopic parameters. Combined with pull-out test and non-destructive testing data, a multi-dimensional evaluation system was constructed, and a comprehensive score was output through the prediction model. Finally, data from multiple projects were integrated through a cloud computing platform, and data mining was used to optimize model parameters to form an intelligent decision-making system.

[0028] Compared to existing technologies, traditional methods evaluate bonding effectiveness solely based on pull-out strength, while this approach integrates microscopic parameters such as chemical bond density, adsorption work, and roughness, establishing a cross-scale correlation model. Existing technologies rely on long-term field observation to assess durability, while this approach significantly shortens the evaluation cycle through multi-factor coupled accelerated testing and a life prediction model. Traditional evaluation systems lack data accumulation and sharing mechanisms, while the cloud platform built in this approach enables multi-project data integration and intelligent analysis, supporting adhesive selection and construction scheme optimization.

[0029] Through the above technical solutions, this application achieves a quantitative correlation between the microscopic mechanism and macroscopic performance of the interface, improving the accuracy of adhesive compatibility prediction; by combining in-situ monitoring and accelerated testing, the evaluation cycle of repair effect is significantly shortened; the constructed multidimensional evaluation index system can comprehensively reflect the interface bonding quality, mechanical properties and durability; the cloud computing-based knowledge base platform supports data sharing and continuous model optimization, providing a reliable basis for scientific decision-making.

[0030] Example 2: Please refer to Figure 1 The specific method for step S1 is as follows: S1.1 The wavenumber of old asphalt at 1700 cm⁻¹ was determined using a Fourier transform infrared spectrometer. -1The carbonyl index was calculated by measuring the absorption peak intensity, the softening point temperature was determined by differential scanning calorimetry to assess the degree of aging, the surface morphology of the aggregate was observed by scanning electron microscopy to obtain the roughness value, the mineral composition was analyzed by X-ray diffraction, the chemical shift and content of the active functional groups of the binder were determined by nuclear magnetic resonance spectroscopy, and the number-average molecular weight and polydispersity index were determined by gel permeation chromatography. A multidimensional characteristic parameter set including carbonyl index, softening point increment, surface roughness, mineral composition, functional group content and molecular weight parameters was established. S1.2 Construct a material feature database containing characteristic data and corresponding bonding performance indicators of pavement materials with different aging degrees, aggregates of different mineral types, and different types of binders. Extract carbonyl index, softening point increment, surface free energy, roughness, functional group concentration, and molecular weight as input parameters. Use random forest algorithm or neural network algorithm to establish a prediction model. Use chemical compatibility index and physical matching degree as output variables. Optimize the model through k-fold cross-validation. Input the parameters of the material to be evaluated and output the matching score value. When the score value is greater than or equal to the preset threshold, the matching is judged to be good.

[0031] In this embodiment: the specific method of step S1 is further proposed as follows: the wavenumber of old asphalt at 1700 cm⁻¹ is determined using a Fourier transform infrared spectrometer. -1 The carbonyl index was calculated using the absorption peak intensity. The softening point temperature was determined using differential scanning calorimetry (DSC) to assess the degree of aging. The surface morphology of the aggregate was observed using scanning electron microscopy (SEM) to obtain roughness values. The mineral composition was analyzed using X-ray diffraction (XRD). The chemical shifts and contents of the active functional groups in the binder were determined using nuclear magnetic resonance (NMR) spectroscopy. The number-average molecular weight and polydispersity index were determined using gel permeation chromatography (GPC). A multidimensional feature parameter set was established, including carbonyl index, softening point increment, surface roughness, mineral composition, functional group content, and molecular weight. A material feature database was constructed, containing characteristic data and corresponding bonding performance indicators for pavement materials with different aging degrees, different aggregate types, and different types of binders. Carbonyl index, softening point increment, surface free energy, roughness, functional group concentration, and molecular weight were extracted as input parameters. A prediction model was established using a random forest algorithm or a neural network algorithm, with chemical compatibility index and physical matching degree as output variables. The model was optimized using k-fold cross-validation. Inputting the parameters of the material to be evaluated, a matching score was output. A good matching was determined when the score was greater than or equal to a preset threshold.

[0032] The carbonyl index refers to the carbonyl index obtained by Fourier transform infrared spectroscopy at 1700 cm⁻¹. -1The oxidation degree index, calculated using characteristic peak intensity, can be achieved by scanning the asphalt sample with an infrared spectrometer and calculating the peak area ratio. This index reflects the amount of oxygen-containing functional groups generated during the asphalt aging process. The softening point increment refers to the temperature difference between the softening points of new and old asphalt determined by differential scanning calorimetry (DSC). Specifically, this can be achieved by measuring the phase transition temperature using a DSC at a constant heating rate. This parameter is used to quantify changes in the thermal stability of asphalt materials. The surface roughness value refers to the aggregate surface morphology parameter calculated through three-dimensional reconstruction of scanning electron microscopy images. Specifically, this can be achieved by extracting the arithmetic mean deviation of the surface profile using image processing software. This parameter affects the mechanical interlocking effect between the binder and the aggregate. The mineral composition refers to the proportion of aggregate mineral components obtained through X-ray diffraction pattern analysis. Specifically, this can be quantitatively calculated by comparing the diffraction peak intensities with standard cards. This parameter determines the chemical compatibility between the aggregate and the binder. Functional group content refers to the concentration of active groups calculated from the chemical shift peak area in nuclear magnetic resonance spectroscopy. Specifically, it can be calculated using the ratio of the integrated area of ​​characteristic peaks to the total peak area in proton or carbon spectroscopy. This parameter characterizes the chemical reaction potential between the binder and the old asphalt. Number-average molecular weight refers to the average length of polymer chains determined by gel permeation chromatography. Specifically, it can be calculated using a molecular weight-retention time calibration curve established with standard samples. This parameter affects the flowability and permeability of the binder. A multidimensional feature parameter set refers to a quantitative dataset integrating the physical, chemical, and morphological characteristics of the material. Specifically, it can be converted from different dimensional parameters to dimensionless values ​​through standardization. This dataset provides multidimensional input for machine learning models. k-fold cross-validation is a model optimization method that divides the dataset into k subsets for iterative training and validation. Specifically, it can be done by randomly partitioning the dataset and calculating the average prediction error. This method is used to prevent overfitting and improve generalization ability.

[0033] First, the old asphalt and aggregates were systematically characterized using various analytical instruments, such as infrared spectroscopy to analyze oxidation levels, differential scanning calorimetry to determine thermal stability, electron microscopy to observe surface morphology, and X-ray diffraction to determine mineral composition. Simultaneously, nuclear magnetic resonance (NMR) and gel permeation chromatography (GPC) techniques were used to accurately determine the chemical composition and molecular structure characteristics of the binder. The parameters obtained from these tests were standardized to form a multidimensional dataset containing material aging status, surface properties, chemical composition, and molecular structure. Subsequently, a feature database containing historical engineering data was constructed, and a machine learning algorithm was used to establish a mapping relationship between input parameters and bonding performance. For example, carbonyl index and softening point increment were used as input variables, and laboratory-measured bond strength was used as the output variable for model training. After optimizing the model parameters through cross-validation, the matching score can be obtained by inputting the test data of the material to be evaluated, providing a quantitative basis for binder selection.

[0034] Compared to existing technologies, traditional methods typically only detect a single physical index or chemical component, such as softening point or molecular weight, failing to comprehensively reflect the material interface interaction mechanism. Existing technologies lack synergistic analysis of material aging state, surface morphology, and chemical composition, leading to biases in adhesive selection. Existing databases often rely on manual experience to establish association rules, making it difficult to handle complex nonlinear relationships of multidimensional parameters. This solution integrates multi-source test data to construct a multidimensional feature space and utilizes machine learning algorithms to uncover hidden association patterns, significantly improving the accuracy and reliability of adhesive matching prediction.

[0035] Through the above technical solutions, this application achieves a systematic characterization of the properties of old pavement materials and adhesives, and establishes a scientific and quantitative matching evaluation system. The synergistic analysis of multi-dimensional feature parameters can accurately identify key factors affecting interfacial adhesion, and the establishment of machine learning models effectively solves the subjectivity and limitations of traditional empirical judgments. Cross-validation mechanisms ensure the stability of model prediction results, and the setting of scoring thresholds provides a clear decision-making basis for engineering applications, thereby reducing the risk of interfacial failure due to material mismatch.

[0036] Example 3: Please refer to Figure 1 The specific method for step S2 is as follows: S2.1. The interface between the binder, old asphalt, and aggregate was scanned using an atomic force microscope to obtain the thickness distribution of the interface layer. The interface elements and chemical bonding type were analyzed using X-ray photoelectron spectroscopy. The interface chemical bonding density was calculated by peak fitting. The contact angle of the droplets on the old asphalt and aggregate surfaces was measured using a contact angle meter. The adhesion work was calculated based on the Young-Dupré equation, and the interface free energy was calculated using the Owens-Wendt method. The three-dimensional morphology of the aggregate surface was measured using a three-dimensional laser scanner. The surface roughness parameters Ra, Rz, and surface area ratio were extracted, and the correlation between roughness and mechanical interlocking strength was established. S2.2 Based on the experimental data in S2.1, molecular simulation software was used to construct models of binder molecules, aged asphalt, and aggregate surfaces. Periodic boundary conditions were set for molecular dynamics simulation. The COMPASS force field was used for energy minimization and equilibrium simulation. The interfacial binding energy and interaction energy were calculated, the interfacial stress distribution was analyzed, the contributions of hydrogen bonds and van der Waals forces were identified, and the interfacial layer thickness, number of bonding sites, and adsorption energy were calculated. The reliability of the experimental results was confirmed when the relative error between the simulation results and the experimental data was less than 15%.

[0037] In this embodiment, the specific method of step S2 is further proposed as follows: The interface between the binder, old asphalt, and aggregate is scanned using an atomic force microscope to obtain the interface layer thickness distribution; the interface elements and chemical bonding type are analyzed using X-ray photoelectron spectroscopy; the interface chemical bonding density is calculated through peak fitting; the contact angle of the droplet on the old asphalt and aggregate surfaces is measured using a contact angle meter; the adhesion work is calculated based on the Young-Dupré equation; the interface free energy is calculated using the Owens-Wendt method; the three-dimensional morphology of the aggregate surface is measured using a three-dimensional laser scanner; the surface roughness parameters Ra, Rz, and surface area ratio are extracted; and the correlation between roughness and mechanical interlocking strength is established. Based on the experimental data, molecular simulation software is used to construct models of the binder molecules, aged asphalt, and aggregate surface; periodic boundary conditions are set for molecular dynamics simulation; energy minimization and equilibrium simulation are performed using the COMPASS force field; the interface binding energy and interaction energy are calculated; the interface stress distribution is analyzed; the contributions of hydrogen bonds and van der Waals forces are identified; the interface layer thickness, the number of bonding sites, and the adsorption energy are calculated; and the reliability of the experimental results is confirmed when the relative error between the simulation results and the experimental data is less than 15%.

[0038] Atomic force microscopy (AFM) scanning of interface layer thickness distribution involves obtaining nanoscale morphological information by measuring changes in interatomic forces between the probe and the sample surface. Specifically, contact or tapping modes can be used to scan the interface transition zone, and the average interface layer thickness is calculated from the height distribution curve. X-ray photoelectron spectroscopy (XPS) analysis of chemical bonding types involves determining the chemical state of elements by measuring photoelectron kinetic energy. Specifically, monochromatic Al Kα rays can be used to excite the sample surface, and the chemical bonding density can be calculated by fitting the characteristic peak areas of C1s and O1s. The Young-Dupré equation calculation of adhesion work involves calculating the solid-liquid interface adhesion energy based on the solid-liquid contact angle. Specifically, droplet shape analysis can be used to determine the contact angle, and the physical adsorption strength can be calculated by combining surface tension parameters. Three-dimensional laser scanning of surface roughness involves obtaining three-dimensional morphological data through laser triangulation. Specifically, a blue laser source can be used to scan the aggregate surface, and Ra and Rz parameters can be extracted through point cloud data processing. Molecular dynamics simulation verification of interfacial interactions involves studying the molecular-scale interaction mechanism through numerical simulation. Specifically, Materials Studio software can be used to construct a pitch molecular model, and the interfacial binding energy can be calculated through equilibrium state simulation.

[0039] In the interface layer analysis, the nanoscale thickness distribution of the binder-old asphalt interface was first obtained using atomic force microscopy. Combined with X-ray photoelectron spectroscopy to identify the chemical bonding type, the interfacial chemical bonding strength was quantitatively characterized. Contact angle test data were converted into an adhesion work index using the Young-Dupré equation, reflecting the intensity of physical adsorption. Three-dimensional laser scanning data, through correlation analysis between surface roughness parameters and mechanical interlocking strength, revealed the microscopic formation mechanism of macroscopic mechanical properties. The asphalt-aggregate interface model constructed using molecular dynamics simulations, through force field parameter settings and equilibrium calculations, verified the consistency between the experimentally measured interfacial bonding energy and the theoretical prediction. When the relative error was controlled within a threshold range, the reliability of the experimental data was confirmed.

[0040] Compared with existing technologies, traditional methods only evaluate interfacial performance through a single mechanical test, failing to distinguish the contributions of chemical bonding and physical adsorption, and lacking verification of the molecular-level mechanism of action. This scheme, through the synergistic application of multi-scale characterization techniques, achieves for the first time the simultaneous quantitative analysis of interfacial chemical bonding density, physical adsorption work, and mechanical interlocking degree, and combines molecular simulation to verify the reliability of experimental data, forming a complete analytical chain from the nanoscale to the macroscale.

[0041] Through the above technical solutions, this application effectively solves the problem of biased adhesive performance evaluation caused by the unclear mechanism of microscopic interface interaction. The cross-validation mechanism of experiments and simulations significantly improves the accuracy of interfacial interaction analysis, providing a reliable data foundation for the subsequent establishment of a micro-macro correlation model. The multi-parameter synergistic analysis method can distinguish the contribution of different mechanisms to interfacial strength, providing precise directions for adhesive formulation optimization.

[0042] Example 4: Please refer to Figure 1 The specific method for step S3 is as follows: S3.1 Extract the interfacial chemical bonding density, physical adsorption work, surface roughness index and penetration depth from step S2, prepare a composite specimen containing an old asphalt base layer, an adhesive interface layer and a repair material layer, conduct a pull-out test to determine the pull-out strength, conduct a shear test to determine the shear strength, conduct a dynamic fatigue test to record the number of cycles of failure, repeat the test for no less than 5 specimens in each group, calculate the average value and standard deviation of each performance index, and establish a sample dataset containing 4 micro parameters as input variables and 3 macro performance indexes as output variables; S3.2. Key micro-parameters were screened using grey relational analysis, and a prediction model Y=f(X1,X2,X3,X4) was established using multivariate nonlinear regression. The regression coefficients were fitted using the least squares method. When the coefficient of determination R... 2The model is considered effective when it is greater than or equal to 0.85. The model is then optimized using support vector machine, random forest, or neural network algorithms. The training and test sets are divided into a 7:3 or 8:2 ratio. The hyperparameters are optimized using k-fold cross-validation. The mean absolute percentage error (MAPE) is calculated. When the MAPE is less than or equal to 10%, the final association prediction model is established.

[0043] In this embodiment: interfacial chemical bond density, physical adsorption work, surface roughness index, and penetration depth are extracted from the interface layer as microscopic parameters. Composite specimens containing an old asphalt base layer, an adhesive interface layer, and a repair material layer are prepared. Pull-out tests are conducted to determine pull-out strength, shear tests are conducted to determine shear strength, and dynamic fatigue tests are conducted to record the number of cycles leading to failure. Each group is tested repeatedly with no fewer than 5 specimens. The average value and standard deviation of each performance index are calculated, and a sample dataset containing 4 microscopic parameters as input variables and 3 macroscopic performance indicators as output variables is established. Grey relational analysis is used to screen key microscopic parameters, and multiple nonlinear regression is used to establish a prediction model. The regression coefficients are fitted using the least squares method. When the coefficient of determination R² is greater than or equal to 0.85, the model is confirmed to be effective. Support vector machine, random forest, or neural network algorithms are used to optimize the model. The training set and test set are divided proportionally, and k-fold cross-validation is used to optimize the hyperparameters. The mean absolute percentage error is calculated, and when the error is less than or equal to 10%, the final correlation prediction model is established.

[0044] Interfacial chemical bond density refers to the number of chemical bonds per unit area, which can be calculated using X-ray photoelectron spectroscopy peak fitting to characterize the tightness of interfacial chemical bonding. Physical adsorption work refers to the adhesion work generated when a liquid comes into contact with a solid, which can be calculated using contact angle testing combined with the Young-Dupré equation to reflect the intensity of physical adsorption. Surface roughness index is a quantitative parameter of the three-dimensional morphology of the aggregate surface, which can be obtained by extracting parameters such as Ra and Rz using three-dimensional laser scanning to evaluate the effectiveness of mechanical interlocking. Penetration depth refers to the depth to which the binder penetrates the old asphalt base layer, which can be measured by microscopic observation or laser scanning to characterize the physical penetration effect of interfacial bonding. Grey relational analysis is a method of screening key parameters by calculating the correlation between sequences, which can be achieved by data standardization and correlation coefficient calculation to identify microscopic parameters that significantly affect macroscopic performance. A multivariate nonlinear regression model establishes a nonlinear relationship model between multiple independent and dependent variables, which can be implemented using polynomial or exponential functions to describe the complex relationship between microscopic parameters and macroscopic performance.

[0045] By preparing composite specimens and conducting tensile, shear, and fatigue tests, macroscopic mechanical property data of interfacial bonding can be systematically obtained. After establishing the datasets of microscopic parameters and macroscopic properties, grey relational analysis is first used to screen out the key parameters that have the greatest impact on tensile strength, shear strength, and fatigue life, eliminating interference from redundant variables. Subsequently, a multivariate nonlinear regression model is established, and the quantitative relationship between microscopic parameters and macroscopic properties is obtained by fitting the model using the least squares method. The effectiveness of the model is confirmed when the coefficient of determination reaches a preset threshold. To further improve the prediction accuracy, machine learning algorithms are used to optimize the regression model. The model's generalization ability is verified by dividing the training and test sets, and hyperparameters are adjusted using cross-validation methods. Finally, a correlation prediction model with controllable error is established.

[0046] Compared to existing technologies, current methods typically rely solely on single regression models or empirical formulas to establish micro- and macro-level correlations, neglecting parameter selection and model optimization, resulting in insufficient prediction accuracy. This approach uses grey relational analysis to select key parameters and combines multiple nonlinear regression with machine learning algorithms to construct a composite model, which can more accurately capture the nonlinear impact of micro-level parameters on macro-level performance. Furthermore, existing technologies lack quantitative validation standards for model effectiveness; this approach ensures the model's prediction results have engineering application value by setting a coefficient of determination and an error threshold.

[0047] Through the above technical solution, this application solves the technical problem of the disconnect between microscopic parameters and macroscopic performance, and realizes the quantitative prediction of interfacial bonding performance. By establishing a multi-parameter correlation model, the macroscopic mechanical properties of the repair material can be quickly evaluated based on microscopic test data, providing a theoretical basis for adhesive selection and formulation optimization. At the same time, the model verification mechanism ensures the reliability of the prediction results, avoids the subjectivity and inconsistency of traditional experience-based judgments, and significantly shortens the evaluation cycle of repair effect.

[0048] Example 5: Please refer to Figure 1 The specific method for step S4 is as follows: S4.1 Fiber Bragg grating strain sensors and temperature sensors are embedded in the adhesive interface layer of the repair test section at intervals of 2m to 5m. The strain measurement range is -3000με to +3000με, and the temperature measurement range is -40℃ to +80℃. Soil moisture sensors and triaxial accelerometers are also installed. Data acquisition frequency is 1 hour to 24 hours. A temperature cycling test is designed, with a high-temperature stage of 60℃ to 80℃ for 4 to 8 hours and a low-temperature stage of -20℃ to -10℃ for 4 to 8 hours, for a total of at least 30 cycles. A water immersion test is designed, with immersion for 24 to 72 hours followed by freeze-thaw cycles. The freezing temperature is -18℃ to -15℃ for 12 hours, and the thawing temperature is 20℃ to 25℃ for 12 hours, for a total of at least 15 cycles. A dynamic load test is designed, with a load amplitude of 0.5MPa to 0.9MPa and a frequency of 10Hz to 20Hz, for a total of 10 load cycles. 4 Next to 10 6 Next, the above experiments were combined into a multi-factor coupled accelerated experimental scheme according to the time series. S4.2. At least three specimens shall be sampled at the 0%, 25%, 50%, 75%, and 100% time points of the accelerated test, as well as at failure. The interfacial chemical bonding density, adsorption work, roughness, and penetration depth shall be determined using the method in step S2. The attenuation rate shall be calculated. Microcracks shall be observed using scanning electron microscopy to calculate the crack density. The porosity growth rate shall be determined using mercury intrusion porosimetry. Pull-out and shear tests shall be performed to calculate the strength retention rate. An accelerated aging model and a fitted apparent activation energy Ea shall be established based on the Arrhenius equation. The degradation kinetic model is as follows: t_real=t_test×exp[Ea / R×(1 / T_real-1 / T_test)]; P(t) = P0 × exp(-k × t^n); Where t_real is the time under actual service conditions, t_test is the time under accelerated test conditions, Ea is the apparent activation energy, R is the molar gas constant, T_real is the absolute temperature of the actual service environment, T_test is the absolute temperature of the accelerated test, P(t) is the performance parameter value at time t, P0 is the performance parameter value at the initial time, k is the degradation rate constant, and n is the degradation index. Failure is determined when the bond density decay rate is ≥50%, the strength retention rate is ≤60%, the crack density is ≥20mm / cm², or the porosity growth rate is ≥30%. The failure time t_failure is calculated, and the actual service life is calculated in combination with the acceleration factor. The specific method is as follows: L_service=t_failure×exp[Ea / R×(1 / T_real-1 / T_test)]; Where L_service is the actual service life and t_failure is the failure time under accelerated testing conditions.

[0049] In this embodiment: This application further proposes embedding fiber Bragg grating strain sensors and temperature sensors in the adhesive interface layer of the repair test section, with a embedding spacing of 2m to 5m. The strain measurement range is -3000με to +3000με, and the temperature measurement range is -40℃ to 80℃. Soil moisture sensors and triaxial accelerometers are also deployed. The data acquisition frequency is 1 hour to 24 hours. A temperature cycling test is designed, with a high temperature stage of 60℃ to 80℃ for 4 to 8 hours and a low temperature stage of -20℃ to -10℃ for 4 to 8 hours, for a total of no less than 30 cycles. A water immersion test is designed, with immersion for 24 to 72 hours followed by freeze-thaw cycles. The freezing temperature is -18℃ to -15℃ for 12 hours, and the thawing temperature is 20℃ to 25℃ for 12 hours, for a total of no less than 15 cycles. A dynamic load test is designed, with a load amplitude of 0.5MPa to 0.9MPa and a frequency of 10Hz to 20Hz, for a total of 10 loads. 4 Next to 10 6 Next, the above experiments were combined into a multi-factor coupled accelerated testing scheme according to the time series. At least three specimens were sampled at the 0%, 25%, 50%, 75%, and 100% time points of the accelerated test, as well as at failure. The interfacial chemical bond density, adsorption work, roughness, and penetration depth were measured, the attenuation rate was calculated, microcracks were observed and crack density was calculated, the porosity growth rate was measured, and pull-out and shear tests were conducted to calculate the strength retention rate. An accelerated aging model was established based on the Arrhenius equation, and the apparent activation energy was fitted. A degradation kinetic model was established. When the bond density attenuation rate was ≥50%, the strength retention rate was ≤60%, or the crack density was ≥20 mm / cm², the degradation kinetic model was established. 2 Failure is determined when the porosity growth rate is ≥30%, and the failure time is calculated and the actual service life is calculated in combination with the acceleration rate.

[0050] Fiber Bragg grating strain sensors are optical sensors that measure strain by utilizing changes in grating wavelength. They can be implemented using fiber Bragg grating encapsulation technology and are used to monitor the strain response of the interface layer in real time under different environments. Temperature sensors, such as thermocouples or thermistors, are used to synchronously record the impact of temperature changes on interface performance. Multi-factor coupled accelerated testing schemes combine temperature cycling, water immersion freeze-thaw cycles, and dynamic loading, and can be implemented through test timing sequences to simulate the synergistic effects of multiple factors in actual service. Interface degradation characteristic parameters include chemical bond density decay rate, crack density, and porosity growth rate, which can be determined using scanning electron microscopy image analysis and mercury intrusion porosimetry to quantify the degree of interface damage. The accelerated aging model is a temperature-time equivalent model based on the Arrhenius equation, implemented through activation energy calculations, and used to extrapolate accelerated test results to actual service conditions.

[0051] By deploying a distributed sensor network at the repair interface, multi-dimensional data such as temperature, strain, and humidity can be collected in real time. The designed multi-factor coupled test simulates the synergistic destructive effects of complex environments on the interface layer by alternately applying high and low temperatures, water immersion, freeze-thaw cycles, and dynamic loads. After sampling at key test points, microscopic characterization techniques are used to quantitatively analyze the changes in parameters such as interfacial chemical bonding density and adsorption work. Combined with macroscopic mechanical property test results, an accelerated aging model based on activation energy is established. This model calculates the difference in reaction rates under different temperature conditions, converting the accelerated test time into actual service time, and thus predicting the service life of the repair layer. When the microscopic parameters reach the preset failure threshold, the system automatically triggers a lifespan termination judgment and outputs the predicted results.

[0052] Compared to existing technologies, traditional methods typically employ single-factor accelerated testing and lack in-situ monitoring capabilities, failing to accurately reflect the interface degradation process under the coupled effects of multiple factors. Existing lifetime prediction models are mostly based on empirical formulas, neglecting the nonlinear influence of temperature on material aging rates. This proposed solution acquires real-time data through a multi-sensor in-situ monitoring network, combined with multiphysics coupling experimental design, enabling more accurate simulation of actual service conditions. The established degradation kinetics model incorporates activation energy parameters, effectively addressing the impact of temperature differences on the acceleration rate and significantly improving the accuracy of lifetime prediction.

[0053] Through the above technical solutions, this application achieves real-time monitoring and multi-dimensional evaluation of the performance of the repair interface under complex environments, solving the problems of long test cycles and high distortion of environmental simulation in traditional evaluation methods. The established accelerated aging model can accurately correlate laboratory accelerated test results with actual service life, providing a scientific basis for maintenance decisions. By setting multi-dimensional failure criteria, the risk of interface peeling can be warned in advance, avoiding repair layer failure due to detection lag.

[0054] Example 6: Please refer to Figure 1 The specific method for step S5 is as follows: S5.1. Based on the correlation prediction model established in step S3, determine the target values ​​of pull-out strength, shear strength, and fatigue life. Solve the target values ​​of micro-parameters in reverse. Determine the design variables of the binder formulation, including the content of base asphalt, the amount of modifier, the amount of emulsifier, and the amount of filler. Set the value range and constraints of each component. Establish a multi-objective optimization function and solve it using a genetic algorithm or a particle swarm optimization algorithm. For the genetic algorithm, set the population size to 50 to 200, the crossover probability to 0.6 to 0.9, and the mutation probability to 0.01 to 0.1. For the particle swarm optimization algorithm, set the number of particles to 30 to 100 and the learning factor to 1.5 to 2.5. Obtain the Pareto optimal solution set and select the optimal formulation using the ideal point method or the weighted method. S5.2 Prepare binder samples according to the optimal formula. Heat the base asphalt to 160℃ to 180℃, and shear it at high speed at 3000rpm to 5000rpm for 30 to 60 minutes. The emulsification temperature is 120℃ to 140℃, and the emulsification speed is 8000rpm to 12000rpm. Prepare no less than 5 sets of parallel samples. Use the method in step S2 to determine the optimized micro parameters, and use the method in step S3.1 to determine the optimized macro properties. Calculate the improvement rate, input the measured micro parameters into the correlation model to calculate the predicted value, and compare it with the measured value. When the pull-out strength improvement rate is ≥20%, the fatigue life improvement rate is ≥30%, and the cost increase rate is ≤15%, the optimization is deemed effective.

[0055] In this embodiment: the portable Fourier transform infrared spectrometer refers to a spectral analysis device that can quickly detect the chemical characteristics of materials on site. Specifically, it can be implemented by a portable instrument equipped with an attenuated total reflectance accessory. The carbonyl index is calculated by scanning the absorption peak intensity within a specific wavenumber range, which is used to evaluate the chemical compatibility between the binder and the old asphalt.

[0056] Penetration depth measurement refers to observing the penetration distance of the binder in old asphalt using a microscope. Specifically, it can be achieved using the scale measurement function of a digital microscope, and is used to characterize the degree of interfacial bonding between the binder and the old material.

[0057] The three-level evaluation index system refers to an evaluation system that integrates microscopic interface parameters, macroscopic mechanical properties, and durability indicators. Specifically, the weight coefficients of each level of indicators can be determined by the analytic hierarchy process, and the dimensional differences can be eliminated through normalization to achieve a comprehensive evaluation of multi-dimensional data.

[0058] Machine learning prediction models refer to intelligent scoring models trained on historical data. Specifically, they can be implemented using random forest or neural network algorithms. By inputting micro-parameters and macro-performance data from on-site detection, they output a comprehensive score and grade classification of the restoration effect.

[0059] At the repaired asphalt pavement site, portable testing equipment is used to quickly acquire microscopic parameters such as interfacial chemical properties, penetration depth, and roughness. These are combined with macroscopic performance data such as pull-out strength and flexural recovery rate, as well as durability indicators such as permeability coefficient and friction coefficient, to form a multi-dimensional testing dataset. Using pre-established correlation models and life prediction models, the on-site testing data is input to calculate the predicted service life. Based on a three-level evaluation index system, each index is standardized and weighted, ultimately outputting a comprehensive score and performance level, generating a complete evaluation report containing both quantitative data and qualitative assessments.

[0060] Compared to existing technologies, traditional methods rely on laboratory testing and evaluation based on single mechanical indicators, failing to rapidly acquire multi-dimensional data on-site and lacking a standardized scoring system. This solution integrates portable testing equipment with an intelligent prediction model, achieving a combination of rapid on-site testing and comprehensive evaluation, thus solving the problems of low efficiency and limited evaluation dimensions inherent in traditional methods.

[0061] Through the above technical solutions, this application realizes rapid on-site evaluation of the asphalt pavement repair effect. By using a multi-dimensional index system and intelligent scoring model, it improves the objectivity and accuracy of the evaluation results, and provides a scientific basis for the quality control of the repair project.

[0062] Example 7: Please refer to Figure 1 The specific method for step S6 is as follows: S6.1. Use a portable Fourier transform infrared spectrometer to scan the repaired section of the road surface after repair, with a wavenumber range of 400 cm⁻¹. -1 Up to 4000cm -1 The carbonyl index was determined, and the interfacial adhesion state was observed using a portable digital microscope at magnifications of 50x to 200x. The penetration depth of the adhesive was measured, and the arithmetic mean roughness Ra value was determined using a portable surface roughness meter, calculated per 100 μm. 2 Pull-out tests were conducted at no fewer than three test points. A portable pull-out tester was used to determine the pull-out strength at a loading rate of 0.5 MPa / s to 1.0 MPa / s. A deflectometer was used to determine the rebound deflection value and calculate the deflection recovery rate. Ground penetrating radar was used to detect interlayer continuity at an antenna frequency of 1 GHz to 2 GHz. An infrared thermal imager was used to detect the temperature distribution. A permeability coefficient meter was used to test the water tightness. A pendulum friction coefficient meter was used to determine the pendulum value BPN. The microscopic parameters were input into the correlation prediction model in step S3 and the life prediction model in step S4 to calculate the predicted service life. S6.2 Establish a three-level evaluation index system encompassing microscopic, macroscopic, and durability levels. Set the qualification standards as follows: carbonyl index <0.8, penetration depth ≥1.0mm, roughness Ra ≥0.5mm, tensile strength ≥1.5MPa, deflection recovery rate ≥80%, defect rate <5%, water permeability coefficient <120ml / min, pendulum value BPN ≥45, and predicted lifespan ≥5 years. Use the analytic hierarchy process (AHP) or entropy weighting method to determine the weight coefficients for each level of index. Normalize each index to convert it into a score value from 0 to 100. Calculate the comprehensive score using the weighted summation method, or use the machine learning prediction model from step S3 to output the comprehensive score and performance level. Based on the comprehensive score, classify the levels as follows: 90-100 points (Excellent), 80-90 points (Good), 60-80 points (Medium), and below 60 points (Poor). Generate an evaluation report containing test data, scoring results, level determination, and improvement suggestions.

[0063] In this embodiment: The standardized data collection template refers to an electronic form with predefined data field structures and formats, specifically implemented using XML or JSON formats. By setting required fields and data type constraints, the integrity and standardization of data collection are ensured. The hybrid storage architecture refers to a storage solution integrating multiple database technologies. Specifically, it can use MySQL to store structured data, MinIO to store unstructured files, and InfluxDB to store time-series data, achieving efficient access to different types of data. The incremental learning method refers to a training approach that dynamically updates model parameters based on newly added data. Specifically, it can be implemented using an online gradient descent algorithm, enabling the prediction model to continuously adapt to newly generated engineering data. Association rule mining refers to discovering frequently occurring associations in a dataset. Specifically, it can use the FP-Growth algorithm to quickly generate frequent itemsets and identify potential association patterns between different material parameters and repair effects.

[0064] By constructing a cloud computing knowledge base platform to integrate scattered engineering data, a microservice architecture is adopted to achieve modular functional deployment. Hybrid storage technology is used to classify and manage structured data, inspection reports, and real-time monitoring data. The data entry interface intercepts outliers by setting data validation rules, and the review process ensures data quality. The multi-dimensional query function supports searching for cases by combination of conditions such as material type and climate conditions, and the GIS map visualization function displays the geospatial distribution of repair projects. Association rule mining extracts the correlation between adhesive formulations and performance indicators from historical data, cluster analysis identifies case groups with similar failure characteristics, and the incremental learning mechanism enables the predictive model to continuously absorb new data and optimize parameters. The intelligent decision-making system matches successful cases in the knowledge base with the input project requirements and generates customized repair solutions by combining the calculation results of the predictive model.

[0065] Compared to existing technologies, which rely on scattered paper records and local storage, resulting in chaotic data formats and difficulties in sharing, this solution achieves standardized data storage and cross-project sharing through standardized templates and a cloud platform. Traditional technologies lack intelligent analysis tools, and decision-making relies on human experience, while this solution achieves automatic knowledge extraction and continuous model optimization through data mining and machine learning. Conventional methods cannot dynamically update evaluation models, while this solution uses an incremental learning mechanism to keep the model adaptable to new technologies and materials. Ordinary systems lack comprehensive decision support functions, while this solution integrates a knowledge base and predictive models to achieve intelligent recommendations for repair solutions.

[0066] Through the above technical solutions, this application solves the problem of insufficient decision-making basis caused by isolated and scattered data in traditional methods, and realizes standardized management and multi-dimensional analysis of repair project data; overcomes the subjectivity and limitations of human experience judgment, and discovers potential patterns through data mining to support scientific decision-making; eliminates the problem of model updates lagging behind technological development, and maintains prediction accuracy through incremental learning; fills the gap of lack of intelligent auxiliary tools in repair scheme design, and generates optimized schemes through systematic analysis, reducing the cost of trial and error in engineering.

[0067] Example 8: Please refer to Figure 1 The specific method for step S7 is as follows: S7.1 Establish a standardized data collection template, which includes fields for basic project information, disease characteristics, material parameters, construction technology, microscopic characterization data, macroscopic performance data, environmental monitoring data, and usage effect data. Define the data type, value range, and required field identifiers for each field. Construct a cloud-based knowledge base platform using a B / S architecture, a microservice architecture for the backend, and a hybrid storage architecture for the database, including relational databases, object storage, and time-series databases. Set up a multi-level user permission management mechanism, use TLS encryption protocol and AES algorithm to ensure data security, establish a daily automatic backup mechanism, develop a data entry interface that supports manual input and batch import, set up data verification rules and review processes, and develop multi-dimensional query, statistical analysis, and GIS map visualization functions. S7.2. Descriptive statistical analysis is performed on historical data according to climate zone and traffic level to identify the optimal adhesive type and construction process parameters. Association rule mining is performed using the Apriori algorithm and FP-Growth algorithm, with a minimum support of 5% to 10% and a minimum confidence of 60% to 80%. K-means or hierarchical clustering algorithms are used to cluster failure cases to identify typical failure modes. Based on new data, the prediction models in steps S3 and S5 are updated periodically using incremental learning methods. An intelligent decision support system is developed. The system takes project information, material information, environmental conditions and performance requirements as input, calls the knowledge base and prediction model to recommend adhesive formulations, predicts repair effects and service life, and generates a decision report that includes recommended solutions, construction suggestions, risk warnings and cost estimates.

[0068] In this embodiment: A standardized data collection template refers to a predefined, uniformly formatted data collection form, specifically using XML or JSON for field definition. Data integrity and consistency are ensured by setting required fields and data type constraints. A hybrid storage architecture combines multiple database technologies, such as relational databases for structured data, object storage for unstructured files, and time-series databases for sensor monitoring data, thus meeting the storage and retrieval needs of different data types. A multi-level user access control mechanism divides data access into hierarchical levels based on roles. For example, administrators can edit data, engineers can only query and analyze, and external personnel can only view statistical results, with access control implemented through an RBAC model. Incremental learning methods utilize online learning algorithms to update model parameters on new data, such as using stochastic gradient descent to progressively optimize model weights, avoiding the computational overhead of retraining with all data.

[0069] Standardized data collection templates ensure the comparability and analyzability of data generated from different projects by standardizing field definitions and input rules. The cloud computing knowledge base platform is modularly deployed based on a microservice architecture, with each functional module operating independently and communicating via API. In the hybrid storage architecture, a relational database stores project attribute data, object storage stores large files such as scanning electron microscope images, and a time-series database records continuous change data monitored by sensors. For data security, the transport layer uses TLS encryption, the storage layer uses the AES algorithm to encrypt sensitive data, and daily automatic backups to an off-site disaster recovery center prevent data loss. The data entry interface provides form filling and batch Excel import functions, and validation rules automatically detect outliers and trigger manual review processes. The statistical analysis module visualizes the distribution of repair effects in different areas using GIS maps, identifies high-success-rate combinations of formulas and construction parameters through association rule mining, and classifies failure cases according to patterns such as interface stripping and insufficient penetration through cluster analysis. The intelligent decision-making system calls the optimized prediction model, combines it with real-time environmental data to generate personalized repair plans, and outputs reports including material ratios, key points for construction temperature control, and detailed cost estimates.

[0070] Compared to existing technologies, which rely on paper records and scattered spreadsheets, resulting in disorganized data formats and difficulties in integration and analysis, this solution achieves standardized data management through standardized templates and a unified platform. Traditional technologies use a single database for storage, which cannot effectively handle multi-source heterogeneous data; this solution improves data access efficiency through a hybrid storage architecture. Existing systems lack dynamic update mechanisms, and fixed model parameters lead to prediction biases; this solution uses incremental learning to continuously optimize the model as data accumulates. Past decision-making relied on human experience, resulting in subjectivity and inconsistency; this solution extracts objective patterns through association rule mining and cluster analysis, supporting data-driven intelligent decision-making.

[0071] Through the above technical solutions, this application solves the problem of insufficient decision-making basis caused by data silos in traditional methods, and realizes standardized collection and centralized management of repair case data. The hybrid storage architecture and multi-dimensional analysis tools improve the processing efficiency and knowledge mining depth of massive amounts of data. The incremental learning-based model update mechanism ensures that the predictive model continuously adapts to changes in new materials and processes, improving the accuracy of recommended solutions. The intelligent decision support system integrates historical data and real-time monitoring information to provide personalized repair solutions for different working conditions, reducing reliance on human experience and decision-making risks.

[0072] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0073] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A method for evaluating the repair effect of asphalt pavement based on binders, characterized in that: The specific steps are as follows: S1. Fourier transform infrared spectroscopy and scanning electron microscopy were used to characterize the old pavement materials, nuclear magnetic resonance was used to analyze the molecular structure of the binder, and a material feature database was established using machine learning algorithms to predict the compatibility between the binder and the pavement materials. S2. The thickness of the interface layer was measured using atomic force microscopy, the chemical bonding density was quantitatively detected by X-ray photoelectron spectroscopy, the physical adsorption work was calculated based on the contact angle test, the mechanical interlocking degree was evaluated by three-dimensional laser scanning, and the interface interaction was verified by molecular dynamics simulation. S3. Extract microscopic parameters including interfacial bonding density, adsorption work, roughness index and penetration depth, conduct pull-out tests and fatigue tests to obtain macroscopic performance data, and use multivariate nonlinear regression and machine learning algorithms to establish a correlation prediction model between microscopic parameters and macroscopic performance. S4. Install fiber optic grating sensors in the repair test section for in-situ monitoring, design multi-factor coupled tests of temperature cycling, water immersion and dynamic load, take samples regularly to characterize interface degradation features, and establish a life prediction model based on the accelerated aging coefficient. S5. Determine the optimization objective based on the correlation model, use genetic algorithm or particle swarm algorithm to perform multi-objective optimization of the formula, and verify the optimization effect through microscopic characterization and macroscopic performance testing. S6. Use portable testing equipment on site to quickly determine the microscopic parameters of the interface, conduct pull-out tests and non-destructive testing, establish a multi-dimensional evaluation index system that includes microscopic, macroscopic and durability, and use a prediction model to output a comprehensive score. S7. Establish standardized data collection templates, build a cloud-based knowledge base platform for repair effects, optimize prediction models through big data mining and analysis, and develop an intelligent decision support system.

2. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 1, characterized in that, The specific method for step S1 is as follows: S1.1 The wavenumber of old asphalt at 1700 cm⁻¹ was determined using a Fourier transform infrared spectrometer. -1 The carbonyl index was calculated by measuring the absorption peak intensity, the softening point temperature was determined by differential scanning calorimetry to assess the degree of aging, the surface morphology of the aggregate was observed by scanning electron microscopy to obtain the roughness value, the mineral composition was analyzed by X-ray diffraction, the chemical shift and content of the active functional groups of the binder were determined by nuclear magnetic resonance spectroscopy, and the number-average molecular weight and polydispersity index were determined by gel permeation chromatography. A multidimensional characteristic parameter set including carbonyl index, softening point increment, surface roughness, mineral composition, functional group content and molecular weight parameters was established. S1.2 Construct a material feature database containing characteristic data and corresponding bonding performance indicators of pavement materials with different aging degrees, aggregates of different mineral types, and different types of binders. Extract carbonyl index, softening point increment, surface free energy, roughness, functional group concentration, and molecular weight as input parameters. Use random forest algorithm or neural network algorithm to establish a prediction model. Use chemical compatibility index and physical matching degree as output variables. Optimize the model through k-fold cross-validation. Input the parameters of the material to be evaluated and output the matching score value. When the score value is greater than or equal to the preset threshold, the matching is judged to be good.

3. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 2, characterized in that, The specific method for step S2 is as follows: S2.

1. The interface between the binder, old asphalt, and aggregate was scanned using an atomic force microscope to obtain the thickness distribution of the interface layer. The interface elements and chemical bonding type were analyzed using X-ray photoelectron spectroscopy. The interface chemical bonding density was calculated by peak fitting. The contact angle of the droplets on the old asphalt and aggregate surfaces was measured using a contact angle meter. The adhesion work was calculated based on the Young-Dupré equation, and the interface free energy was calculated using the Owens-Wendt method. The three-dimensional morphology of the aggregate surface was measured using a three-dimensional laser scanner. The surface roughness parameters Ra, Rz, and surface area ratio were extracted, and the correlation between roughness and mechanical interlocking strength was established. S2.2 Based on the experimental data in S2.1, molecular simulation software was used to construct models of binder molecules, aged asphalt, and aggregate surfaces. Periodic boundary conditions were set for molecular dynamics simulation. The COMPASS force field was used for energy minimization and equilibrium simulation. The interfacial binding energy and interaction energy were calculated, the interfacial stress distribution was analyzed, the contributions of hydrogen bonds and van der Waals forces were identified, and the interfacial layer thickness, number of bonding sites, and adsorption energy were calculated. The reliability of the experimental results was confirmed when the relative error between the simulation results and the experimental data was less than 15%.

4. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 3, characterized in that, The specific method for step S3 is as follows: S3.1 Extract the interfacial chemical bonding density, physical adsorption work, surface roughness index and penetration depth from step S2, prepare a composite specimen containing an old asphalt base layer, an adhesive interface layer and a repair material layer, conduct a pull-out test to determine the pull-out strength, conduct a shear test to determine the shear strength, conduct a dynamic fatigue test to record the number of cycles of failure, repeat the test for no less than 5 specimens in each group, calculate the average value and standard deviation of each performance index, and establish a sample dataset containing 4 micro parameters as input variables and 3 macro performance indexes as output variables; S3.

2. Key micro-parameters were screened using grey relational analysis, and a prediction model Y=f(X1,X2,X3,X4) was established using multivariate nonlinear regression. The regression coefficients were fitted using the least squares method. When the coefficient of determination R... 2 The model is considered effective when it is greater than or equal to 0.

85. The model is then optimized using support vector machine, random forest, or neural network algorithms. The training and test sets are divided into a 7:3 or 8:2 ratio. The hyperparameters are optimized using k-fold cross-validation. The mean absolute percentage error (MAPE) is calculated. When the MAPE is less than or equal to 10%, the final association prediction model is established.

5. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 4, characterized in that, The specific method for step S4 is as follows: S4.1 Fiber Bragg grating strain sensors and temperature sensors are embedded in the adhesive interface layer of the repair test section at intervals of 2m to 5m. The strain measurement range is -3000με to +3000με, and the temperature measurement range is -40℃ to +80℃. Soil moisture sensors and triaxial accelerometers are also installed. Data acquisition frequency is 1 hour to 24 hours. A temperature cycling test is designed, with a high-temperature stage of 60℃ to 80℃ for 4 to 8 hours and a low-temperature stage of -20℃ to -10℃ for 4 to 8 hours, for a total of at least 30 cycles. A water immersion test is designed, with immersion for 24 to 72 hours followed by freeze-thaw cycles. The freezing temperature is -18℃ to -15℃ for 12 hours, and the thawing temperature is 20℃ to 25℃ for 12 hours, for a total of at least 15 cycles. A dynamic load test is designed, with a load amplitude of 0.5MPa to 0.9MPa and a frequency of 10Hz to 20Hz, for a total of 10 load cycles. 4 Next to 10 6 Next, the above experiments were combined according to time series to form a multi-factor coupled accelerated experimental scheme; S4.

2. At least three specimens shall be sampled at the 0%, 25%, 50%, 75%, and 100% time points of the accelerated test, as well as at failure. The interfacial chemical bonding density, adsorption work, roughness, and penetration depth shall be determined using the method in step S2. The attenuation rate shall be calculated. Microcracks shall be observed using scanning electron microscopy to calculate the crack density. The porosity growth rate shall be determined using mercury intrusion porosimetry. Pull-out and shear tests shall be performed to calculate the strength retention rate. An accelerated aging model and a fitted apparent activation energy Ea shall be established based on the Arrhenius equation. The degradation kinetic model is as follows: t_real=t_test×exp[Ea / R×(1 / T_real-1 / T_test)]; P(t) = P0 × exp(-k × t^n); Where t_real is the time under actual service conditions, t_test is the time under accelerated test conditions, Ea is the apparent activation energy, R is the molar gas constant, T_real is the absolute temperature of the actual service environment, T_test is the absolute temperature of the accelerated test, P(t) is the performance parameter value at time t, P0 is the performance parameter value at the initial time, k is the degradation rate constant, and n is the degradation index. Failure is determined when the bond density decay rate is ≥50%, the strength retention rate is ≤60%, the crack density is ≥20mm / cm², or the porosity growth rate is ≥30%. The failure time t_failure is calculated, and the actual service life is calculated in combination with the acceleration factor. The specific method is as follows: L_service=t_failure×exp[Ea / R×(1 / T_real-1 / T_test)]; Where L_service is the actual service life and t_failure is the failure time under accelerated testing conditions.

6. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 5, characterized in that, The specific method for step S5 is as follows: S5.

1. Based on the correlation prediction model established in step S3, determine the target values ​​of pull-out strength, shear strength, and fatigue life. Solve the target values ​​of micro-parameters in reverse. Determine the design variables of the binder formulation, including the content of base asphalt, the amount of modifier, the amount of emulsifier, and the amount of filler. Set the value range and constraints of each component. Establish a multi-objective optimization function and solve it using a genetic algorithm or a particle swarm optimization algorithm. For the genetic algorithm, set the population size to 50 to 200, the crossover probability to 0.6 to 0.9, and the mutation probability to 0.01 to 0.

1. For the particle swarm optimization algorithm, set the number of particles to 30 to 100 and the learning factor to 1.5 to 2.

5. Obtain the Pareto optimal solution set and select the optimal formulation using the ideal point method or the weighted method. S5.2 Prepare binder samples according to the optimal formula. Heat the base asphalt to 160℃ to 180℃, and shear it at high speed at 3000rpm to 5000rpm for 30 to 60 minutes. The emulsification temperature is 120℃ to 140℃, and the emulsification speed is 8000rpm to 12000rpm. Prepare no less than 5 sets of parallel samples. Use the method in step S2 to determine the optimized micro parameters, and use the method in step S3.1 to determine the optimized macro properties. Calculate the improvement rate, input the measured micro parameters into the correlation model to calculate the predicted value, and compare it with the measured value. When the pull-out strength improvement rate is ≥20%, the fatigue life improvement rate is ≥30%, and the cost increase rate is ≤15%, the optimization is deemed effective.

7. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 6, characterized in that, The specific method for step S6 is as follows: S6.

1. Use a portable Fourier transform infrared spectrometer to scan the repaired section of the road surface after repair, with a wavenumber range of 400 cm⁻¹. -1 Up to 4000cm -1 The carbonyl index was determined, and the interfacial adhesion state was observed using a portable digital microscope at magnifications of 50x to 200x. The penetration depth of the adhesive was measured, and the arithmetic mean roughness Ra value was determined using a portable surface roughness meter, calculated per 100 μm. 2 Pull-out tests were conducted at no fewer than three test points. A portable pull-out tester was used to determine the pull-out strength at a loading rate of 0.5 MPa / s to 1.0 MPa / s. A deflectometer was used to determine the rebound deflection value and calculate the deflection recovery rate. Ground penetrating radar was used to detect interlayer continuity at an antenna frequency of 1 GHz to 2 GHz. An infrared thermal imager was used to detect the temperature distribution. A permeability coefficient meter was used to test the water tightness. A pendulum friction coefficient meter was used to determine the pendulum value BPN. The microscopic parameters were input into the correlation prediction model in step S3 and the life prediction model in step S4 to calculate the predicted service life. S6.2 Establish a three-level evaluation index system encompassing microscopic, macroscopic, and durability levels. Set the qualification standards as follows: carbonyl index <0.8, penetration depth ≥1.0mm, roughness Ra ≥0.5mm, tensile strength ≥1.5MPa, deflection recovery rate ≥80%, defect rate <5%, water permeability coefficient <120ml / min, pendulum value BPN ≥45, and predicted lifespan ≥5 years. Use the analytic hierarchy process (AHP) or entropy weighting method to determine the weight coefficients for each level of index. Normalize each index to convert it into a score value from 0 to 100. Calculate the comprehensive score using the weighted summation method, or use the machine learning prediction model from step S3 to output the comprehensive score and performance level. Based on the comprehensive score, classify the levels as follows: 90-100 points (Excellent), 80-90 points (Good), 60-80 points (Medium), and below 60 points (Poor). Generate an evaluation report containing test data, scoring results, level determination, and improvement suggestions.

8. The method for evaluating the effect of asphalt pavement repair based on binder according to claim 7, characterized in that, The specific method for step S7 is as follows: S7.1 Establish a standardized data collection template, which includes fields for basic project information, disease characteristics, material parameters, construction technology, microscopic characterization data, macroscopic performance data, environmental monitoring data, and usage effect data. Define the data type, value range, and required field identifiers for each field. Construct a cloud-based knowledge base platform using a B / S architecture, a microservice architecture for the backend, and a hybrid storage architecture for the database, including relational databases, object storage, and time-series databases. Set up a multi-level user permission management mechanism, use TLS encryption protocol and AES algorithm to ensure data security, establish a daily automatic backup mechanism, develop a data entry interface that supports manual input and batch import, set up data verification rules and review processes, and develop multi-dimensional query, statistical analysis, and GIS map visualization functions. S7.

2. Descriptive statistical analysis is performed on historical data according to climate zone and traffic level to identify the optimal adhesive type and construction process parameters. Association rule mining is performed using the Apriori algorithm and FP-Growth algorithm, with a minimum support of 5% to 10% and a minimum confidence of 60% to 80%. K-means or hierarchical clustering algorithms are used to cluster failure cases to identify typical failure modes. Based on new data, the prediction models in steps S3 and S5 are updated periodically using incremental learning methods. An intelligent decision support system is developed. The system takes project information, material information, environmental conditions and performance requirements as input, calls the knowledge base and prediction model to recommend adhesive formulations, predicts repair effects and service life, and generates a decision report that includes recommended solutions, construction suggestions, risk warnings and cost estimates.

Images