Roadbed and pavement compactness detection and regulation method for road engineering

Abstract

The invention discloses a roadbed and pavement compactness detection and regulation method for road engineering. The method comprises the following steps: deploying an intrusive sensor array to monitor an internal physical state and water-heat-force coupling phase change data; synchronously acquiring operation data of intelligent compaction equipment and scanning data of nondestructive detection equipment, and identifying a phase change critical point of the material; a disturbance propagation coefficient is calculated through a mechanical transmission chain model after data fusion, a target point is determined, and automatic micro sampling is carried out; and fusing all data to establish a macro-medium-micro three-scale fused compaction prediction model, generating a real-time regulation and control instruction to realize closed-loop control, and finally generating a three-dimensional compaction degree distribution digital twin model. According to the method, multi-source data are integrated, material phase change and mechanical transfer rules are associated, and the stability and the intelligent level of compaction quality are improved.

Description

Technical Field

[0001] This invention belongs to the field of road engineering technology, specifically relating to a method for detecting and controlling the compaction degree of roadbed and pavement in road engineering. Background Technology

[0002] The compaction degree of roadbed and pavement is a core indicator for evaluating the construction quality of road engineering, directly affecting the road's load-bearing capacity, stability, and service life. Traditional compaction degree testing and control methods mainly rely on manual sampling, laboratory tests, and empirical parameter adjustments. While these methods can meet basic quality requirements, they suffer from limitations such as low testing efficiency, high destructiveness, and data lag, making them unsuitable for the demands of large-scale, high-precision road construction. With the development of intelligent technologies, the industry has gradually introduced non-destructive testing equipment such as ground-penetrating radar and intelligent compaction instruments, as well as automated compaction equipment. By collecting data such as compaction trajectories and surface settlement, these devices assist in decision-making, improving the continuity of testing and the timeliness of control to some extent.

[0003] However, existing technologies still have significant shortcomings: on the one hand, the integration of detection data is insufficient, failing to effectively link the internal physical state of the subgrade and pavement with the phase transformation characteristics of materials, resulting in a one-sided understanding of the mechanical behavior of materials during compaction and difficulty in accurately capturing the impact of the phase transformation critical point on the compaction effect; on the other hand, the control logic is mostly based on empirical models of a single scale, lacking multi-scale analysis of the compaction energy transfer law, and unable to achieve a dynamic closed loop of detection-prediction-control, resulting in large fluctuations in compaction quality, especially in complex environments such as frozen soil areas and high-temperature areas, where problems such as insufficient compaction or over-compaction are prone to occur.

[0004] Therefore, the invention proposes a method for detecting and controlling the compaction degree of roadbed and pavement in road engineering to at least partially solve the above problems. Summary of the Invention

[0005] To address the aforementioned problems in existing technologies, this invention provides a method for detecting and controlling the compaction degree of roadbed and pavement in road engineering. This method solves the problems in existing technologies, such as insufficient data fusion, failure to correlate internal physical states with material phase transformation characteristics, and the inability to achieve dynamic closed-loop control due to the single-scale empirical model and lack of multi-scale analysis, which leads to large fluctuations in compaction quality under complex environments.

[0006] The objective of this invention can be achieved through the following technical solution: a method for detecting and controlling the compaction degree of roadbed and pavement in road engineering, comprising: Step S1: Deploy an invasive sensor array within the subgrade or pavement structure of the road section to be inspected, and simultaneously monitor the internal physical state data and water-thermal-mechanical coupling phase change data. The phase change data includes at least the unfrozen water content in the freeze-thaw cycle or the glass transition temperature of the asphalt mixture. Step S2: During the compaction process, the operation process data of the intelligent compaction equipment and the scanning data of the non-destructive testing equipment are acquired simultaneously. The non-destructive testing equipment is equipped with a dielectric constant dispersion characteristic analyzer to identify the critical point of material phase transition. Step S3: Based on the real-time fusion analysis of the internal physical state data, operation process and scanning data, calculate the disturbance propagation coefficient through the mechanical transmission chain model, and determine the target point sampling depth according to the dynamic sampling depth calculation model, and determine at least one verification point; Step S4: Perform automated micro-sampling at the target point to obtain physical parameter data for model calibration; Step S5: Integrate all data to establish a compaction prediction model that integrates macro, meso, and micro scales, and generate real-time control commands for compaction equipment based on the model prediction results to achieve closed-loop quality control.

[0007] As a preferred embodiment of the present invention, the invasive sensor array is a fiber Bragg grating sensor array, and the frozen water content is inverted by the wavelength offset of the fiber Bragg grating. The internal physical state data includes at least strain or temperature.

[0008] As a preferred technical solution of the present invention, in step S2, the operation process data of the intelligent compaction equipment includes at least one or more of the following: compaction trajectory, number of compaction passes, compactor value, or surface temperature; the non-destructive testing equipment is a ground penetrating radar, which identifies the phase transition critical point in the temperature range of -5℃ to 5℃ by analyzing the dielectric constant dispersion characteristics.

[0009] As a preferred embodiment of the present invention, the method for determining the target verification point in step S3 includes: constructing a mechanical transmission chain model, the transmission path of which is compaction energy → surface strain → deep stress wave attenuation → subgrade modulus; and calculating the disturbance propagation coefficient η in real time based on the mechanical transmission chain model using edge computing devices deployed on-site, wherein the formula for calculating the disturbance propagation coefficient η is: in, The surface layer represents the real-time strain change of the roadbed surface material obtained through an invasive sensor array; the deep layer E represents the elastic modulus of the deep roadbed material obtained by inversion based on the attenuation law of deep stress waves; σ vibration represents the vibration stress generated during the operation of the compaction equipment; f vibration represents the current real-time vibration frequency of the compaction equipment; and f reference represents the preset reference vibration frequency.

[0010] In a preferred embodiment of the present invention, in step S4, the automated micro-sampling is performed using an automated micro-drilling device. The automated micro-drilling device automatically adjusts the drilling depth based on the sampling depth parameters calculated in real time, and the sampling depth h satisfies the following formula: Where v is the real-time travel speed of the intelligent compaction equipment; η is the response delay time of the invasive sensor array; k is the disturbance compensation factor; and η is the disturbance propagation coefficient calculated in step S3.

[0011] As a preferred embodiment of the present invention, in step S5, the established compaction degree prediction model adopts a macro-meta-micro three-scale fusion architecture, specifically including: Macro-scale layer: The Transformer model is used to process the full-field scanning data of ground penetrating radar, and the macro-scale stress distribution σ of the region is output. Mesoscale layer: Based on graph neural network, a particle contact force chain network of roadbed material is constructed. The strain data obtained by invasive sensor array is used to reconstruct the force transmission path between particles and output the mesoscale elastic modulus Emedium and stress distribution σmedium. Microscale layer: By simulating the phase transition process of materials through molecular dynamics, the molecular potential function of asphalt mixture or soil is embedded, and the microscopic elastic modulus Emicro is output. The macro-meta-micro three-scale model is trained through cross-level loss function coupling, whereby the loss function is: Where α is the stress coupling coefficient and β is the modulus coupling coefficient. This is the L2 norm squaring operation.

[0012] As a preferred embodiment of the present invention, the Leonard-Jones potential function or other empirical potential function describing intermolecular interactions embedded in asphalt mixtures or soil has a macro-metascale data resolution of 0.5m×0.5m, and the stress coupling coefficient α and modulus coupling coefficient β in the cross-level loss function are calibrated by a compaction work transfer efficiency test.

[0013] As a preferred embodiment of the present invention, in step S5, the real-time control command includes: When the unfrozen water content is detected to be greater than the critical value, a compensation strategy for the frozen soil area is triggered: reduce the vibration frequency and increase the amplitude. When the glass transition temperature of asphalt deviates from the set range, a high-temperature zone compensation strategy is triggered: shorten the duration of a single compaction pass and increase the number of compaction passes.

[0014] As a preferred technical solution of the present invention, the method ultimately generates a three-dimensional digital twin model of compaction distribution, which integrates and displays the material phase change state layer and the mechanical transmission chain path, and is used to identify phase change risk areas with different colors.

[0015] The beneficial effects of this invention are as follows: By combining an invasive sensor array with non-destructive testing equipment, synchronous monitoring of multi-dimensional data on the roadbed interior and surface is achieved, breaking the limitations of traditional manual sampling and ensuring comprehensiveness, accuracy, and efficiency of detection, while accurately capturing the critical point of material phase transition; the compaction prediction model, which integrates macro-meta-micro scales, deeply understands the mechanical behavior of materials through cross-level analysis and training, and can accurately locate the root cause of compaction problems, overcoming the limitations of traditional single-scale empirical models; real-time control commands based on model predictions can dynamically adjust equipment parameters for complex scenarios such as frozen soil areas and high-temperature areas, reducing insufficient or excessive compaction and lowering rework rates; the final generated three-dimensional compaction degree distribution digital twin model visualizes the phase transition state and mechanical path, making it easy for construction personnel to intuitively grasp the quality status and potential risks, improving management convenience; the overall solution integrates multi-source data acquisition, intelligent analysis, and real-time closed-loop control technologies, realizing an intelligent upgrade of compaction degree detection and control in road engineering, accelerating progress and saving costs while ensuring construction quality, and providing strong support for high-quality road construction. Attached Figure Description

[0016] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart of the steps of the present invention; Figure 2 This is a flowchart illustrating the operation process of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0019] In this invention, the molecular potential function refers to the empirical potential function used in molecular dynamics simulations to describe non-bonded interactions between atoms or molecules, including but not limited to the Lennard-Jones potential and the Morse potential. This term is used specifically to refer to such classical, parameterized potential function models, rather than to all possible mathematical models of intermolecular forces.

[0020] Please see Figures 1-2 This embodiment provides a method for detecting and controlling the compaction degree of roadbed and pavement in road engineering, as disclosed in this invention. The core technical points of this invention will be described in detail below.

[0021] In this embodiment, the invasive sensor array mentioned in step S1 is preferably a fiber Bragg grating (FBG) sensor array, because it has the advantages of anti-electromagnetic interference, high precision, and distributed measurement capability.

[0022] Deployment method: During the roadbed filling process, after each layer is laid, an FBG sensor array is deployed in a grid pattern at a depth of approximately 10cm from the top surface of that layer. The specific deployment spacing is adjusted according to the engineering accuracy requirements, for example, a longitudinal spacing of 5m and a transverse spacing of 3m along the road. This layered, gridded deployment method can construct a three-dimensional internal state perception network.

[0023] Data Acquisition and Transmission: Each FBG sensor is connected to a fiber optic demodulator deployed on-site. The demodulator acquires the reflected wavelength data of the grating at high frequency. The acquired data is transmitted in real time via a wireless transmission module to edge computing devices deployed on the construction site for preliminary processing and storage.

[0024] Inversion Algorithm Implementation: FBG sensors are sensitive to both temperature and strain, and their wavelength shift is a function of both. To decouple these factors, temperature-compensated FBGs are typically used, or temperature-only FBG sensors are deployed nearby. For frozen soil subgrades, the specific implementation of inverting unfrozen water content using wavelength shift is as follows: First, soil samples with different moisture contents and freezing temperatures are prepared in the laboratory for the target soil type, and FBG sensors are embedded. The unfrozen water content of the soil samples at different sub-zero temperatures is accurately measured using methods such as nuclear magnetic resonance, while simultaneously recording the FBG wavelength shift. Based on this, a calibration database and empirical formulas or machine learning models are established to determine the relationship between wavelength shift, temperature, and unfrozen water content for a specific soil type. In field applications, the unfrozen water content at the corresponding location can be inverted based on the real-time measured wavelength shift and temperature.

[0025] Step S2 aims to acquire macroscopic data from external devices and combine it with internal sensor information to identify key critical points of material phase transition. Specifically, the intelligent compaction equipment acquires data in real time from a modern intelligent road roller equipped with an onboard control system and sensors via the roller's controller LAN bus or proprietary data interface. This data includes real-time position coordinates, travel speed, vibration frequency, vibration amplitude, and compaction degree indicators, and is synchronized with timestamps before being sent to the edge computing device. For non-destructive testing equipment, a vehicle-mounted ground-penetrating radar is preferred. Installed in front of the compaction equipment or on an independent inspection vehicle, it pre-scans the area to be compacted, generating a dielectric constant profile reflecting material uniformity, moisture content distribution, and potential defects. The scan data is then compared with RTK data. GPS coordinate binding; in terms of material phase transformation critical point identification, the frozen soil phase transformation is combined with the internal temperature measured by the FBG sensor in step S1 and the inverted unfrozen water content to set a freezing sensitivity threshold. For example, when the temperature of silty clay is below -0.5℃ and the unfrozen water content drops sharply, it is determined to have entered the phase transformation critical zone; the asphalt phase transformation uses an infrared thermometer on the compaction equipment to monitor the surface temperature of the asphalt mixture, presets the glass transition temperature according to the asphalt material type, defines the optimal compaction temperature range, and identifies the phase transformation critical state when the real-time temperature exceeds the threshold, at which point the compaction effect will deteriorate sharply.

[0026] The core of step S3 lies in constructing a mechanical transmission chain model and calculating the disturbance propagation coefficient η, which is crucial for achieving intelligent analysis and positioning in this invention. The calculation formula is as follows: The physical meaning and acquisition method of each parameter are explained in detail below: Real-time strain change of subgrade surface material: This value is directly measured by FBG strain sensors deployed on the subgrade surface. When the compaction equipment passes over the sensor, the demodulator records the peak strain change, which is the value of the strain. surface layer.

[0027] Edeep subgrade material elastic modulus: This parameter is obtained through inversion based on the attenuation law of deep stress waves. Specifically, FBG sensors are also deployed in the deep subgrade. When the compaction equipment operates on the surface, its vibration energy propagates into the depth as stress waves. By comparing the amplitude attenuation and phase delay of the stress wave signals measured by surface and deep sensors, and combining this with wave theory, the equivalent elastic modulus Edeep along the wave propagation path in the medium can be calculated.

[0028] σvibration stress generated by the compaction equipment: This value is provided by the sensor system built into the intelligent compaction equipment itself. Modern intelligent road rollers are usually equipped with accelerometers to measure the vibration acceleration of the vibrating steel drum. Combined with the mass of the steel drum and the contact model with the ground, the dynamic vibration stress σvibration applied to the roadbed surface can be estimated.

[0029] f. Current real-time vibration frequency of the compaction equipment: This parameter can be directly read from the local area network bus of the compaction equipment's controller and is the current operating parameter of the equipment.

[0030] f. Preset reference vibration frequency: A preset parameter used for normalization. Its value is preset in the system according to the compaction specifications and material type. For example, it can be set to 30Hz, which is commonly used in the industry.

[0031] The physical meaning of η: η is a dimensionless coefficient that intuitively reflects the efficiency of compaction energy transfer from the surface to the deeper layers. A larger η value indicates relatively large surface strain and lower deep layer modulus, or smaller vibration energy input, resulting in severe energy attenuation and low efficiency during propagation to deeper layers. Therefore, areas with large η values ​​may indicate a risk of insufficient deep compaction, requiring further sampling and verification.

[0032] Step S4 uses an automated micro-drilling device for precise verification, and its workflow is as follows: The equipment structure is integrated on a small tracked or wheeled platform and mainly includes: a high-precision real-time dynamic (RTK-GPS) positioning system, an automatically controlled drilling system, a sample packaging system, and a module for communicating with the main control system. Its automation process is as follows: Upon receiving instructions, the edge computing device determines the coordinates of one or more target points to be verified based on the calculated η value, and then... Calculate the sampling depth h. Where v is the compactor's travel speed. The sensor response delay is denoted by k, and the disturbance compensation factor is denoted by 'h'. The coordinates and depth h are sent to the micro-drilling equipment. Upon receiving the command, the micro-drilling equipment automatically travels to the target point using an RTK-GPS system, with errors controllable to the centimeter level. Upon arrival, the drilling system starts, drilling at the preset rotation speed and feed rate according to the set depth h. After drilling to the designated depth h, a sampling tube collects an undisturbed soil sample at that depth. The sample is automatically ejected and sealed in a sample box with a QR code label, the label information being associated with the sampling location, depth, time, and other data. The retrieved sample is transported to a mobile testing vehicle next to the construction site. The vehicle is equipped with rapid testing instruments, such as a microwave moisture meter and a small density meter, which can determine the actual physical parameters of the soil sample within minutes. The measurement results are automatically uploaded via scanning the QR code and compared with the predicted data in the system for calibration and correction of subsequent prediction models.

[0033] The macro-meta-micro three-scale fusion model constructed in step S5 is one of the core prediction features of this invention, and its data flow and construction details are as follows: The macroscale layer model and its input / output: A Transformer model is employed. Its input is full-field scanning data acquired by a non-destructive testing device. This data is preprocessed into a series of profile image patches representing information such as dielectric constant and conductivity; for example, each image patch corresponds to a 0.5m × 0.5m region. The Transformer model, through its self-attention mechanism, captures spatial correlations over a large area and outputs the macroscopic stress distribution σmacro of that region.

[0034] Mesoscale Layer Model and Input / Output: A particle contact force chain network for roadbed materials is constructed based on a graph neural network. In this network, each FBG sensor is considered a node, and its measured real-time strain data serves as the node feature. The spatial relationships between sensors define the edges of the graph. Through training the graph neural network, the force transmission path between particles is reconstructed, thereby outputting a more refined mesoscale elastic modulus Eintermediate and stress distribution σintermediate.

[0035] Microscale layer modeling and input / output: Simulating phase transition processes in materials through molecular dynamics. For example, when dealing with asphalt mixtures, specialized software such as Materials Studio can be used to construct atomic models containing thousands of asphalt and aggregate molecules. Using mature force fields such as molecular potential functions, different temperature and pressure conditions are simulated under isothermal and isochoric ensembles to calculate the microscopic elastic modulus Emicro of the material. The simulation results provide physical mechanism support for mesoscale models.

[0036] Model training and coupling employ an offline pre-training plus online fine-tuning approach: large-scale pre-training is first conducted using historical engineering data and laboratory data, followed by online fine-tuning of the model on-site using calibration data obtained from automated micro-drilling equipment to adapt to the current working conditions; cross-level loss function. The model is used to couple three scales, where α is the stress coupling coefficient and β is the modulus coupling coefficient. This involves L2 norm squaring. L2 norm squaring is the sum of the squares of the elements of a vector, used to quantify the difference between two vectors. Furthermore, this operation allows for the measurement of errors in stress and modulus parameters at different scales, providing a quantitative basis for cross-level model coupling. The Leonard-Jones potential function or other empirical potential functions describing intermolecular interactions are embedded in the asphalt mixture or soil. The macro-to-mesoscale data resolution reaches 0.5m × 0.5m. The coupling coefficients α and β are calibrated through compaction work transfer efficiency tests. Specifically, compaction operations with different parameters (vibration frequency, amplitude, compaction speed) are performed at the test site, while simultaneously measuring the total work input to the compactor and the increase in strain energy within the subgrade measured by a sensor array. By analyzing the relationship between the input work and the increase in strain energy, the weight of energy transfer between different scales is calibrated, thereby determining the values ​​of α and β.

[0037] Based on the prediction results of the three-scale model, the system generates real-time control commands for the intelligent compaction equipment to cope with complex working conditions: Compensation strategy for frozen soil areas: When the model predicts or sensors detect that the unfrozen water content in a certain area exceeds a preset critical value, the system determines that compaction in that area is difficult. At this point, a compensation strategy is triggered, sending a command to the compaction equipment: "Reduce the current vibration frequency from the standard 30Hz to 22Hz, and simultaneously increase the amplitude from 1.8mm to 2.2mm." This low-frequency, high-amplitude vibration mode helps overcome the additional resistance caused by soil freezing, increasing the penetration depth of compaction energy.

[0038] High-Temperature Compensation Strategy: In asphalt pavement construction, the glass transition temperature (Tg) of asphalt is a critical parameter. This invention sets an optimal compaction temperature range, for example, for a certain type of styrene-butadiene-styrene block copolymer modified asphalt, it is set to Tg + 90℃ ± 5℃. When the asphalt temperature predicted by the infrared thermometer on the compactor surface or the model deviates from this range, a corresponding strategy is triggered. If the temperature is too high, the instruction is to shorten the duration of a single compaction pass and increase the number of compaction passes to avoid excessive asphalt fluidity at high temperatures leading to shoving and shoving; if the temperature is too low, the instruction is to accelerate the compaction pace or stop the operation. The critical value for unfrozen water content is calibrated based on soil type through frost heave tests; the asphalt glass transition temperature setting range is determined based on the asphalt grade through indoor compaction effect tests to ensure that the range matches the material properties.

[0039] The resulting three-dimensional digital twin model of compaction distribution provides managers with intuitive decision support. This model can be rendered and interacted with in three dimensions on web-based graphics libraries or in client software developed using professional engines such as Unity3D. Its information fusion display includes a phase change state layer and a mechanical transmission chain path. The phase change state layer is colored according to the material's phase change risk; for example, in the frozen soil subgrade model, a gradient heat map from blue to red is used to mark areas with high unfrozen water content, visually presenting the risk of frost heave. The mechanical transmission chain path is displayed by overlaying three-dimensional pipeline diagrams with arrows and varying thicknesses. The pipeline thickness represents stress magnitude, and the color depth represents energy attenuation, allowing managers to intuitively grasp the distribution of compaction energy within the subgrade. In summary, this invention, by integrating multi-source sensing, multi-scale modeling, and closed-loop intelligent control technologies, achieves refined and intelligent management of the road engineering compaction process, significantly improving construction quality and efficiency.

[0040] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for detecting and controlling the compaction degree of roadbed and pavement in road engineering, characterized in that, include: Step S1: Deploy an invasive sensor array within the subgrade or pavement structure of the road section to be inspected, and simultaneously monitor the internal physical state data and water-thermal-mechanical coupling phase change data. The phase change data includes at least the unfrozen water content in the freeze-thaw cycle or the glass transition temperature of the asphalt mixture. Step S2: During the compaction process, the operation process data of the intelligent compaction equipment and the scanning data of the non-destructive testing equipment are acquired simultaneously. The non-destructive testing equipment is equipped with a dielectric constant dispersion characteristic analyzer to identify the critical point of material phase transition. Step S3: Based on the real-time fusion analysis of the internal physical state data, operation process and scanning data, calculate the disturbance propagation coefficient through the mechanical transmission chain model, and determine the target point sampling depth according to the dynamic sampling depth calculation model, and determine at least one verification point; Step S4: Perform automated micro-sampling at the target point to obtain physical parameter data for model calibration; Step S5: Integrate all data to establish a compaction prediction model that integrates macro, meso, and micro scales, and generate real-time control commands for compaction equipment based on the model prediction results to achieve closed-loop quality control.

2. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, The invasive sensor array is a fiber Bragg grating sensor array, which inverts the frozen water content by measuring the wavelength offset of the fiber Bragg grating. The internal physical state data includes at least strain or temperature.

3. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, In step S2, the operation process data of the intelligent compaction equipment includes at least one or more of the following: compaction trajectory, number of compaction passes, compactor value, or surface temperature; the non-destructive testing equipment is a ground-penetrating radar, which identifies the phase transition critical point in the temperature range of -5℃ to 5℃ by analyzing the dielectric constant dispersion characteristics.

4. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, In step S3, the method for determining the target verification point includes: constructing a mechanical transmission chain model, the transmission path of which is compaction energy → surface strain → deep stress wave attenuation → subgrade modulus; and using edge computing devices deployed on-site to calculate the disturbance propagation coefficient η in real time based on the mechanical transmission chain model, the formula for calculating the disturbance propagation coefficient η is as follows: in, The surface layer represents the real-time strain change of the roadbed surface material obtained through an invasive sensor array; the deep layer E represents the elastic modulus of the deep roadbed material obtained by inversion based on the attenuation law of deep stress waves; σ vibration represents the vibration stress generated during the operation of the compaction equipment; f vibration represents the current real-time vibration frequency of the compaction equipment; and f reference represents the preset reference vibration frequency.

5. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 4, characterized in that, In step S4, the automated micro-sampling is performed using an automated micro-drilling device. The automated micro-drilling device automatically adjusts the drilling depth based on the real-time calculated sampling depth parameters. The sampling depth h satisfies the following formula: Where v is the real-time travel speed of the intelligent compaction equipment; η is the response delay time of the invasive sensor array; k is the disturbance compensation factor; and η is the disturbance propagation coefficient calculated in step S3.

6. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, In step S5, the established compaction degree prediction model adopts a macro-meta-micro three-scale fusion architecture, specifically including: Macro-scale layer: The Transformer model is used to process the full-field scanning data of ground penetrating radar, and the macro-scale stress distribution σ of the region is output. Mesoscale layer: Based on graph neural network, a particle contact force chain network of roadbed material is constructed. The strain data obtained by invasive sensor array is used to reconstruct the force transmission path between particles and output the mesoscale elastic modulus Emedium and stress distribution σmedium. Microscale layer: By simulating the phase transition process of materials through molecular dynamics, the molecular potential function of asphalt mixture or soil is embedded, and the microscopic elastic modulus Emicro is output. The macro-meta-micro three-scale model is trained through cross-level loss function coupling, whereby the loss function is: Where α is the stress coupling coefficient and β is the modulus coupling coefficient. This is the L2 norm squaring operation.

7. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 6, characterized in that, The Leonard-Jones potential function or other empirical potential function describing intermolecular interactions embedded in the asphalt mixture or soil has a macro-metascale data resolution of 0.5m×0.5m. The stress coupling coefficient α and modulus coupling coefficient β in the cross-level loss function are calibrated through a compaction work transfer efficiency test.

8. The method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, In step S5, the real-time control instructions include: When the unfrozen water content is detected to be greater than the critical value, a compensation strategy for the frozen soil area is triggered: reduce the vibration frequency and increase the amplitude. When the glass transition temperature of asphalt deviates from the set range, a high-temperature zone compensation strategy is triggered: shorten the duration of a single compaction pass and increase the number of compaction passes.

9. A method for detecting and controlling the compaction degree of roadbed and pavement in road engineering according to claim 1, characterized in that, The method ultimately generates a three-dimensional digital twin model of compaction distribution, which integrates the material phase change state layer and mechanical transmission chain path to identify phase change risk areas with different colors.

Images