A multi-dimensional intelligent detection system and method for highway asphalt mixture performance

Abstract

The application provides a highway asphalt mixture performance multi-dimensional intelligent detection system and method, and belongs to the technical field of road engineering. The system comprises: a simulation detection platform having a filling bin capable of replacing different simulation fillers and a modular detection plate for bearing a test piece to simulate different roadbeds; an environment simulation and loading module for applying a comprehensive environmental load to the test piece; a multi-source information acquisition module for acquiring a multi-dimensional image sequence of the test piece in real time during loading; and an intelligent analysis and processing module for analyzing performance according to the image sequence. The method comprises: configuring simulation fillers and test pieces; applying a time-controllable comprehensive environmental load; acquiring and correlating a multi-dimensional image sequence; extracting deformation, seepage and spalling features to form a feature set through a feature analysis strategy; and obtaining a comprehensive performance based on the feature set through a performance evaluation strategy. The application realizes intelligent detection and evaluation of multiple working conditions in parallel and multi-dimensional synchronization in a single test.

Description

Technical Field

[0001] This invention relates to the field of road engineering technology, and specifically to a multi-dimensional intelligent testing system and method for the performance of asphalt mixtures for highways. Background Technology

[0002] As my country's highway network extends into areas with complex geological conditions and traffic loads continue to increase, higher demands are being placed on the durability and service reliability of asphalt pavement materials. Performance evaluation of asphalt mixtures is a crucial step in ensuring pavement quality. Its core lies in accurately simulating the complex mechanical and climatic environmental effects experienced by materials in actual roads and quantitatively assessing their response under various failure modes. Traditional laboratory testing methods, such as rutting tests, immersion Marshall tests, and freeze-thaw splitting tests, constitute the current mainstream evaluation system, aiming to evaluate the high-temperature stability, water stability, and freeze-thaw resistance of materials, respectively.

[0003] However, existing technologies have significant limitations. First, each test is usually conducted independently, simulating relatively simple loading conditions and environmental factors. This fails to reproduce the comprehensive stress state resulting from the long-term coupled effects of multiple factors such as traffic load, precipitation infiltration, and temperature changes on a real road surface, leading to discrepancies between indoor evaluation results and actual road performance. Second, traditional tests primarily provide single, "endpoint" performance indicators, lacking simultaneous, dynamic observation and quantitative analysis of microscopic and macroscopic phenomena such as deformation development, moisture migration, and aggregate spalling during the damage process, making it difficult to reveal the intrinsic mechanisms of performance degradation. Furthermore, existing methods struggle to conduct convenient and efficient differentiated simulations and comparative tests in the laboratory to address the differences in subgrade support conditions, resulting in a lack of scientific basis for material selection and mix design tailored to specific road conditions. Summary of the Invention

[0004] This invention provides a multi-dimensional intelligent testing system and method for the performance of highway asphalt mixtures, which solves the problems of single testing dimensions, unrealistic working condition simulation, and inability to simultaneously evaluate the comprehensive performance of materials under multi-dimensional load coupling in a single test in the existing technology.

[0005] To achieve the above objectives, one embodiment of the present invention provides a multi-dimensional intelligent testing system for the performance of highway asphalt mixtures, comprising: a simulation testing platform including several filling chambers for filling with simulated fillers of different physical properties to simulate different roadbeds; several modular testing plates that can be spliced ​​together are arranged above the filling chambers for supporting asphalt mixture specimens; an environmental simulation and loading module for applying a controllable comprehensive environmental load to the asphalt mixture specimens placed on the simulation testing platform; a multi-source information acquisition module including image acquisition devices arranged above and to the sides of the simulation testing platform for real-time acquisition of multi-dimensional image sequences of the asphalt mixture specimens on each filling chamber and each simulation testing plate during the application of the comprehensive environmental load; and an intelligent analysis and processing module for performing performance testing on the asphalt mixture specimens based on the multi-dimensional image sequences to obtain the comprehensive performance of the asphalt mixture specimens under different road conditions.

[0006] Furthermore, the comprehensive environmental load includes mechanical stress, permeation stress, and temperature stress; the environmental simulation and loading module is configured to include: a servo loading mechanism for applying the mechanical stress; a spraying mechanism for simulating rainwater to generate the permeation stress; and a temperature control mechanism for adjusting the ambient temperature to generate the temperature stress.

[0007] Furthermore, each filling chamber corresponds to a preset number of modular testing plates, and each modular testing plate has a unique identification code; the multi-dimensional image sequence includes an asphalt image sequence and a water seepage image sequence of the modular testing plates, and the intelligent analysis and processing module is also used to associate the testing plate image sequence corresponding to each simulated testing plate according to the unique identification code.

[0008] On the other hand, a multi-dimensional intelligent detection method for highway asphalt mixture performance is also provided, applied to the aforementioned multi-dimensional intelligent detection system for highway asphalt mixture performance. The method is characterized by the following steps: Step S1: In different filling chambers of the simulated testing platform, simulated fillers with different physical properties are configured, and asphalt mixture specimens are laid on the surface of the testing plates; Step S2: According to preset testing rules, a time-controllable comprehensive environmental load is applied to the asphalt mixture specimens; Step S3: A multi-dimensional image sequence of the asphalt mixture specimens under the comprehensive environmental load is obtained, and each modular testing plate is associated with the corresponding water seepage image sequence in the multi-dimensional image sequence; Step S4: Based on the multi-dimensional image sequence corresponding to each simulated filler, a feature analysis strategy is used to extract the deformation characteristics, surface water diffusion characteristics, and aggregate particle detachment characteristics of the asphalt mixture specimens on each simulated filler to form a feature set; Based on the feature set and combined with a preset performance evaluation strategy, the comprehensive performance of the asphalt mixture specimens under different road conditions is obtained.

[0009] Furthermore, the asphalt image sequence includes a deformation image sequence and a surface state image sequence. The feature analysis strategy includes a deformation analysis sub-strategy, which includes: calculating the vertical displacement field of the asphalt mixture specimen based on the deformation image sequence corresponding to each filler, and extracting the rutting depth and vertical displacement standard deviation from the vertical displacement field; calculating the horizontal displacement field of the surface of the asphalt mixture specimen based on the surface state image sequence corresponding to the same filler, and extracting the maximum horizontal shear displacement and surface roughness evolution rate from the horizontal displacement field; and using the set composed of the rutting depth, vertical displacement standard deviation, maximum horizontal shear displacement, and surface roughness evolution rate as the deformation feature to characterize the deformation performance of the asphalt mixture specimen.

[0010] Furthermore, the feature analysis strategy also includes a seepage analysis sub-strategy, which includes: based on the seepage image sequence corresponding to each filling chamber, using a time-series image segmentation algorithm to identify and track the diffusion contour of water traces in the seepage image sequence, and calculating the rate of change of its diffusion area and perimeter over time; quantifying its surface runoff characteristics and internal infiltration depth by analyzing the diffusion path and final steady-state distribution of the water traces; and determining the seepage characteristics based on the rate of change of diffusion area, surface runoff characteristics, and internal infiltration depth.

[0011] Furthermore, the feature analysis strategy also includes a spalling monitoring sub-strategy, which includes: precisely synchronizing the surface state image sequence with the time sequence of the applied comprehensive environmental load to establish a correspondence between image frames and load values ​​and load types; for image frames corresponding to different load stages, using an adaptive threshold segmentation algorithm to identify aggregate particles detached from the surface of the asphalt mixture specimen; counting the cumulative number of aggregate particles identified under a specific load and calculating the average spalling rate per unit time; and determining the aggregate particle detachment characteristics based on the cumulative number of aggregate particles, the average spalling rate, and their correspondence with the load to characterize the aggregate retention capacity and bond failure critical conditions of the asphalt mixture specimen under different comprehensive environmental loads.

[0012] Furthermore, the performance evaluation strategy includes: for each feature in the feature set, querying a preset benchmark database to obtain the benchmark threshold corresponding to each feature based on the roadbed type simulated by the current simulated infill; comparing the parameter value of each feature with the corresponding benchmark threshold, and calculating the performance score corresponding to each feature according to a preset scoring function; obtaining the dynamic weight coefficient of each feature from a preset weight knowledge base based on the roadbed type simulated by the current simulated infill; and obtaining the comprehensive performance through weighted fusion calculation based on the performance score of each feature and its corresponding dynamic weight coefficient.

[0013] Furthermore, step S4 also includes: determining weak features based on the performance scores corresponding to each feature, determining the target raw material corresponding to the weak feature by combining the preset raw material component-performance mapping relationship; and calculating the adjustment direction and adjustment amount of the target raw material based on the difference between the performance score of the weak feature and the target value.

[0014] This invention provides a multi-dimensional intelligent testing system and method for highway asphalt mixture performance. By constructing a highly integrated physical-digital fusion testing platform in the laboratory, it improves the realism, efficiency, and intelligence of asphalt mixture performance evaluation. Specifically, the system, through a modular testing plate with replaceable filling chambers, achieves convenient and high-throughput simulation of different roadbed conditions for the first time. This allows for parallel comparison of the adaptability of a material under various road conditions, greatly improving the efficiency of research and evaluation. The method, by applying programmable comprehensive environmental loads and utilizing image acquisition and computer vision algorithms, achieves for the first time the simultaneous, dynamic, and visualized quantitative extraction and correlation analysis of the three key dimensions of material performance—deformation, water permeability, and spalling—during a single test. This breaks through the limitations of traditional isolated, static, and "black box" testing, enabling performance evaluation to leap from relying on single endpoint data to intelligent diagnosis and prediction based on the entire process and multi-index coupled response. This provides unprecedented scientific tools and decision-making basis for the design and precise selection of high-performance asphalt mixtures for complex service environments. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0016] Figure 1 This is a schematic diagram of the filling chamber structure provided in an embodiment of the present invention; Figure 2 This is a flowchart of the multi-dimensional intelligent detection method for the performance of highway asphalt mixtures provided in this embodiment of the invention. Figure 3 This is a flowchart of the deformation analysis sub-strategy provided in an embodiment of the present invention; Figure 4 This is a flowchart of the seepage analysis sub-strategy provided in an embodiment of the present invention; In the diagram, 1 is the filling chamber; 2 is the modular testing plate; 3 is the asphalt mixture specimen; and 4 is the simulated filling material. Detailed Implementation

[0017] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0018] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0019] The following is combined with Figures 1-4 This invention is described in detail.

[0020] like Figure 1 As shown in the figure, this invention provides a multi-dimensional intelligent testing system for the performance of highway asphalt mixtures, comprising: a simulation testing platform including several filling chambers for filling with simulated fillers of different physical properties to simulate different roadbeds; several modular testing plates that can be spliced ​​together are arranged above the filling chambers for supporting asphalt mixture specimens; an environmental simulation and loading module for applying a controllable comprehensive environmental load to the asphalt mixture specimens placed on the simulation testing platform; a multi-source information acquisition module including image acquisition devices arranged above and to the sides of the simulation testing platform for real-time acquisition of multi-dimensional image sequences of the asphalt mixture specimens on each filling chamber and each simulation testing plate during the application of the comprehensive environmental load; and an intelligent analysis and processing module for performing performance testing on the asphalt mixture specimens based on the multi-dimensional image sequences to obtain the comprehensive performance of the asphalt mixture specimens under different road conditions.

[0021] Specifically, simulated filler refers to the material filled into the filler chamber to replace the subgrade or base course materials in the actual road structure. Its selection criteria are based on its physical properties, such as stiffness, modulus of elasticity, hydration, and coefficient of thermal expansion, rather than its chemical composition. By changing the simulated filler with different properties, such as from high-modulus cement-stabilized crushed stone to low-modulus sand, or saturated materials with freeze-thaw sensitivity, the system can reproduce a variety of typical or extreme subgrade support conditions under laboratory conditions, ranging from solid to weak, from dry to waterlogged, and from normal temperature to freeze-thaw cycles.

[0022] Specifically, the comprehensive environmental load is a set of loads composed of various types of physical actions that change over time according to a specific program. It is used to simulate the coupled effects of factors such as traffic wheel loads, temperature changes, and precipitation infiltration on pavement structures in the real world. A multi-dimensional image sequence refers to a collection of ordered image frames continuously captured by an image acquisition device throughout the entire time history of the comprehensive environmental load application.

[0023] Preferably, the comprehensive environmental load includes mechanical stress, permeation stress, and temperature stress; the environmental simulation and loading module is configured to include: a servo loading mechanism for applying the mechanical stress; a spraying mechanism for simulating rainwater to generate the permeation stress; and a temperature control mechanism for adjusting the ambient temperature to generate the temperature stress.

[0024] The servo loading mechanism generates and applies programmable mechanical stress to simulate the vertical pressure and / or horizontal shear force exerted by vehicle tires on the road surface. The main body of this mechanism is vertically mounted directly above each filling chamber, and its loading head can move downwards to directly contact and press against the surface of the asphalt mixture specimen placed on the modular testing plate. The spraying mechanism simulates the effect of rainfall on the road surface by spraying water. Penetration stress refers to the dynamic water pressure and pore water pressure generated under load and temperature changes after water seeps into the pores of the mixture. The spray heads of this mechanism are horizontally arranged above each modular testing plate to ensure that the sprayed water evenly covers the specimen surface. The temperature control mechanism simulates the diurnal temperature range and seasonal changes in nature by changing the ambient temperature. Temperature stress mainly refers to the internal stress generated by the thermal expansion and contraction of the asphalt mixture and its underlying structure due to temperature changes. This mechanism can precisely control the temperature of the entire testing space.

[0025] Preferably, each filling chamber corresponds to a preset number of modular testing plates, and each modular testing plate has a unique identification code; the multi-dimensional image sequence includes an asphalt image sequence and a water seepage image sequence of the modular testing plates, and the intelligent analysis and processing module is also used to associate the testing plate image sequence corresponding to each simulated testing plate according to the unique identification code.

[0026] The unique identifier of the modular testing plate is a unique numerical or symbolic code assigned to each physical testing plate within the system. The asphalt image sequence refers to the image sequence obtained by the image acquisition device with the asphalt mixture specimen itself as the target, aiming to record the evolution of the material's own state, such as deformation, texture changes, and aggregate spalling on the specimen surface. The water seepage image sequence refers to the image sequence obtained by the image acquisition device with the testing plate and the specimen it supports as the observation area, specifically capturing the presence, diffusion, and flow phenomena of liquids. Its core objective is to quantify the behavior of water near the interface between the specimen and the testing plate, such as lateral diffusion, downward seepage, and surface runoff.

[0027] Specifically, the system associates the image sequence of each simulated test plate with the image sequence of the test plate. The specific operation is that the system automatically collects, indexes and encapsulates all asphalt image sequences and water seepage image sequences originating from the same modular test plate based on the unique identifier code that was bound to the image data during acquisition. This provides accurate and orderly data input for subsequent independent and complete multi-dimensional performance analysis of the specimens on the test plate.

[0028] The multi-dimensional intelligent testing system for highway asphalt mixture performance provided in this invention constructs a highly integrated and automated comprehensive material performance evaluation platform, fundamentally revolutionizing the traditional testing paradigm that relies on single, isolated testing equipment. Through the collaborative design of its core modular testing plate and filling chamber, the system achieves, for the first time in the laboratory, flexible and parallel simulation of various roadbed physical properties. This allows for simultaneous comparative tests under different road conditions on the same platform, greatly improving the coverage and efficiency of the testing. Simultaneously, the integrated environmental simulation and loading module, through the coordinated operation of servo loading, precision spraying, and programmed temperature control, can apply precisely controllable comprehensive environmental loads, realistically replicating the coupled effects of traffic, precipitation, and temperature changes experienced by real road surfaces, significantly enhancing the realism and rigor of the test conditions. The seamless integration of the multi-source information acquisition module and the intelligent analysis and processing module enables multi-view, high spatiotemporal resolution image recording and automated feature extraction throughout the entire test process. This upgrades performance evaluation from a "black box" interpretation that relies on endpoint data to intelligent diagnosis based on the entire cycle and visualized damage evolution process. As a result, it provides unprecedented, reliable, and efficient hardware support for the performance evaluation, quality control, and applicability research of asphalt mixtures.

[0029] like Figure 2As shown, this embodiment of the invention also provides a multi-dimensional intelligent detection method for the performance of highway asphalt mixtures, applied to the aforementioned multi-dimensional intelligent detection system for highway asphalt mixture performance. The method includes: Step S1: Simulated fillers with different physical properties are configured in different filling chambers of the simulated testing platform, and asphalt mixture specimens are laid on the surface of the testing plates; Step S2: A time-controlled comprehensive environmental load is applied to the asphalt mixture specimens according to preset testing rules; Step S3: A multi-dimensional image sequence of the asphalt mixture specimens under the comprehensive environmental load is obtained, and each modular testing plate is associated with the corresponding water seepage image sequence in the multi-dimensional image sequence; Step S4: Based on the multi-dimensional image sequence corresponding to each simulated filler, the deformation characteristics, surface water diffusion characteristics, and aggregate particle detachment characteristics of the asphalt mixture specimens on each simulated filler are extracted using a feature analysis strategy to form a feature set; Based on the feature set and combined with a preset performance evaluation strategy, the comprehensive performance of the asphalt mixture specimens under different road conditions is obtained.

[0030] Among them, time-controllable refers to the precise scheduling of loading mechanisms to perform loading actions in a coordinated or alternating manner by the central controller according to preset absolute time points or relative time intervals. The multi-dimensional image sequence corresponding to each simulated filler refers to the set of image sequences taken from the modular detection plate area associated with the specific simulated filler, retrieved and extracted from the overall image database based on the spatial layout and identification information of the simulation detection platform.

[0031] Specifically, the detection rules are a program script consisting of a series of logical and timing instructions, which specifies the order, intensity, duration, and coupling relationship of the three load components: mechanical stress, permeation stress, and temperature stress. For example, a testing rule simulating water damage after freeze-thaw cycles under heavy traffic can include a time sequence: First, at a baseline ambient temperature of 15°C, a servo loading mechanism is activated to apply 100,000 standard wheel loads to asphalt mixture specimens at a pressure of 0.7 MPa and a frequency of 60 cycles / minute, simulating the initial traffic compaction and stabilization stage; then, while maintaining the loading pressure, a spraying mechanism is simultaneously activated to simulate continuous moderate rain, while a temperature control mechanism is controlled to uniformly lower the ambient temperature from 15°C to -10°C over 4 hours, simulating the autumn and winter cooling and rainfall process, where moisture gradually seeps in and eventually freezes in the pores; next, the temperature is maintained at -10°C for 12 hours to allow the internal moisture of the specimen to fully freeze and generate frost heave stress; then, the temperature control mechanism is controlled to raise the temperature back to 5°C over 8 hours, simulating the spring warming and snow melting process, and when the temperature reaches 2°C, the servo loading mechanism is restarted to continue 50,000 loads, while intermittent short-term spraying can be used to simulate the effect of snow melt water. This rule, by precisely arranging the order, intensity, and coupling timing of the application of mechanical, penetrating, and temperature stresses, comprehensively stimulates the possible rutting deformation, freeze-thaw damage, water damage, and their coupling effects in a single experiment.

[0032] The multi-dimensional intelligent detection method for highway asphalt mixture performance provided in this invention constructs a complete and automated technical closed loop, encompassing test configuration, process simulation, data acquisition, and intelligent evaluation. Through programmable detection rules, this method can accurately reproduce complex and variable real-world service conditions in a single test, achieving simultaneous, multi-dimensional, and quantitative analysis of the performance response of asphalt mixtures under comprehensive environmental loads. It integrates traditionally isolated and fragmented single-performance tests into a systematic evaluation process that dynamically reveals the interactions between multiple failure modes such as deformation, water seepage, and spalling. This not only significantly improves detection efficiency and the realism of scenario coverage but also upgrades performance evaluation from empirical judgment relying on endpoint data to a traceable and interpretable intelligent diagnosis based on full-cycle damage evolution characteristics, thus providing unprecedented scientific decision-making basis for material design optimization and engineering selection.

[0033] like Figure 3As shown, preferably, the asphalt image sequence includes a deformation image sequence and a surface state image sequence. The feature analysis strategy includes a deformation analysis sub-strategy, which includes: calculating the vertical displacement field of the asphalt mixture specimen based on the deformation image sequence corresponding to each filler, and extracting the rutting depth and vertical displacement standard deviation from the vertical displacement field; calculating the horizontal displacement field of the surface of the asphalt mixture specimen based on the surface state image sequence corresponding to the same filler, and extracting the maximum horizontal shear displacement and surface roughness evolution rate from the horizontal displacement field; and using the set composed of the rutting depth, vertical displacement standard deviation, maximum horizontal shear displacement, and surface roughness evolution rate as the deformation feature to characterize the deformation performance of the asphalt mixture specimen.

[0034] The deformation image sequence refers to the set of image data acquired to quantify the macroscopic and three-dimensional deformation of asphalt mixture specimens. This sequence is obtained by multiple high-resolution industrial cameras positioned at specific angles to the side of the specimen. The surface condition image sequence refers to the set of image data acquired to observe and analyze the texture, particle movement, and horizontal deformation of the upper surface of asphalt mixture specimens. This sequence is obtained by multiple cameras positioned vertically above the specimen.

[0035] Specifically, the vertical displacement field is calculated using the digital image correlation method. This method tracks the pixel displacements of natural textures or artificial speckle patterns on the specimen's side surface in a series of images before and after loading with high precision. Using a correlation coefficient matching algorithm, it calculates the displacement components of thousands of feature points on the specimen surface perpendicular to the loading direction, thereby reconstructing a two-dimensional vertical displacement distribution cloud map of the entire observation surface. The rut depth extracted from this displacement field specifically refers to the global maximum value of the displacement values ​​at all points in the displacement field. The standard deviation of the vertical displacement is a statistical measure of the dispersion of the displacement values ​​at all points in the displacement field relative to their average value. It characterizes the uniformity or local concentration of rut deformation; a larger standard deviation indicates more uneven deformation and a potential risk of local shear failure.

[0036] Specifically, the horizontal displacement field can be calculated using optical flow or image correlation methods targeting surface texture. Optical flow estimates the velocity vector of each pixel within the image plane by analyzing grayscale or color variations between consecutive frames, thus obtaining a two-dimensional horizontal motion field on the specimen surface. The maximum horizontal shear displacement extracted from this field specifically refers to the maximum relative displacement between two adjacent points within the entire horizontal displacement field. It directly reflects the maximum local shear slip under load and is a key indicator for evaluating the shear deformation resistance and internal interlocking structure stability of the mixture. The surface roughness evolution rate is quantified by analyzing the rate of change of texture features (such as contrast and entropy values ​​of the grayscale co-occurrence matrix) over time in a sequence of surface images. It characterizes the dynamic rate at which the microtexture of the specimen surface becomes smooth or rough due to wear, polishing, or particle rearrangement under combined environmental loads, and is an important indirect parameter for evaluating the anti-skid performance and durability of the mixture.

[0037] For example, when an asphalt mixture rutting specimen is laid on a filler bed simulating a frozen soil subgrade, and loading is applied, the deformation image sequence captured by the side camera, after digital image correlation processing, generates a full-field vertical displacement cloud map. The maximum rutting depth of 8.2 mm is automatically extracted from this map, and the standard deviation of the vertical displacement is calculated to be 1.5 mm, indicating concentrated deformation at the bottom of the rutting groove. Simultaneously, the surface condition image sequence captured by the camera directly above, analyzed using optical flow, calculates the maximum horizontal shear displacement of 0.8 mm, revealing significant relative slippage between aggregates. Texture analysis shows that the surface roughness evolution rate at the end of the test is -15%, indicating significant wear and smoothing of the surface texture under the coupled effects of load and water freezing. Finally, the set of four quantified values ​​{8.2 mm, 1.5 mm, 0.8 mm, -15%} represents the comprehensive deformation characteristics of the specimen under this simulated road condition.

[0038] In a preferred embodiment of this invention, a full-field, multi-dimensional quantitative analysis of the three-dimensional deformation behavior of asphalt mixture specimens is achieved by fusing deformation image sequences obtained from the side of the specimen and surface state image sequences obtained from directly above. This method not only accurately measures the macroscopic rut depth but also simultaneously assesses the uniformity of the deformation field, the maximum shear deformation within the material, and the durability evolution of surface texture. This expands the traditional single rut depth index into a comprehensive set of deformation characteristics that fully and deeply reflects the material's resistance to permanent deformation, shear stability, and surface functional degradation trends, providing refined and visualized data support for the scientific evaluation of the high-temperature stability and long-term service performance of asphalt mixtures.

[0039] like Figure 4As shown, preferably, the feature analysis strategy further includes a seepage analysis sub-strategy, which includes: based on the seepage image sequence corresponding to each filling chamber, using a time-series image segmentation algorithm to identify and track the diffusion contour of water traces in the seepage image sequence, and calculating the rate of change of its diffusion area and perimeter over time; quantifying its surface runoff characteristics and internal infiltration depth by analyzing the diffusion path and final steady-state distribution of the water traces; and determining the seepage characteristics based on the rate of change of diffusion area, surface runoff characteristics, and internal infiltration depth.

[0040] The water seepage image sequence is acquired by a moisture-sensitive image acquisition device, such as a camera sensitive to moisture absorption at a specific wavelength. Temporal image segmentation algorithms are key technologies for automatic identification of moisture regions. For example, a threshold segmentation method based on the HSV color space can be used to automatically separate pixels representing wet areas from pixels in the dry background in each frame, generating a series of time-varying binary mask images. Identifying and tracking the water trail diffusion contours involves automatically detecting connected regions representing water trails in each frame of the aforementioned binary mask sequence and calculating their geometric features, such as area and perimeter. Calculating the rate of change of the diffusion area and perimeter over time is based on the temporal data of these contours. Differential calculations are used to obtain the rate of increase in water trail coverage area and the rate of expansion of contour complexity per unit time. These two rates directly reflect the dynamic process of moisture intrusion.

[0041] Specifically, the diffusion path can be obtained by tracing the movement of the centroid of the water trail region across multiple frames of images, used to determine whether the water infiltrates uniformly or if a dominant flow channel exists. Surface runoff characteristics can be quantified by analyzing parameters such as the duration of contact between the water trail profile and the specimen edge, and the length of the contact line, characterizing the ability of water to flow away along the surface when it fails to infiltrate in time. Internal infiltration depth refers to the depth of the water front's downward movement measured through the side observation window image of the specimen.

[0042] For example, when asphalt mixture specimens were laid on fillers of roadbeds in simulated rainy areas and sprayed with water, the infiltration image sequence captured by a moisture-sensitive camera was processed by a time-series image segmentation algorithm. The algorithm automatically tracked the increase in the water stain diffusion area from 0 square millimeters to 8500 square millimeters within 300 seconds after the spraying began, calculating an average diffusion area change rate of 28.3 square millimeters per second. By analyzing the centroid trajectory of the water stain, it was found that its diffusion path showed a clear tendency to converge towards one edge of the specimen, rather than uniform infiltration. At the same time, the algorithm detected that the water stain outline and the edge of the specimen came into contact 120 seconds after the spraying began and lasted for 180 seconds, with the contact line length accounting for 40% of the total edge length, which quantitatively characterizes significant surface runoff features. Combined with the side observation window image, the internal infiltration depth of water at the end of the experiment was measured to be 8 millimeters. Ultimately, the set of three quantitative indicators, {28.3 mm² / s, significant lateral convergence and runoff, 8 mm}, represents the comprehensive seepage characteristics of this specimen under simulated rainy road conditions.

[0043] A preferred embodiment of this invention achieves a comprehensive, quantifiable, and multi-dimensional assessment of the water damage resistance of asphalt mixture specimens through automated time-series image processing and analysis of permeation image sequences reflecting dynamic moisture migration. This method can accurately track and quantify the diffusion rate, spatial distribution path, surface runoff tendency, and final penetration depth of water intrusion, thus upgrading the traditional single, final-state "permeability coefficient" test to a comprehensive evaluation that reveals the dynamic mechanism of moisture damage. This provides a refined, intuitive, and repeatable scientific basis for accurately diagnosing the water sensitivity of materials and assessing their long-term durability risk under actual rain, freeze-thaw, and other multi-water environments.

[0044] Preferably, the feature analysis strategy further includes a spalling monitoring sub-strategy, which includes: precisely synchronizing the surface state image sequence with the time sequence of the applied comprehensive environmental load to establish a correspondence between image frames and load values ​​and load types; for image frames corresponding to different load stages, using an adaptive threshold segmentation algorithm to identify aggregate particles detached from the surface of the asphalt mixture specimen; counting the cumulative number of aggregate particles identified under a specific load and calculating the average spalling rate per unit time; and determining the aggregate particle detachment characteristics based on the cumulative number of aggregate particles, the average spalling rate, and their correspondence with the load to characterize the aggregate retention capacity and bond failure critical conditions of the asphalt mixture specimen under different comprehensive environmental loads.

[0045] Specifically, precise synchronization is achieved by assigning a unified, high-precision timestamp to the image acquisition device and load sensors (such as force sensors and temperature and humidity sensors). Each frame of image and each moment of load data carries this timestamp, enabling alignment in the database based on time and establishing a correspondence between image frames, load values, and load types. For example, it can be known that an image was taken under 1.1 MPa mechanical stress, 5°C temperature, and with the sprinkler system on. Adaptive threshold segmentation algorithms, such as thresholding algorithms based on local image characteristics, can automatically calculate the optimal segmentation threshold based on the overall or local brightness and contrast of each frame of image. This effectively addresses dynamic changes in the image background caused by changes in lighting, water reflections, and surface wetting during the experiment, ensuring stable separation of the foreground and background at different load stages (such as dry loading, wet loading, and low-temperature loading). Identified "aggregate particles" appear in the image as newly emerging, small, bright connected regions.

[0046] Specifically, a specific load can be a load range, a load type, or a load combination. Based on the established synchronization relationship, the system automatically filters all image frames corresponding to the specific load condition and accumulates all identified particles in these frames to obtain a cumulative count. The average spalling rate is then calculated by dividing this cumulative count by the total duration of the corresponding load condition. For example, the average spalling rate can be calculated separately for the "high temperature (60°C) heavy load (1.0MPa) stage" and the "freeze-thaw cycle (-10°C to 5°C) stage".

[0047] A preferred embodiment of this invention achieves load-condition-triggered diagnosis of asphalt mixture bond failure behavior by performing in-depth correlation analysis between dynamic visual observation of aggregate particle detachment and precise temporal sequence of applied comprehensive environmental loads. This method not only quantifies the total amount and average rate of detached particles but also precisely reveals the causal relationship between detachment and specific mechanical, penetrating, or temperature stresses, thereby identifying the critical load conditions leading to material failure. This surpasses the limitations of traditional detachment tests, which only provide the final mass loss rate, and provides a scientifically sound, mechanistic basis for evaluating the aggregate retention capacity of asphalt mixtures under different harsh environments, predicting the long-term durability of their bond structure, and specifically optimizing materials to resist specific environmental stresses.

[0048] Preferably, the performance evaluation strategy includes: for each feature in the feature set, querying a preset benchmark database to obtain a benchmark threshold corresponding to each feature based on the subgrade type simulated by the current simulated infill; comparing the parameter value of each feature with the corresponding benchmark threshold, and calculating the performance score corresponding to each feature according to a preset scoring function; obtaining the dynamic weight coefficient of each feature from a preset weight knowledge base based on the subgrade type simulated by the current simulated infill; and obtaining the comprehensive performance through weighted fusion calculation based on the performance score of each feature and its corresponding dynamic weight coefficient.

[0049] The benchmark database is a pre-generated knowledge base that stores the acceptable or excellent ranges of various performance characteristic parameters for different typical subgrade types. For example, for infill materials simulating weak subgrades, considering the risk of uneven settlement, the database sets more stringent allowable thresholds for deformation characteristics such as rut ​​depth than for rigid base courses. For infill materials simulating subgrades in rainy or freeze-thaw regions, the benchmark thresholds for seepage and spalling characteristics are even more stringent. The system automatically calls the corresponding benchmark threshold set by identifying the identifier of the current simulated infill material. The scoring function is a mathematical relationship that maps measured characteristic values ​​to standard scores. Its core logic is that a high score is obtained when the characteristic value is equal to or better than the benchmark threshold; when the characteristic value is worse than the threshold, the score decreases according to the functional relationship. Through this step, original characteristics of different dimensions and orders of magnitude are uniformly transformed into dimensionless, comparable performance scores. The weighted knowledge base defines the relative importance of various performance dimensions in the overall evaluation under different subgrade types. For example, in the evaluation of subgrades in high-temperature and heavy-load areas, deformation characteristics have the highest weight coefficient; in the evaluation of subgrades in seasonally frozen soil areas, the weight coefficients of seepage characteristics and spalling characteristics will be significantly increased. The dynamic weight coefficient is a sum vector, the sum of its elements is one, and each element corresponds to the weight of a feature dimension. The weighted fusion is the sum of the products of each feature performance score and its corresponding dynamic weight coefficient. The result is a quantitative comprehensive performance value, such as a score between zero and one hundred. This value comprehensively reflects the overall performance level of the asphalt mixture specimen on multiple key performance dimensions under specific simulated road conditions.

[0050] Specifically, the scoring function formula is as follows: For features where smaller values ​​generally indicate better performance, such as rut ​​depth, the performance score is... The calculation formula is:

[0051] in, This is the measured value of this feature. The preset excellent threshold, The preset non-compliance threshold is, and < .

[0052] For features where higher values ​​generally indicate better performance, such as processed positive metrics, the performance score... The calculation formula is:

[0053] in, This is the measured value of this feature. The preset excellent threshold, The preset non-compliance threshold is, and .

[0054] For features where the value must fall within a specific range to be considered excellent, the formula for calculating the performance score S is as follows:

[0055] in, This is the measured value of this feature. The preset lower limit for acceptance. The preset upper limit of qualification constitutes the ideal range. ], The non-compliance threshold is too low. The non-compliance threshold is excessively high, and < < < .

[0056] For example, suppose we analyze an asphalt mixture specimen in a simulated "seasonally freeze-thawed weak subgrade" infill. The extracted features are: rut depth of 6.8 mm (deformation feature), moisture diffusion area change rate of 25 mm² / s (seepage feature), and average spalling rate of 1.2 specimens / minute during the freeze-thaw cycle (stripping feature). Based on the identifier of the current simulated infill, the system automatically retrieves the corresponding benchmark thresholds from the benchmark database. The benchmark thresholds are set as follows: Deformation Feature =5mm, =10mm; seepage characteristics =15mm² / s, =40mm² / s; spalling characteristics =0.5 pieces / min =2.0 pieces / min; For the performance score of deformation characteristics, since 5 < 6.8 < 10, therefore S = Similarly, the performance scores for seepage characteristics and spalling characteristics were calculated to be 60 and 53.3, respectively. Subsequently, the system retrieved the dynamic weight coefficient corresponding to this "seasonally frozen-thawed weak subgrade" from the weight knowledge base and set it as: deformation weight W. RD =0.4, seepage weight W AR =0.35, peeling weight W DR =0.25. Finally, the weighted fusion calculation is performed to obtain the comprehensive performance value P: P=64×0.4+60×0.35+53.3×0.25=25.6+21+13.325≈59.9.

[0057] A preferred embodiment of this invention establishes a personalized, quantifiable, and intelligent comprehensive material performance evaluation engine. By introducing a benchmark database for different road conditions and a dynamic weight knowledge base, it achieves automatic adaptation of evaluation standards and focuses to simulated road conditions, solving the problem of traditional evaluation methods being detached from actual engineering scenarios. Simultaneously, the scoring function unifies the original characteristics of different physical dimensions into comparable and calculable performance scores, and then integrates these sub-scores into a single, intuitive comprehensive performance value through a weighted fusion model. This ultimately transforms multi-dimensional and complex material response data into a decision-making basis that can scientifically and objectively reflect the applicability and durability level of asphalt mixtures under specific target road conditions, thus providing an efficient and reliable intelligent evaluation tool for material selection, quality control, and engineering design.

[0058] Preferably, step S4 further includes: determining the weak feature based on the performance score corresponding to each feature, determining the target raw material corresponding to the weak feature by combining the preset raw material component-performance mapping relationship; and calculating the adjustment direction and adjustment amount of the target raw material based on the difference between the performance score of the weak feature and the target value.

[0059] Specifically, weak features refer to feature dimensions whose performance scores are below a target threshold, which is a pre-set value in the system. The raw material component-performance mapping relationship establishes the qualitative and quantitative influence relationships between various key raw material components and material properties in asphalt mixtures. For example, high-temperature deformation performance is mainly affected by the high-temperature viscosity of asphalt, aggregate skeleton structure, and anti-rutting agents; water damage performance is mainly affected by the adhesion between asphalt and aggregate, porosity, and anti-stripping agents; and anti-stripping performance is closely related to the thickness of the asphalt film, aggregate surface properties, and anti-stripping agents. When the system identifies a stripping dimension as a weak feature, it will automatically lock onto the target raw material that may cause the problem based on this mapping relationship, such as "insufficient adhesion between asphalt and aggregate" or "low anti-stripping agent dosage."

[0060] Specifically, the adjustment direction refers to whether the dosage of the target raw material should be increased or decreased, or its quality grade should be improved. The adjustment amount refers to the specific increase or decrease in value, and its calculation relies on a preset adjustment amount calculation model. This model is a machine learning model trained on a large amount of historical test data. The input of the model is the performance score difference, i.e., the gap between the actual score of the weak feature and the target threshold, and the output is the recommended adjustment amount and direction for the determined target raw material. For example, if the system determines that the "stripping score" is 20 points lower than the target value, and the mapping relationship points to "insufficient liquid anti-stripping agent content", the adjustment amount calculation model recommends "increasing the liquid anti-stripping agent content by 0.2%" based on the relationship fitted from historical data. The raw material component-performance mapping relationship and adjustment amount calculation model are established by data mining and regression analysis on a large number of asphalt mixture formulations with known performance and their test results. In a preferred embodiment of this invention, the weak links in performance are traced back to specific material components through a raw material-performance mapping relationship, and the performance gap is transformed into specific raw material adjustment directions and dosage recommendations using an adjustment calculation model. This provides quantifiable and actionable intelligent decision support for the formulation optimization and customized design of asphalt mixtures.

[0061] In summary, the present invention provides a multi-dimensional intelligent testing system and method for highway asphalt mixture performance, creatively constructing a highly integrated, programmable, visualized, and intelligent material performance evaluation platform. This system, by simulating various real roadbed conditions in parallel in the laboratory and applying programmable coupled environmental loads, revolutionizes traditional isolated and static testing methods into a comprehensive experimental paradigm that can simultaneously stimulate and quantify the material's multi-dimensional performance responses in deformation, water permeability, and spalling. The accompanying methodological processes, particularly intelligent feature extraction based on multi-view image sequences, personalized performance evaluation tailored to road conditions, and reverse optimization recommendation of raw materials based on data models, together constitute a complete technical closed loop from physical simulation and data perception to intelligent decision-making. This fundamentally solves the industry pain points of existing technologies, such as fragmented testing dimensions, distorted working condition simulation, and evaluation results that cannot effectively guide engineering practice. It provides an unprecedentedly efficient, scientific, and reliable intelligent solution for the high-performance design, precise selection, and quality control of asphalt pavement materials.

[0062] The above description is merely a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-dimensional intelligent testing system for the performance of asphalt mixtures for highways, characterized in that, include: The simulation testing platform includes several filling chambers for filling with simulated fillers of different physical properties to simulate different roadbeds; several modular testing plates that can be spliced ​​are set above the filling chambers to support asphalt mixture specimens. An environmental simulation and loading module is used to apply a controllable comprehensive environmental load to asphalt mixture specimens placed on the simulation testing platform. The multi-source information acquisition module includes image acquisition devices arranged above and to the side of the simulation testing platform, which are used to acquire multi-dimensional image sequences of asphalt mixture specimens and simulation testing plates on each filling bin in real time during the application of the comprehensive environmental load. The intelligent analysis and processing module is used to perform performance testing on asphalt mixture specimens based on the multi-dimensional image sequence, and obtain the comprehensive performance of asphalt mixture specimens under different road conditions.

2. The multi-dimensional intelligent testing system for the performance of highway asphalt mixtures according to claim 1, characterized in that, The comprehensive environmental load includes mechanical stress, permeability stress, and temperature stress; The environment simulation and loading module is configured to include: A servo loading mechanism is used to apply the mechanical stress; A spray system is used to simulate rainwater to generate the aforementioned permeation stress; A temperature control mechanism is used to regulate the ambient temperature to generate the temperature stress.

3. The multi-dimensional intelligent testing system for the performance of highway asphalt mixtures according to claim 1, characterized in that, Each filling chamber corresponds to a preset number of modular testing plates, and each modular testing plate has a unique identification code; the multi-dimensional image sequence includes an asphalt image sequence and a water seepage image sequence of the modular testing plates, and the intelligent analysis and processing module is also used to associate the testing plate image sequence corresponding to each simulated testing plate according to the unique identification code.

4. A multi-dimensional intelligent testing method for the performance of highway asphalt mixtures, applied to the multi-dimensional intelligent testing system for the performance of highway asphalt mixtures as described in any one of claims 1-3, characterized in that, include: Step S1: In different filling chambers of the simulation testing platform, simulated fillers with different physical properties are configured, and asphalt mixture specimens are laid on the surface of the testing plate. Step S2: Apply a time-controlled comprehensive environmental load to the asphalt mixture specimens according to the preset testing rules; Step S3: Obtain the multi-dimensional image sequence of the asphalt mixture specimen under comprehensive environmental load, and associate each modular test plate with the corresponding seepage image sequence in the multi-dimensional image sequence; Step S4: Based on the multi-dimensional image sequence corresponding to each simulated filler, the deformation characteristics, surface water diffusion characteristics, and aggregate particle detachment characteristics of the asphalt mixture specimen on each simulated filler are extracted through feature analysis strategy to form a feature set; based on the feature set and combined with the preset performance evaluation strategy, the comprehensive performance of the asphalt mixture specimen under different road conditions is obtained.

5. The multi-dimensional intelligent detection method for the performance of highway asphalt mixtures according to claim 4, characterized in that, The asphalt image sequence includes a deformation image sequence and a surface state image sequence. The feature analysis strategy includes a deformation analysis sub-strategy, which includes: Based on the deformation image sequence corresponding to each filling bin, the vertical displacement field of the asphalt mixture specimen is calculated, and the rutting depth and vertical displacement standard deviation are extracted from the vertical displacement field. Based on the surface state image sequence corresponding to the same filling bin, the horizontal displacement field of the asphalt mixture specimen surface is calculated, and the maximum horizontal shear displacement and surface roughness evolution rate are extracted from the horizontal displacement field. The set of rut depth, vertical displacement standard deviation, maximum horizontal shear displacement, and surface roughness evolution rate is used as the deformation characteristics to characterize the deformation performance of asphalt mixture specimens.

6. The multi-dimensional intelligent detection method for the performance of highway asphalt mixtures according to claim 4, characterized in that, The feature analysis strategy further includes a seepage analysis sub-strategy, which includes: Based on the seepage image sequence corresponding to each filling chamber, the diffusion contour of water traces in the seepage image sequence is identified and tracked by the time-series image segmentation algorithm, and the rate of change of its diffusion area and perimeter over time is calculated. By analyzing the diffusion path and final steady-state distribution of water traces, the surface runoff characteristics and internal infiltration depth are quantified. The seepage characteristics are determined based on the rate of change of diffusion area, surface runoff characteristics, and internal infiltration depth.

7. The multi-dimensional intelligent detection method for the performance of highway asphalt mixtures according to claim 5, characterized in that, The feature analysis strategy also includes a peeling detection sub-strategy, which includes: The surface state image sequence is precisely synchronized with the time sequence of the applied comprehensive environmental load to establish the correspondence between image frames and load values ​​and load types. For image frames corresponding to different load stages, an adaptive threshold segmentation algorithm is used to identify aggregate particles that have detached from the surface of the asphalt mixture specimen. The cumulative number of aggregate particles identified under a specific load is counted, and the average spalling rate per unit time is calculated. Based on the cumulative number of aggregate particles, the average spalling rate, and their correspondence with the load, the aggregate particle detachment characteristics are determined to characterize the aggregate retention capacity and critical conditions for bond failure of asphalt mixture specimens under different comprehensive environmental loads.

8. The multi-dimensional intelligent detection method for the performance of highway asphalt mixtures according to claim 4, characterized in that, The performance evaluation strategy includes: For each feature in the feature set, based on the roadbed type simulated by the current simulated filling material, a preset benchmark database is queried to obtain the benchmark threshold corresponding to each feature. The parameter values ​​of each feature are compared with the corresponding benchmark thresholds, and the performance score corresponding to each feature is calculated according to the preset scoring function. Based on the roadbed type simulated by the current simulated filling material, the dynamic weight coefficients of each feature are obtained from the preset weight knowledge base; The overall performance is obtained by weighted fusion calculation based on the performance scores of each feature and their corresponding dynamic weight coefficients.

9. The multi-dimensional intelligent detection method for the performance of highway asphalt mixtures according to claim 8, characterized in that, Step S4 further includes: Based on the performance scores corresponding to each feature, weak features are identified, and the target raw materials corresponding to the weak features are determined by combining the preset raw material component-performance mapping relationship. Based on the difference between the performance score of the weak feature and the target value, the adjustment direction and adjustment amount of the target raw material are calculated.

Images