Intelligent asphalt pavement rolling control method and system

Abstract

The invention relates to the field of highway construction, in particular to an intelligent asphalt pavement rolling control method and system, and the method comprises the steps: deploying machine vision equipment to collect an image of a just-paved asphalt mixture, recognizing the segregation degree in combination with a convolutional neural network, and generating a segregation distribution thermodynamic diagram with geographic coordinates; the road roller is integrated with a positioning sensor, a temperature sensor, a vibration sensor and a speed sensor, relevant parameters are collected, and the real-time compaction degree is calculated. The thermodynamic diagram and the compaction degree are subjected to space-time matching, optimal operation parameters are calculated through a dynamic coupling model on the basis of data such as the segregation degree and the temperature of a target area, the optimal operation parameters are wirelessly sent to a road roller control system to adjust speed and vibration parameters, and a road roller feeds back the new compaction degree after execution to form closed-loop control. The problem that separation and compaction control are disjointed in the prior art is solved, separation information and the compaction process are deeply fused, and underpressure of a separation area and overpressure of a uniform area are avoided.

Description

Technical Field

[0001] This invention relates to the field of highway construction, specifically to an intelligent rolling control method and system for asphalt pavement. Background Technology

[0002] The compaction quality of asphalt pavement directly determines the durability and driving safety of highway structures. With the development of intelligent construction technology, real-time monitoring and dynamic control have become the core directions for improving compaction accuracy. The CN112575656A patent published by Southeast University, "A Multifunctional Sensor and Arrangement Method for Asphalt Pavement Compaction Degree Detection," still faces significant technical bottlenecks in practical applications: its core control logic is still guided by a single compaction degree index, failing to consider the impact of differences in the initial state of the asphalt mixture on the compaction effect, especially the interference of segregation, a key defect. Segregation leads to uneven gradation of the mixture, causing under the same compaction parameters, segregated areas often require higher energy and thus experience insufficient compaction, while uniform areas may experience aggregate breakage due to excess energy, creating a dual problem of quality risks and energy waste. Furthermore, existing technologies lack a linkage mechanism between segregation detection methods such as machine vision and the compaction control system. Segregation information is only used for post-event evaluation and cannot provide feedforward guidance for adjusting compaction parameters, resulting in the roller's operating parameters always being in a passive response state, making it difficult to achieve a balance between quality and efficiency. Summary of the Invention

[0003] The purpose of this invention is to provide an intelligent compaction control method and system for asphalt pavement, in order to solve the problem that existing asphalt pavement compaction technology only takes compaction degree as the core control target and does not link segregation detection results with the compaction process, resulting in insufficient compaction of segregation areas and over-compaction of uniform areas, making it difficult to balance compaction quality and construction efficiency.

[0004] To achieve the above objectives, the following technical solution is adopted.

[0005] A method for intelligent compaction control of asphalt pavement includes: deploying machine vision equipment to acquire images of the surface of freshly laid but uncompacted asphalt mixture; identifying the degree of segregation using a pre-set convolutional neural network and generating a segregation distribution heat map with geographic coordinates; integrating a positioning module, temperature sensor, vibration sensor, and speed sensor on a road roller to acquire location, number of compaction passes, speed, vibration parameters, and temperature, and calculating real-time compaction values; performing spatiotemporal matching between the segregation distribution heat map and the real-time compaction values; calculating the optimal combination of operating parameters based on the degree of segregation, temperature, number of compaction passes, and current compaction value of the area to be compacted by the road roller, using a dynamic coupling model; wirelessly transmitting the optimal combination of operating parameters to the road roller control system to automatically adjust the travel speed and vibration parameters; and collecting new compaction values ​​after the road roller performs the adjustment and feeding them back to the dynamic coupling model for evaluation and triggering secondary adjustments, forming a closed-loop control.

[0006] Optionally, generating a segregation distribution heat map with geographic coordinates includes: acquiring images of the asphalt mixture surface using a linear or area array camera behind the paver, inputting the images into a convolutional neural network to output the segregation level of each area, and mapping the segregation level to a geographic coordinate system using RTK / GNSS positioning data to generate a spatial heat map covering the construction surface with the degree of segregation marked at each point, which serves as the spatial input basis for the dynamic coupling model.

[0007] Optionally, the calculation of the real-time compaction value includes: the roller recording its own position through a GNSS positioning module, collecting vibration waveforms through a vibration sensor, collecting the temperature of the mixture through a temperature sensor, inputting the vibration waveforms into an intelligent compaction algorithm module, calculating the compaction value at the current position by combining the temperature and position, and updating the number of rolling passes for each grid according to the number of times the trajectory coincides with the grid, forming a multi-dimensional compaction state dataset containing position, compaction degree, temperature, number of passes, speed and vibration parameters.

[0008] Optionally, the step of calculating the optimal combination of operating parameters through a dynamic coupling model includes: the central system matching the roller trajectory with the heat map, extracting the segregation degree, temperature, number of compaction passes, and current compaction degree of the target area, and inputting them into the dynamic coupling model; the dynamic coupling model sets the energy demand weight based on the segregation degree, corrects the attenuation coefficient based on the temperature, calculates the energy gap based on the number of passes and compaction degree, and outputs recommended driving speed, vibration frequency, and vibration amplitude values ​​to form a complete parameter combination for roller adjustment.

[0009] Optionally, the automatic adjustment of driving speed and vibration parameters includes: the central system encapsulates the recommended driving speed value, vibration frequency value, and vibration amplitude value into control commands, and sends them to the roller's on-board controller via wireless communication; after the controller parses the commands, it drives the walking system to adjust the speed and drives the vibration hydraulic system to adjust the frequency and amplitude, so that the roller performs parameter matching before entering the target area or during compaction, in order to avoid over-compaction in the uniform zone or under-compaction in the segregation zone.

[0010] Optionally, the feedback and triggering of secondary adjustments includes: after the roller compacts according to the adjusted parameters, it re-collects vibration and temperature data, calculates a new compaction value, and uploads it; the central system compares the new compaction value with the target threshold, and if the deviation exceeds the limit, it calls the dynamic coupling model again, combines the updated segregation degree, temperature, number of passes, and compaction value, calculates the optimization parameters for the next stage, and issues instructions again until the compaction degree stabilizes and meets the target, thus achieving closed-loop iterative control.

[0011] Optionally, when calculating parameters in the dynamic coupling model, a differentiated strategy is implemented: when the area is identified as a severely segregated region, the driving speed is reduced, the vibration frequency or amplitude is increased, or the upper limit of the number of compaction passes is increased; when the area is identified as a mild or uniform region, the driving speed is increased, the vibration frequency or amplitude is reduced, or the upper limit of the number of compaction passes is decreased; when the area is identified as a low-temperature region, the driving speed is forcibly reduced and the vibration frequency or amplitude is increased to compensate for the reduction in compaction efficiency caused by temperature.

[0012] An intelligent asphalt pavement compaction control system includes: The segregation perception subsystem, including machine vision equipment and segregation recognition algorithm module, is used to generate a heat map of segregation distribution with geographic coordinates; The compaction monitoring subsystem includes a positioning module, a temperature sensor, a vibration sensor, a velocity sensor, and a compaction calculation module, which is used to collect compaction status data. The central processing and decision-making subsystem runs a dynamically coupled model to match heat maps and compaction data and calculate optimal parameters; The road roller control subsystem receives commands and adjusts the travel speed and vibration parameters; the human-machine interface is used to display heat maps, compaction maps, commands, and early warning information.

[0013] Optionally, the central processing and decision-making subsystem performs the following: matching the roller trajectory with the heat map, extracting the segregation degree, temperature, number of passes, and compaction degree value of the target area, and inputting the dynamic coupling model to calculate the recommended speed, frequency, and amplitude value; receiving feedback compaction degree value, and if it does not meet the standard, recalculating and issuing new instructions to form a closed loop; the central processing and decision-making subsystem interacts with the segregation sensing subsystem, compaction monitoring subsystem, and roller control subsystem in real time via a wireless network to ensure that instructions are synchronized with the field.

[0014] Optionally, the dynamic coupling model embeds a differentiated control rule base: severely segregated regions trigger speed reduction, frequency / amplitude increase, or number of passes increase commands; mild or uniform regions trigger speed increase, frequency / amplitude decrease, or number of passes decrease commands; low-temperature regions force speed reduction and frequency / amplitude increase; the rule base supports online updates or manual overwriting via a human-computer interaction interface.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This application systematically addresses the core limitations of existing technologies by constructing a dynamic coupling mechanism between segregation and compaction states. It generates a segregation distribution heatmap with geographic coordinates using machine vision, combines it with multi-parameter monitoring data from the road roller for spatiotemporal matching, and calculates optimal operating parameters through a dynamic coupling model to form a closed-loop control. This achieves, for the first time, a deep integration of segregation information and the compaction process, enabling the road roller to proactively adjust its operating strategy for areas with different degrees of segregation, fundamentally avoiding the problems of under-compaction in segregation zones and over-compaction in uniform zones. Building upon this foundation, RTK / GNSS positioning technology is used to accurately map segregation levels to a geographic coordinate system, providing a high-precision spatial reference for the dynamic coupling model. Through the fusion analysis of vibration waveforms and temperature data, a dataset encompassing dimensions such as location, compaction degree, and number of passes is constructed, enhancing the comprehensiveness of compaction state assessment. Energy demand weighting and temperature attenuation coefficient correction mechanisms enable the dynamic coupling model to accurately calculate the compaction energy gap in different regions. Wireless command transmission and execution mechanisms ensure real-time parameter adjustment, allowing the roller to complete parameter matching before entering the target area. Closed-loop feedback logic achieves iterative parameter optimization through continuous compaction degree deviation assessment. Differentiated strategies develop specific control rules for severely segregated and low-temperature areas, further improving adaptability to complex working conditions. The system, through the collaborative efforts of subsystems for segregation sensing, compaction monitoring, central decision-making, and execution control, constructs a complete technical implementation framework. The real-time data interaction capabilities of the central processing subsystem and the rule base design of the dynamic coupling model ensure the engineering practicality of the technical solution, ultimately achieving comprehensive optimization of asphalt pavement compaction quality stability, construction efficiency, and energy consumption costs. Attached Figure Description

[0016] Figure 1 This is a schematic flowchart of an embodiment of an intelligent compaction control method for asphalt pavement according to the present invention.

[0017] Figure 2 This is a schematic diagram of the module structure of an embodiment of an intelligent asphalt pavement compaction control system according to the present invention. Detailed Implementation

[0018] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other.

[0019] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0020] Example 1

[0021] like Figure 1 As shown in the figure, this embodiment focuses on the construction scenario of the asphalt lower layer of highways and elaborates on the specific execution process of the intelligent rolling control method for asphalt pavement.

[0022] In the real-time perception of segregation distribution, the deployment and debugging of machine vision equipment must be completed first. The image acquisition device is fixed to the center of the crossbeam at the rear of the paver using an adjustable rigid bracket. The bracket height is adjusted according to the height of the paver's screed, ensuring the lens axis forms a 15°-20° angle with the paving surface, covering the entire paving width without blind spots. If the paving width exceeds the coverage of a single camera, multiple cameras can be installed at intervals along the crossbeam, with adjacent cameras overlapping their acquisition areas by 10%-15% to avoid missed images. The choice of camera type should be considered in conjunction with the paving speed: when the paver speed is ≥2m / min, a linear scan camera is selected, as it has continuous scanning capability and can avoid image fragmentation caused by excessive speed; when the paving speed is <2m / min or the construction section is a complex terrain such as curves or slopes, an area scan camera is selected, using multi-frame image stitching to eliminate edge distortion. The camera lens needs to be fitted with a dirt-proof protective cover. The surface of the protective cover is coated with a hydrophobic and oil-resistant coating to reduce the splashing and adhesion of asphalt mixture. It is also equipped with a small blowing device (driven by compressed air) that automatically blows the lens every 5 minutes to ensure the lens is clean.

[0023] The construction of a convolutional neural network involves four stages: sample collection, annotation, training, and optimization. Sample collection covers different construction environments (sunny days, cloudy days, nighttime supplemental lighting) and different aggregate states (normal, mild segregation, moderate segregation, severe segregation), collecting no fewer than 10,000 valid images. Image resolution is set according to camera type to ensure clear representation of aggregate distribution details. During annotation, a combination of manual and machine-assisted methods is used to classify segregation levels based on characteristics such as aggregate particle size uniformity and surface porosity: mild segregation is characterized by small differences in local aggregate particle size and porosity meeting design requirements; moderate segregation is characterized by aggregate agglomeration or dispersion in some areas, with porosity exceeding design tolerances; severe segregation is characterized by large areas of uneven aggregate particle size, with obvious concentration of coarse aggregate or accumulation of fine aggregate, and severely excessive porosity. After annotation, the samples are divided into training, validation, and test sets in a 7:2:1 ratio. A convolutional neural network (such as the ResNet series or MobileNet series) is used for training. Data augmentation techniques (image rotation, brightness adjustment, noise addition) are introduced during training to improve the model's adaptability to different lighting and dust environments. After model training, on-site debugging is required in the test section. Model parameters are optimized based on actual recognition results to ensure consistency between segregation level recognition results and manual detection results, meeting construction requirements, and that the processing time of a single image does not affect real-time performance—that is, the interval from image acquisition to outputting the segregation level does not exceed the time required for the paver to advance 0.5m. The convolutional neural network can be the ResNet series (ResNet-50, ResNet-101) or the MobileNet series (MobileNetV3-Large). These algorithms extract local image features (such as aggregate edges and particle size distribution differences) through multi-layer convolutional kernels, making them suitable for recognizing small to medium-sized segregation areas. Model structure: The input layer receives a 224×224 pixel image (single-channel grayscale image to highlight the contrast between aggregates and voids), and the convolutional layer uses a 3×3 convolutional kernel (stride 1, padding 1). Parameter redundancy is reduced through residual connections (ResNet) or depthwise separable convolution (MobileNet). Applicable scenarios: Straight road construction with paving speed ≥ 2m / min (continuous scanning images by line scan camera), and sunny / cloudy days without significant changes in light intensity; Training parameters: The optimizer used is Adam (initial learning rate 1e-4, decaying by 10% every 5 epochs), the loss function is cross-entropy loss (adapted to dissociative multi-class classification tasks), and training is performed until the accuracy on the validation set is ≥97%.

[0024] The RTK / GNSS module and camera need to be timestamped to ensure that each frame of image matches the corresponding geographic coordinates. The module is installed next to the camera bracket to avoid signal obstruction by metal structures. Its positioning data is transmitted in real time to the image processing unit via a data cable. The processing unit converts the image pixel coordinates into actual geographic coordinates: first, the mapping relationship between camera pixels and the actual road surface is determined through on-site calibration; then, combined with the planar coordinates (latitude and longitude) and elevation data provided by RTK / GNSS, the segregation areas in each frame of image are marked to their corresponding geographic locations. Subsequently, an interpolation algorithm is used to generate a continuous segregation distribution heat map from the discrete segregation level data. The heat map is based on the design coordinate system of the construction section and is divided into fixed grids (the grid size is adapted to the width of the roller's steel wheel, usually 2m×2m). Each grid is marked with the corresponding segregation level and distinguished by different colors (e.g., blue for mild, yellow for moderate, and red for severe). The heat map update frequency is consistent with the camera acquisition frequency to ensure that the segregation status of the latest paved area is reflected in real time.

[0025] Real-time monitoring of compaction status requires the integration of sensors and data synchronization on the road roller. The positioning module is installed in the center of the top of the road roller cab, secured with a metal bracket. A shock-absorbing pad is added to the bottom of the bracket to reduce the impact of road roller vibration on the module. The module is an industrial-grade product supporting multiple satellite systems (BeiDou, GPS), ensuring stable signal reception even in complex construction environments such as those with tall buildings or trees nearby, and meeting construction accuracy requirements. The temperature sensor is installed on the side bracket of the front steel wheel of the road roller. It is a non-contact infrared sensor to avoid direct contact with the high-temperature asphalt mixture and potential damage. The sensor's detection direction is perpendicular to the paved surface, and the detection distance is controlled at 15cm-20cm to ensure that the surface temperature of the mixture before compaction is measured. The acquisition frequency is the same as that of the positioning module (1 time / second), and the data is transmitted in real time to the road roller's data acquisition unit.

[0026] Roller Steel Drum Parameter Determination and Grid Matching Logic: This embodiment uses a double steel drum roller (suitable for the compaction requirements of asphalt sub-layers on highways), and its core steel drum parameters are as follows: Steel wheel width: 2.0m (directly matches the side length of the 2m×2m fixed grid, ensuring that a single compaction can cover the width of a single grid, avoiding statistical deviations in the number of passes caused by compaction across grids); Steel wheel diameter: 1.6m (to meet the compaction depth requirement of 10-12cm for the asphalt base layer); Steel wheel edge chamfer: 5mm radius (to avoid shear damage to the mixture edge during compaction and prevent increased segregation). The design basis for grid division and steel wheel matching is: 2m × 2m grid side length = steel wheel width × (1 - compaction overlap rate), where the compaction overlap rate is set to 10% (a standard industry value, meeting specifications). Therefore, the actual effective compaction width of a single steel wheel is 2.0m × 0.9 = 1.8m. A grid side length slightly larger than the effective compaction width ensures seamless grid coverage between adjacent compaction tracks, while avoiding errors in pass counts caused by covering multiple grids in a single compaction (e.g., the criterion of ≥90% grid area coverage can be accurately applied).

[0027] The vibration sensor is bolted to the front steel wheel bearing housing of the road roller. A piezoelectric accelerometer is selected, its range adapted to the road roller's vibration parameters, capable of capturing the complete acceleration waveform of the steel wheel vibration, including the forward peak, reverse peak, and vibration period. The sampling frequency is set to 1000Hz to ensure no vibration details are missed. The speed sensor is connected to the rear steel wheel axle of the road roller via a coupling. The travel speed is calculated by correlating the axle rotation speed with the steel wheel diameter. During installation, it is necessary to ensure that the coupling is coaxial with the axle to avoid speed measurement errors caused by eccentricity. The sensor output signal is processed by the signal conditioning module and then transmitted to the data acquisition unit.

[0028] Real-time compaction calculation requires establishing a correlation model through field tests. In the test section before construction, test points are selected at 20m intervals. The actual compaction degree at each test point is obtained using core drilling (according to the highway engineering quality inspection and evaluation standards). Simultaneously, the vibration parameters (peak acceleration, vibration frequency), temperature data, and number of compaction passes are recorded for each test point. This data is imported into data analysis software, and a multivariate regression analysis method is used to establish a correlation model between compaction degree and vibration parameters, temperature, and number of compaction passes. The model must consider the impact of temperature on compaction efficiency; that is, as the temperature decreases, the rate of increase in compaction degree decreases under the same vibration parameters, therefore a temperature correction term needs to be introduced. After the model is established, it needs to be verified in the test section. Additional test points are selected, and the compaction degree calculated by the model is compared with the measured value by core drilling to ensure that the deviation is within the allowable range (≤1.5%). If the deviation exceeds the standard, the model parameters are readjusted until the requirements are met.

[0029] The compaction degree calculation model is constructed through a process of "test section calibration - algorithm training - verification and optimization" to ensure reproducibility by those skilled in the art. The specific steps are as follows: Step 1: Data Acquisition for the Test Section. A test section (200m long, 12m wide) with the same as-mixed asphalt (AC-25C type) and the same compaction standard (target compaction degree ≥96%) as in this embodiment was selected. Five temperature gradients (120℃, 130℃, 140℃, 150℃, 160℃, covering the effective compaction temperature range of the asphalt mixture), four segregation grades (uniform, light, medium, heavy, classified according to the segregation identification standard in this embodiment), and six compaction passes (1-6 passes, covering the design pass range) were set. Each test was repeated three times. Vibration waveform data (peak acceleration, vibration frequency), temperature data, and corresponding core drilling measured compaction degree values ​​were collected under each working condition (tested according to Article 8.4.2 of the "Highway Engineering Quality Inspection and Evaluation Standard" JTG F80 / 1-2017), resulting in 360 sets of valid data.

[0030] Step 2: Feature Screening and Preprocessing. Screen feature parameters (peak vibration acceleration, vibration frequency, temperature, number of compaction passes) with a correlation ≥ 0.8 with compaction degree, perform normalization processing (eliminate dimensional differences), and use the Laida criterion to remove outlier data (data points with deviation > 3σ) to ensure data validity.

[0031] Step 3: The model algorithm training adopts a hybrid algorithm of "multiple linear regression + BP neural network": First, a preliminary relationship is established through multiple linear regression (expression: K=a1A+a2f+a3T+a4N+b, where K is the compaction degree, A is the peak acceleration, f is the vibration frequency, T is the temperature, and N is the number of passes). Then, the linear result is used as the input of the BP neural network (4 neurons in the input layer, 8 neurons in the hidden layer, and 1 neuron in the output layer). 70% of the data is used for training, 20% for validation, and 10% for testing. The iteration is repeated 500 times until the root mean square error (RMSE) of the prediction is ≤0.5%.

[0032] Step 4: Calibration and Storage. Compare the measured values ​​from the core drilling method with the predicted values ​​from the model, adjust the neural network parameters to a prediction accuracy of ≥98%, and store the final regression coefficients, neural network weights, etc., as "test section calibration parameters" in the compaction calculation module for real-time calculation and retrieval.

[0033] The number of compaction passes is counted using a grid counting method. The construction area is divided into fixed 2m×2m grids (consistent with the grid of the segregation heat map). The data acquisition unit records the roller's trajectory in real time (coordinate sequence provided by the positioning module). When the trajectory completely covers a grid (coverage area ≥ 90% of the grid area), it is determined that the grid has completed one effective compaction pass, and the pass counter is incremented by 1. If the trajectory only partially covers the grid, it is not counted as an effective pass, avoiding counting errors caused by the roller turning or moving backward. At the same time, the data acquisition unit summarizes the real-time collected location, compaction degree, temperature, number of compaction passes, travel speed, and vibration parameters (frequency, amplitude) to form a multi-dimensional compaction state dataset. The dataset is sorted by timestamp, with one record generated every second, and uploaded to the central processing system in real time via a wireless transmission module.

[0034] To cover different construction scenarios and ensure the compaction quality of the entire road section, new grid divisions and compaction rules have been added for straight sections and curves: For straight-line compaction (curvature radius ≥ 500m), the 2m × 2m grid remains unchanged. The roller adopts a "longitudinal reciprocating" compaction method: it travels parallel to the centerline of the road, and moves laterally by 1.8m (effective compaction width of the steel wheel) after each round trip, advancing from one side of the road to the other to avoid misalignment of adjacent compaction zones; GNSS positioning is used to ensure that the deviation between each trajectory and the grid boundary is ≤ 0.1m.

[0035] For curve compaction (curvature radius < 500m), due to the small inner radius and large outer radius of the curve, the original 2m×2m grid easily led to insufficient compaction on the inner side and over-compaction on the outer side. Therefore, the grid was reduced to 1m×1m (to adapt to curvature changes), and the compaction sequence was adjusted to "spiral advancement from the inner side to the outer side": first, the first round of compaction was completed with the inner edge of the curve as the reference, and then the second round was carried out by shifting outward by 1.8m (effective width) until the entire width of the curve was covered; the compaction overlap rate was increased to 15%, and the trajectory was corrected in real time through GNSS to ensure that the curvature of each round was consistent with the design curvature.

[0036] In the spatiotemporal matching process between the segregation distribution heat map and real-time compaction values, it is necessary to first unify the coordinates and time reference. Both the segregation heat map and the compaction data adopt the design coordinate system of the construction section to ensure the consistency of spatial location. In terms of time, the system records the timestamp of segregation detection (the time when the camera acquires the image) and the timestamp of compaction data (the time when the sensor acquires the data). Considering the time difference between the paver and the roller (usually 5-10 minutes), a time compensation algorithm is used to adjust—that is, when the roller is about to enter a certain area, the system calls the segregation heat map data generated when the paver passes through that area, and combines it with the existing compaction data of that area (such as the number of passes already rolled, the current compaction degree) to achieve accurate matching of "segregation state + compaction state". During the matching process, if the segregation heat map data of a certain area is missing (such as due to temporary camera failure), the system automatically uses the segregation data of adjacent areas for interpolation to supplement it and triggers an early warning to prompt the operator to check the equipment; if the compaction data is missing, the parameter calculation of that area is paused and restarted after the data is recovered.

[0037] The operation of the dynamic coupling model and the calculation of optimal operating parameters are the core components. The central processing system first extracts key information about the area to be compacted by the roller: it obtains the segregation level of the area from the segregation thermogram, and obtains the current temperature, number of compaction passes, current compaction degree, and target compaction degree (set according to design requirements, such as ≥96%) from the compaction dataset. The model first sets the energy demand weights according to the segregation level: the energy demand weight for severely segregated areas is 1.2-1.5, for slightly segregated areas it is 0.8-1.0, and for uniform areas it is 0.6-0.8. The weight values ​​are determined through verification on test sections to ensure that areas with more severe segregation can obtain higher compaction energy. Next, the attenuation coefficient is adjusted according to temperature: when the temperature is between 140℃ and 160℃ (the optimal compaction temperature range for asphalt mixtures), the attenuation coefficient is 1.0; for every 10℃ decrease in temperature, the attenuation coefficient increases by 0.1-0.2 to compensate for the decrease in compaction efficiency caused by the temperature drop; when the temperature is below 100℃, the attenuation coefficient no longer increases linearly, but is set to an upper limit value to avoid excessive reliance on high-energy compaction leading to aggregate breakage.

[0038] Subsequently, the model calculates the energy gap: based on the difference between the current compaction degree and the target compaction degree, combined with the energy demand weight and temperature decay coefficient, the additional compaction energy required for the area to reach the target compaction degree is determined. Then, the optimal combination of operating parameters is calculated based on the energy gap: driving speed is negatively correlated with compaction energy; the larger the energy gap, the lower the driving speed (extending rolling time and increasing the number of compaction passes per unit area); vibration frequency and amplitude jointly determine the compaction energy per unit time. When the energy gap is large, the vibration frequency (increasing the number of vibrations) or amplitude (increasing the energy per vibration pass), or both, can be adjusted simultaneously; at the same time, the number of compaction passes is considered—if the number of compaction passes is close to the design limit (e.g., the design number is 6 passes, and 5 passes have been completed), energy is supplemented first by adjusting vibration parameters to avoid exceeding the pass limit and causing over-compaction.

[0039] After the parameters are calculated, a rationality check is required: the driving speed must be within the rated speed range of the road roller (usually 2-6 km / h), the vibration frequency must match the vibration system characteristics of the road roller (usually 25-50 Hz), and the amplitude must meet the compaction requirements of the mixture (usually 0.3-0.8 mm). If the calculation results exceed the reasonable range, the model will automatically adjust to the closest reasonable value and mark the reason for the adjustment on the human-machine interface.

[0040] The optimal combination of operating parameters is transmitted to the roller control system via wireless communication. The communication uses industrial-grade wireless transmission protocols (such as LoRa and 5G industrial private networks) to ensure stability and real-time performance in complex electromagnetic environments at the construction site (such as interference from other construction equipment). The central processing system encapsulates the recommended travel speed, vibration frequency, and vibration amplitude into standardized control commands. These commands include the geographical coordinates of the effective area (e.g., a 20m × 12m construction section), ensuring the roller only performs parameter adjustments within the designated area. Upon receiving the command, the roller's onboard controller first performs data verification: checking if the command format is correct and if the parameters are within the equipment's allowable range. If the verification fails, it sends an error message to the central system, requesting a retransmission. After successful verification, the controller parses the command parameters and drives the corresponding actuators to adjust: by adjusting the engine throttle and gearbox gears to control the travel system, it achieves smooth speed adjustments, avoiding uneven compaction caused by sudden speed changes; by controlling the hydraulic pump flow and pressure of the vibration hydraulic system, it adjusts the vibration frequency and amplitude. During the hydraulic system adjustment process, pressure stability must be maintained to prevent vibration parameter fluctuations from affecting the compaction effect. Parameter adjustments must be completed before the roller enters the effective zone. If the roller has already entered the effective zone, the current parameters should be maintained to complete the compaction of the zone. The new parameters should be executed in the next compaction cycle to ensure compaction continuity.

[0041] The closed-loop feedback and secondary adjustment process must ensure that compaction quality consistently meets standards. After the roller completes compaction of a certain area according to the adjusted parameters, the data acquisition unit immediately re-collects vibration and temperature data for that area, derives a new compaction value through the compaction degree calculation model, and uploads it to the central processing system. The central system compares the new compaction degree value with the target compaction degree and calculates the deviation: if the deviation is ≤1% (design allowable deviation), the compaction of that area is deemed satisfactory and no adjustment is needed; if the deviation is >1%, the secondary adjustment process is triggered—the updated information for that area is re-extracted (including the new compaction degree, the increased number of compaction passes, and the current temperature; if there is no significant change in segregation status, the original segregation level is retained), and the updated information is input into the dynamic coupling model to recalculate the optimal parameters. During the secondary adjustment, the model should prioritize adjusting vibration parameters (such as appropriately increasing the amplitude). If the deviation still does not meet the standards after adjustment, then consider reducing the travel speed or increasing the number of compaction passes to avoid efficiency loss due to over-adjustment. This process is repeated until the compaction deviation in the area stabilizes within the allowable range, forming a closed-loop control of "parameter issuance-execution-detection-adjustment". If an area still fails to meet the standard after three adjustments, the system triggers an early warning, prompting the operator to conduct on-site inspections (such as whether the mixture temperature is too low or whether the sensor is faulty). The closed-loop process is restarted after the problem is resolved.

[0042] Furthermore, the differentiated strategy of the dynamic coupling model needs to be implemented throughout the entire construction process: For areas with severe segregation (such as areas with concentrated coarse aggregates due to uneven aggregate distribution during paving), in addition to reducing driving speed and increasing vibration parameters, the upper limit of the number of compaction passes can be increased by 1-2 passes to ensure sufficient aggregate embedment; For areas with mild or uniform segregation, under the premise of meeting the target compaction degree, the driving speed should be appropriately increased (such as from 3km / h to 4-5km / h), and the vibration frequency or amplitude should be reduced to reduce equipment energy consumption and the risk of aggregate breakage; For low-temperature areas (such as areas where the temperature is below 120℃ in the later stages of construction, or areas where heat dissipation is faster at the edge of the road section), regardless of the degree of segregation, the driving speed should be forcibly reduced (such as to 2-3km / h), while the vibration frequency or amplitude should be increased to compensate for the reduction in compaction efficiency caused by temperature and avoid insufficient compaction due to excessively low temperature.

[0043] Example 2 like Figure 2 As shown, this embodiment is based on the control method described in Embodiment 1, and elaborates on the intelligent rolling control system for asphalt pavement in the same asphalt lower layer construction scenario of a highway.

[0044] The segregation perception subsystem consists of machine vision equipment, an RTK / GNSS module, an edge computing unit, and a data transmission module. Its core function is to generate a segregation distribution heat map with geographic coordinates. The machine vision equipment includes a camera, lens, protective device, and mounting bracket: the camera is an industrial-grade product with a wide dynamic range (adapting to bright sunlight and nighttime lighting scenarios) and high sensitivity (reducing image noise in dusty environments); the lens is a fixed-focus lens with a focal length determined according to the paving width and installation height to ensure full coverage of the paving width and image resolution that meets recognition requirements; the protective device is made of metal with an internal heat insulation layer (preventing the high temperature of the paver from being conducted to the camera), and a light-transmitting window on the surface (the window glass is made of anti-glare and scratch-resistant material), with a pre-reserved blow-out interface on the side (connecting to the paver's compressed air system); the mounting bracket is made of aluminum alloy, and its height and angle can be adjusted via bolts. The bottom of the bracket is welded and fixed to the crossbeam at the rear of the paver to ensure no shaking during construction.

[0045] The RTK / GNSS module is an industrial-grade module supporting dual-mode positioning of BeiDou-3 and GPS, with centimeter-level positioning accuracy. The module is equipped with an external antenna mounted on the top of the support frame, higher than other parts of the paver to avoid signal obstruction. It connects to the edge computing unit via an RS232 interface, with a positioning data update frequency of 1Hz. The edge computing unit is an embedded industrial computer installed in the paver's cab, containing a built-in segregation identification algorithm module—integrated in software, supporting online upgrades. It can receive image data from the camera and RTK / GNSS positioning data in real time, completing segregation level identification and heat map generation, while also having a data caching function.

[0046] The data transmission module uses an industrial-grade wireless network card, integrated within the edge computing unit. It is responsible for transmitting the generated discrete distribution heat map to the central processing and decision-making subsystem in real time. During transmission, a data compression algorithm is used (to reduce the amount of data and improve the transmission speed), and the data is encrypted to prevent tampering during transmission.

[0047] The compaction monitoring subsystem consists of a positioning module, temperature sensor, vibration sensor, velocity sensor, compaction calculation module, and data transmission module. Its core function is to collect compaction status data and calculate the real-time compaction degree. The positioning module is the same model as the segregation sensing subsystem and is installed on the top of the roller's cab, fixed by a shock-absorbing bracket. The module's data is transmitted to the compaction calculation module via a data cable, while also providing a coordinate reference for recording the roller's trajectory.

[0048] The temperature sensor is a non-contact infrared temperature sensor, which covers the construction temperature range of asphalt mixture (-20℃-250℃) and has the accuracy to meet the construction requirements. The sensor is installed on the side of the front steel wheel of the road roller by a metal bracket. The bracket can be adjusted in height and angle to ensure that the detection point is located on the paving surface directly in front of the steel wheel. The sensor output signal is processed by the signal conditioning module (filtering noise) and then transmitted to the compaction calculation module.

[0049] The vibration sensor is a piezoelectric accelerometer, with a measurement range matched to the vibration system of the road roller (capable of withstanding the maximum vibration acceleration of the road roller). The frequency response range covers the operating frequency of the road roller (20-60Hz). The sensor is fixed to the steel wheel bearing seat with bolts. The mounting surface must be flat and coated with thermal grease (to reduce vibration interference and heat accumulation). The analog signal output by the sensor is converted into a digital signal by an A / D conversion module (16 bits or higher resolution) and then transmitted to the compaction calculation module.

[0050] The speed sensor is an incremental rotary encoder, and the resolution is determined according to the diameter of the roller's steel wheel and the speed range (ensuring speed measurement accuracy ≤0.1km / h). It is connected to the rear steel wheel axle of the roller through a coupling made of elastic material (to compensate for installation deviations between the wheel axle and the sensor). The pulse signal output by the sensor is transmitted to the compaction calculation module, which calculates the travel speed by counting pulses and the wheel axle circumference.

[0051] The compaction calculation module uses an embedded processor (with multi-channel data acquisition capabilities) and is installed in a waterproof control cabinet inside the roller's cab. It incorporates a compaction degree calculation model and a compaction pass statistics algorithm: the compaction degree calculation model is implemented in software, capable of calling stored test section calibration parameters and combining them with real-time acquired vibration and temperature data to calculate the compaction degree; the compaction pass statistics algorithm analyzes trajectory data provided by the positioning module to achieve grid counting. The module also has data storage capabilities (capable of storing at least 7 days of compaction data) and supports data export via USB interface for later analysis.

[0052] The data transmission module is consistent with the segregation sensing subsystem and is integrated into the compaction calculation module. It uploads multi-dimensional compaction state datasets (location, compaction degree, temperature, number of passes, velocity, vibration parameters) to the central processing and decision-making subsystem in real time. The transmission frequency is synchronized with the data acquisition frequency (1 time / second).

[0053] The central processing and decision-making subsystem is the core of the entire system. It consists of an industrial server, dynamic coupling model software, a data fusion module, an instruction generation module, and a communication module. It is installed in the control room at the construction site (ambient temperature 0℃-40℃, equipped with a regulated power supply and cooling system). The industrial server is a high-performance rack-mount server with a multi-core CPU, large-capacity memory, and hard drive storage. It supports concurrent multi-task processing (simultaneously receiving compaction data from multiple road rollers and segregation data from multiple pavers). The server has built-in redundant power supplies to prevent system interruption due to power failure.

[0054] The dynamic coupling model software is installed on the server as an application. Its core function is to achieve spatiotemporal matching of segregation data and compaction data, energy demand calculation, and optimal parameter output. The software has a built-in differentiated control rule library, which contains parameter adjustment logic corresponding to different segregation levels and temperature ranges. It supports online updates through a human-machine interface—operators can modify the weight values, attenuation coefficients, parameter ranges, etc. in the rule library according to the design requirements of different construction projects (such as different mixture types and different compaction standards). The updated rule library takes effect in real time without restarting the software.

[0055] The data fusion module is responsible for receiving heat map data from the segregation sensing subsystem and compaction data from the compaction monitoring subsystem. First, it preprocesses the data: removing outlier data (such as values ​​outside the range due to sensor malfunction), completing missing data (using interpolation or adjacent data as substitutes), and standardizing the data format (converting it according to a preset standardized format). Then, it performs spatiotemporal matching: based on a unified coordinate system and timestamps, it associates the compaction data with the corresponding segregation heat map grid, generating a "grid-segregation level-compaction state" association data table to provide input for the dynamic coupling model.

[0056] The instruction generation module generates standardized control instructions based on the optimal operating parameters output by the dynamic coupling model. These instructions include information such as the roller number (distinguishing multiple rollers), the coordinates of the effective area, the travel speed, vibration frequency, vibration amplitude, and the instruction activation time. The instructions are in XML or JSON format for easy parsing by the roller's onboard controller. The module also features instruction caching and retransmission capabilities: if instruction transmission fails (e.g., due to a wireless network interruption), the instructions are automatically cached and retransmitted once the network is restored, ensuring no instructions are lost.

[0057] The communication module includes a wireless gateway and a firewall. The wireless gateway supports multi-protocol access (adapting to the transmission protocols of the segregation sensing subsystem, compaction monitoring subsystem, and road roller control subsystem), and the firewall is used to filter unauthorized access to ensure system data security. The module monitors the communication status of each subsystem in real time. If the communication of a subsystem is interrupted, an early warning message is immediately displayed on the human-machine interface, and the interruption time and cause are recorded for later troubleshooting.

[0058] The road roller control subsystem consists of an onboard controller, an actuator drive module, and a status feedback module, integrated into the existing control system of the road roller to achieve automatic parameter adjustment and status feedback. The onboard controller uses an industrial-grade PLC (Programmable Logic Controller), which has high reliability and vibration resistance. It is connected to the communication module of the central processing and decision-making subsystem via a data cable to receive and parse control commands. During the parsing process, the controller compares the parameters in the command with the road roller's own rated parameters (such as maximum speed and maximum amplitude). If they exceed the range, it automatically adjusts to the rated values ​​and feeds back the adjustment results to the central system.

[0059] The actuator drive module includes a travel drive unit and a vibration drive unit: the travel drive unit is connected to the roller's engine throttle and gearbox control module, and adjusts the throttle opening and gearbox gear by outputting PWM signals to achieve smooth adjustment of travel speed. During the adjustment process, the speed change rate is controlled within 0.5 km / h / s to avoid impact; the vibration drive unit is connected to the solenoid valve of the roller's vibration hydraulic system. By controlling the on / off time and frequency of the solenoid valve, the flow and pressure of the hydraulic pump are adjusted, thereby changing the vibration frequency and amplitude. During the adjustment process, the hydraulic system pressure is kept stable (fluctuation range ≤0.5MPa) to ensure smooth transition of vibration parameters.

[0060] The status feedback module works in conjunction with the actuator drive module to collect real-time values ​​of driving speed, vibration frequency, and vibration amplitude, and transmit them to the vehicle controller. The controller compares the actual values ​​with the target values ​​in the command. If the deviation exceeds the allowable range (e.g., speed deviation > 0.2 km / h), it automatically adjusts the output of the drive module until the actual value matches the target value. At the same time, the status feedback module synchronously uploads the actual parameters and the collected compaction data to the central processing and decision-making subsystem, providing a basis for closed-loop feedback.

[0061] The human-machine interface consists of an industrial monitor, a keyboard, and a mouse, installed in the control room and connected to the server of the central processing and decision-making subsystem to realize construction status monitoring, parameter setting, and early warning viewing. The industrial monitor uses a touch screen of 19 inches or larger with a resolution of ≥1920×1080 and supports multi-window display: the main window displays a real-time heat map of segregation distribution and a superimposed map of compaction distribution of the construction section, with different colors representing different segregation levels and compaction ranges. Details of any area can be viewed by zooming and panning; the sub-windows display the real-time location, operating parameters (travel speed, vibration frequency, amplitude), number of passes, current compaction degree, communication status of each subsystem, and sensor operating status of each roller.

[0062] The interface supports data query functionality: operators can query segregation data, compaction data, and control command records by time range (such as a specific day or construction section). Query results can be exported to Excel or PDF formats for use in construction log writing and quality traceability. Simultaneously, the interface has an early warning function: when abnormal situations such as sensor failure, communication interruption, or excessive compaction deviation occur, a red warning window pops up on the interface, accompanied by an audible alert, and displays the location, type, and handling suggestions of the abnormality. The warning information is automatically stored on the server for later analysis.

[0063] In addition, the interface provides a rule base update entry: after the operator verifies the password, they can enter the rule base editing interface to modify the parameter adjustment logic corresponding to different segregation levels and temperature ranges. During the editing process, the interface displays the parameter comparison before and after the modification in real time. After confirming that there are no errors, clicking "Save" will take effect, and a modification record will be automatically generated to ensure traceability.

[0064] After system installation, collaborative debugging is required to ensure smooth data interaction and normal functional integration between subsystems. First, perform individual subsystem debugging: Start the segregation sensing subsystem, compaction monitoring subsystem, central processing and decision-making subsystem, and roller control subsystem respectively. Check if each device is working properly, whether data acquisition is accurate (e.g., comparing segregation identification results with manual inspection, comparing calculated compaction values ​​with core drilling measurements), and whether command issuance is smooth (e.g., the roller executes speed adjustment commands issued by the central system).

[0065] After the individual subsystems are successfully debugged, system-wide integration testing is conducted: all subsystems are started, simulating the actual construction process—the paver begins paving, the segregation sensing subsystem generates and uploads a heat map; the roller begins compaction, the compaction monitoring subsystem collects and uploads data; the central system completes spatiotemporal matching and parameter calculation, and issues commands; the roller executes the commands and provides feedback on the actual parameters; the central system makes secondary adjustments based on the feedback. During the integration testing, the key checks are the spatiotemporal matching accuracy (e.g., the deviation between the roller position and the segregation heat map grid ≤ 0.5m), the parameter adjustment response time from the central system issuing the command to the roller completing the execution ≤ 5 seconds, and the compaction deviation of the closed-loop feedback effect remaining stable within the allowable range.

[0066] After successful commissioning, a field trial run was conducted: a 100m long test section was selected, and the system was run according to the actual construction process. All data during the trial run, including segregation data, compaction data, control commands, and early warning information, were recorded. After the trial run, the test section was inspected for quality by core drilling to test the compaction degree and for visual inspection to check the segregation repair status. If all test results met the design requirements, the system could be officially put into use. If problems were found, the subsystem parameters were adjusted accordingly, such as optimizing the dynamic coupling model weights and calibrating the sensors. The trial run was repeated until the requirements were met.

[0067] As is known from common technical knowledge, this invention can be implemented through other embodiments that do not depart from its spirit or essential characteristics. Therefore, the disclosed embodiments described above are merely illustrative in all respects and are not the only ones. All modifications within the scope of this invention or its equivalents are included in this invention.

Claims

1. A method for intelligent compaction control of asphalt pavement, characterized in that, include: Machine vision equipment is deployed to acquire images of the surface of freshly laid, uncompacted asphalt mixture. A pre-set convolutional neural network identifies the degree of segregation and generates a segregation distribution heatmap with geographic coordinates. A positioning module, temperature sensor, vibration sensor, and speed sensor are integrated on the roller to collect data on location, number of compaction passes, speed, vibration parameters, and temperature, and to calculate real-time compaction values. The segregation distribution heatmap is spatiotemporally matched with the real-time compaction values. Based on the degree of segregation, temperature, number of compaction passes, and current compaction value of the area to be compacted by the roller, an optimal combination of operating parameters is calculated using a dynamic coupling model. This optimal combination of operating parameters is wirelessly transmitted to the roller control system, which automatically adjusts the travel speed and vibration parameters. After the roller performs the adjustment, a new compaction value is collected and fed back to the dynamic coupling model for evaluation and to trigger secondary adjustments, forming a closed-loop control system.

2. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, The process of generating a segregation distribution heat map with geographic coordinates includes: acquiring images of the asphalt mixture surface using a linear or area array camera behind the paver, inputting the images into a convolutional neural network to output the segregation level of each area, and mapping the segregation level to a geographic coordinate system using RTK / GNSS positioning data to generate a spatial heat map covering the construction surface with the degree of segregation marked at each point, which serves as the spatial input basis for the dynamic coupling model.

3. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, The calculation of the real-time compaction value includes: the roller recording its own position through the GNSS positioning module, collecting vibration waveforms through the vibration sensor, collecting the temperature of the mixture through the temperature sensor, inputting the vibration waveforms into the intelligent compaction algorithm module, calculating the compaction value at the current position by combining the temperature and position, and updating the number of rolling passes for each grid according to the number of times the trajectory coincides with the grid, forming a multi-dimensional compaction state dataset containing position, compaction degree, temperature, number of passes, speed and vibration parameters.

4. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, The process of calculating the optimal combination of operating parameters through a dynamic coupling model includes: the central system matching the roller trajectory with the heat map, extracting the segregation degree, temperature, number of compaction passes, and current compaction degree of the target area, and inputting them into the dynamic coupling model; the dynamic coupling model sets the energy demand weight based on the segregation degree, corrects the attenuation coefficient based on the temperature, calculates the energy gap based on the number of passes and compaction degree, and outputs recommended driving speed, vibration frequency, and vibration amplitude values ​​to form a complete parameter combination for roller adjustment.

5. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, The automatic adjustment of driving speed and vibration parameters includes: the central system encapsulates the recommended driving speed value, vibration frequency value and vibration amplitude value into control commands, and sends them to the roller's on-board controller via wireless communication; after the controller parses the commands, it drives the walking system to adjust the speed and drives the vibration hydraulic system to adjust the frequency and amplitude, so that the roller performs parameter matching before entering the target area or during rolling, in order to avoid over-compaction in the uniform area or under-compaction in the segregation area.

6. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, The feedback and triggering of secondary adjustments include: after the roller compacts according to the adjusted parameters, it re-collects vibration and temperature data, calculates a new compaction value, and uploads it; the central system compares the new compaction value with the target threshold, and if the deviation exceeds the limit, it calls the dynamic coupling model again, combines the updated segregation degree, temperature, number of passes, and compaction value, calculates the optimization parameters for the next stage, and issues instructions again until the compaction degree is stably met, thus realizing closed-loop iterative control.

7. The intelligent rolling control method for asphalt pavement according to claim 1, characterized in that, When calculating parameters using the dynamic coupling model, a differentiated strategy is implemented: when the area is identified as a severely segregated region, the driving speed is reduced, the vibration frequency or amplitude is increased, or the upper limit of the number of compaction passes is increased; when the area is identified as a mild or uniform region, the driving speed is increased, the vibration frequency or amplitude is reduced, or the upper limit of the number of compaction passes is decreased; when the area is identified as a low-temperature region with a temperature below 120℃ in the later stage of construction, the driving speed is forcibly reduced and the vibration frequency or amplitude is increased to compensate for the reduction in compaction efficiency caused by temperature.

8. An intelligent rolling control system for asphalt pavement, based on the intelligent rolling control method for asphalt pavement according to any one of claims 1-7, characterized in that, include: The segregation perception subsystem, including machine vision equipment and segregation recognition algorithm module, is used to generate a heat map of segregation distribution with geographic coordinates; The compaction monitoring subsystem includes a positioning module, a temperature sensor, a vibration sensor, a velocity sensor, and a compaction calculation module, which is used to collect compaction status data. The central processing and decision-making subsystem runs a dynamically coupled model to match heat maps and compaction data and calculate optimal parameters; The road roller control subsystem receives commands and adjusts the travel speed and vibration parameters; the human-machine interface is used to display heat maps, compaction maps, commands, and early warning information.

9. The intelligent asphalt pavement compaction control system according to claim 8, characterized in that, The central processing and decision-making subsystem performs the following actions: matching the roller trajectory with the heat map, extracting the segregation degree, temperature, number of passes, and compaction degree value of the target area, and inputting it into the dynamic coupling model to calculate the recommended speed, frequency, and amplitude values; receiving feedback compaction degree values, and if the standard is not met, recalculating and issuing new instructions to form a closed loop; the central processing and decision-making subsystem interacts with the segregation sensing subsystem, compaction monitoring subsystem, and roller control subsystem in real time via a wireless network to ensure that instructions are synchronized with the field.

10. The intelligent asphalt pavement compaction control system according to claim 8, characterized in that, The dynamic coupling model embeds a differentiated control rule base: in areas of severe segregation, it triggers commands to reduce speed, increase frequency / amplitude, or increase the number of passes; in areas of mild or uniform segregation, it triggers commands to increase speed, reduce frequency / amplitude, or decrease the number of passes; in low-temperature areas where the temperature is below 120℃ in the later stages of construction, it forces a reduction in speed and an increase in frequency / amplitude; the rule base supports online updates or manual overwriting through a human-computer interaction interface.

Images