Asphalt pavement microwave heat-induced crack self-healing method and device based on digital twinning

Abstract

The present application relates to a kind of asphalt pavement microwave heat induced crack self-healing method and device based on digital twinning, wherein the method comprises: step S1: establishing the three-dimensional static digital twinning platform of road;Step S2: obtain the aerial result of unmanned aerial vehicle to road, monitor whether there is crack;Step S3: when finding that there is crack, locate crack position, adopt ground penetrating radar to detect crack position and obtain the morphological data and material data of crack;Step S4: based on the morphological data and material data of crack position, combined with three-dimensional static digital twinning platform, using microwave heat induced asphalt self-healing process simulation analysis determines the best frequency and best power;Step S5: the best frequency and best power determined in simulation analysis process are used to repair crack in road.Compared with prior art, the present application has the advantages of improving self-healing effect etc.

Description

Technical Field

[0001] This invention relates to the field of self-healing asphalt pavement, and in particular to a method and apparatus for self-healing microwave thermally induced cracks in asphalt pavement based on digital twins. Background Technology

[0002] Asphalt pavement is widely used due to its excellent smoothness, high driving comfort, low noise level, and good waterproof properties. However, asphalt concrete pavement is prone to fatigue cracking under extreme temperature changes and repeated vehicle loads, and this cracking continues to expand over time, leading to a decline in the service performance of asphalt pavement and a significant reduction in its service life.

[0003] Researchers have continuously explored various methods to improve the self-healing ability of asphalt concrete pavements based on the self-healing mechanism of asphalt concrete. These methods mainly include microcapsule self-healing, polymer-modified self-healing, biomaterial self-healing, and thermally induced self-healing. Among these, microwave thermally induced self-healing is highly regarded for its advantages, such as more uniform internal temperature heating of concrete, deeper effective depth, multiple repair capabilities, and cost savings of 30%-40% compared to traditional repair methods. The mechanism of microwave heating involves adding functional fillers to asphalt concrete. When the functional fillers are hindered in the microwave high-frequency magnetic field, they dissipate and absorb the microwaves, converting the field energy into heat energy, which is then transferred to the asphalt mastic, causing the asphalt mixture to self-heal at high temperatures.

[0004] Existing research indicates that microwave thermally induced self-healing efficiency is closely related to the degree of damage to asphalt concrete, especially in the early stages of crack formation, where timely repair can significantly restore its performance. However, current artificial microwave thermal induction methods have certain limitations, lacking the ability to repair cracks in real time and making it difficult to effectively intervene in the early stages of crack formation to improve self-healing efficiency. Furthermore, these methods theoretically fail to clearly define the means of determining the microwave frequency and power during microwave thermal induction, relying mostly on empirical judgment. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A method for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins, comprising:

[0008] Step S1: Establish a three-dimensional static digital twin platform for the road;

[0009] Step S2: Obtain aerial photographs of the road taken by the drone to monitor for the presence of cracks;

[0010] Step S3: Once a crack is detected, locate the crack position and use ground-penetrating radar to detect the crack position and obtain crack morphology and material data.

[0011] Step S4: Based on the morphological and material data of the crack location, and combined with the three-dimensional static digital twin platform, the optimal frequency and optimal power are determined by microwave thermally induced asphalt self-healing process simulation analysis.

[0012] Step S5: Repair the cracks in the road using the optimal frequency and power determined by the simulation analysis process.

[0013] Step S1 specifically includes:

[0014] Step S11: Use a LiDAR mobile scanning system to scan the road and obtain vehicle point cloud data in the road surface coordinate system. The LiDAR mobile scanning system integrates a global navigation satellite system, an inertial navigation system, a LiDAR, and a time synchronization unit. It obtains trajectory position information through combined navigation of the global navigation satellite system and the inertial navigation system, collects point cloud data through LiDAR, and processes the trajectory data and point cloud data based on the principles of time registration and spatial registration to obtain vehicle point cloud data in the road surface coordinate system.

[0015] Step S12: By using elevation filtering, retain the point cloud data within the target elevation range, and then by using local normal vector filtering, cluster the road surface points to obtain the road surface points. Extract the road feature information based on the road surface points, wherein the road feature information includes position parameters and vector information.

[0016] Step S13: Based on the extracted road feature information, establish a three-dimensional static digital twin digital platform.

[0017] Step S13 specifically includes:

[0018] Step S131: Based on the extracted road feature information, select regular spacing to encrypt the three-dimensional feature points of the road, construct a continuous regular triangular network using the triangular network algorithm, and generate a three-dimensional road surface model through texture mapping and texture parameter recognition.

[0019] Step S132: Set control parameters according to the category and geometric characteristics of the feature model, build a standard feature model library, read the coordinates, geometry and category information of the features and road markings, and construct a three-dimensional static digital twin platform.

[0020] Step S2 specifically includes:

[0021] Step S21: Set the drone's cruise route;

[0022] Step S22: Based on the set drone cruise route, take aerial photos to obtain aerial images, and perform edge detection on the road boundary lines of the aerial photos to obtain feature images containing boundary line features. Perform Hough transform on the feature images and extract boundary line features with geometric features such as straight line length, inclination rate, and parallelism as constraints. Traverse the pixels of the aerial photos, assign the value of pixels outside the boundary line to 0 according to the boundary line features, and keep the pixels inside the boundary line unchanged. Segment the road surface area in the image and generate a detection image based on the road surface area.

[0023] Step S23: Input the detected image into the trained crack recognition model to obtain the crack recognition result.

[0024] In step S21, the basic flight path of the UAV is the center line of the lane. The flight altitude, maximum flight speed, camera baseline length, and flight path interval of the UAV during inspection are as follows:

[0025] H 航 =f·P·R 航 / C 航

[0026] V 航 =K·n·V 车 / L 车

[0027] L 基 = (2-A)·P·R 航

[0028] L 间 =2·VD·V / 2

[0029] Wherein: H 航 R represents the flight altitude of the drone during inspection, f represents the focal length of the camera mounted on the drone, and R represents the focal length of the camera mounted on the drone. 航 For the camera's resolution, C 航 V represents the sensor size of the camera. 航 Where K is the maximum flight speed, n is the forward image frame length, and L is the camera scale. 车 L is the length of the vehicle body. 基 Let A be the baseline length for photography, and L be the forward overlap. 间 D represents the route spacing and D represents the lateral overlap.

[0030] The crack recognition model is a multi-layer neural network with 2,500 input nodes, two output nodes (A and B), and 71 hidden layer nodes.

[0031] Step S4 specifically includes:

[0032] Step S41: Based on the material data at the crack location, read the gradation composition, thermal conductivity, specific heat capacity, and thermal diffusivity of the asphalt mixture at the crack location;

[0033] Step S42: Determine the three-dimensional spatial information of the crack contour points based on the morphological data of the crack location, and construct the physical model of the cracked Marshall specimen;

[0034] Step S43: By configuring different frequencies and input power, simulation experiments are conducted to obtain the heating rate and uniform temperature distribution at the crack under different frequencies and input power.

[0035] Step S44: Based on the heating rate and temperature distribution uniformity at the crack, output the optimal frequency and optimal input power for thermal induction microwave thermal induction.

[0036] The construction process of the Marshall sample physical model specifically includes:

[0037] Step S421: Using the volume of a standard cylindrical Marshall specimen as the total volume, randomly generate aggregates of different particle sizes until the cumulative volume of aggregates in each interval reaches the corresponding target value;

[0038] Step S422: Based on the obtained aggregate particle size, with the constraints of not overlapping with the generated spheres and not overlapping with the specimen boundary, uniformly generate spheres in the standard Marshall specimen space to obtain the physical model of the Marshall specimen.

[0039] The particle size of the aggregate is:

[0040] d = d z +η k (d z+1 -d z )

[0041] Where: d is the particle size of the aggregate, η k Let d be a random number with a given probability distribution. z and d z+1 These represent the upper and lower limits of the aggregate particle size within the interval, respectively.

[0042] A self-healing device for microwave thermally induced cracks in asphalt pavement based on digital twins includes a memory, a processor, and a program stored in the memory, characterized in that the processor executes the method described above.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] 1. Based on a digital twin platform, this invention visualizes asphalt pavement cracks in batches, enabling real-time monitoring and intelligent repair on the digital platform. This effectively improves pavement performance, ensuring driving safety and passenger comfort. Simultaneously, this technology significantly reduces traffic accidents and congestion caused by pavement damage, resulting in a substantial improvement in the highway traffic environment.

[0045] 2. It can effectively control cracks in their early stages, significantly reducing the damage caused by cracks and thus greatly improving the self-repairing ability of asphalt pavement, achieving efficient management and maintenance of pavement damage.

[0046] 3. Based on a digital twin platform and simulation model, this invention achieves precise and characteristic repair strategies for various asphalt pavement types, different functional fillers, and cracks with varying development stages. During the repair process, the invention intelligently selects the most suitable microwave frequency and power to ensure uniform heating of the asphalt and appropriate heating time, thereby addressing the aging problem that easily occurs during asphalt heating to some extent. This refined control not only improves repair efficiency but also significantly enhances the service performance of the repaired asphalt pavement, providing a new approach for intelligent road maintenance. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of the main steps of the method of the present invention. Detailed Implementation

[0048] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0049] A method for self-healing microwave thermally induced cracks in asphalt pavement based on digital twins, such as... Figure 1 As shown, it includes:

[0050] Step S1: Establish a three-dimensional static digital twin platform for the road;

[0051] Step S1 specifically includes:

[0052] Step S11: Use a LiDAR mobile scanning system to scan the road and obtain vehicle point cloud data in the road surface coordinate system. The LiDAR mobile scanning system integrates a global navigation satellite system, an inertial navigation system, a LiDAR, and a time synchronization unit. It obtains trajectory position information through combined navigation of the global navigation satellite system and the inertial navigation system, collects point cloud data through LiDAR, and processes the trajectory data and point cloud data based on the principles of time registration and spatial registration to obtain vehicle point cloud data in the road surface coordinate system.

[0053] Specifically, in this embodiment, the SZT-R1000 vehicle-mounted LiDAR mobile scanning system is used to acquire road point cloud information. This system integrates GNSS, INS (Inertial Navigation System), scanner, panoramic camera, control unit and time synchronization unit. Its working principle is as follows: the system's trajectory and position information is obtained through the combined navigation of the global navigation satellite system and the inertial navigation system. The LiDAR collects point cloud data and processes the trajectory data and point cloud data according to the principles of time registration and spatial registration to finally obtain vehicle-mounted point cloud data in the local coordinate system.

[0054] Step S12: By using elevation filtering, retain the point cloud data within the target elevation range, and then by using local normal vector filtering, cluster the road surface points to obtain the road surface points. Extract the road feature information based on the road surface points, wherein the road feature information includes position parameters and vector information.

[0055] Specifically, in this embodiment, the process of extracting road surface point cloud information includes: 1) Elevation filtering: Arrange the points in the vertical column in descending order of elevation value, take the average elevation of the lowest 75 to 150 points as the minimum elevation, and retain the point cloud data with elevations between the minimum elevation and 2m higher than the minimum elevation; 2) Local normal vector filtering: First, scale the three-dimensional data of the point cloud, then standardize the data distribution center to the origin, and then perform principal component analysis on the processed point cloud data to calculate the point cloud normal vector and eigenvalue corresponding to the third principal component. Select points within a suitable range of the angle between the normal vector and the vertical axis, and take the filtering parameter as 3; 3) Extract road feature information: Adjust the neighborhood E value according to the point cloud density, and use the DBSCAN algorithm to cluster the point cloud into several sets for each grid. The set with the most points is the road surface point, and the other sets are non-road surface points; Input the extracted road surface point cloud into the TopoDOT point cloud data processing and application system to semi-automatically extract road vector information, position parameters, geometric parameters and texture information;

[0056] Step S13: Based on the extracted road feature information, establish a three-dimensional static digital twin platform, specifically including:

[0057] Step S131: Based on the extracted road feature information, select regular spacing to encrypt the three-dimensional feature points of the road, construct a continuous regular triangular network using the triangular network algorithm, and generate a three-dimensional road surface model through texture mapping and texture parameter recognition.

[0058] Step S132: Set control parameters according to the category and geometric characteristics of the feature model, build a standard feature model library, read the coordinates, geometry and category information of the features and road markings, and construct a three-dimensional static digital twin platform.

[0059] Specifically, in this embodiment, the process is based on Dynamo for Revit to build a 3D static digital twin platform, which includes: 1. 3D road surface modeling: Based on the extracted road geometry information, the 3D feature points of the road are densified with regular spacing. A continuous regular triangular mesh with equal base length is automatically constructed in Microstation V8 using a triangular mesh algorithm to generate a DSM (Digital Modeling System), which is then imported into Dynamo for Revit. A 3D road surface model is generated through texture mapping and texture parameter recognition. 2. Parametric ground modeling and construction of a static digital twin platform: Corresponding control parameters are set according to the category and geometric characteristics of the ground feature model. A standard ground feature model library is built in Dynamo for Revit. The coordinates, geometry, and category information of ground features and road markings obtained from point cloud data are read. A batch automatic modeling process is performed using a lofting algorithm, and the model is projected onto the 3D road surface model using the projection principle to construct a 3D static digital twin platform.

[0060] Step S2: Obtain aerial photography of the road from the drone to monitor for cracks, specifically including:

[0061] Step S21: Set the drone's patrol route. In step S21, the basic flight path of the drone is the center line of the lane. The drone's flight altitude, maximum flight speed, camera baseline length, and flight path interval during the inspection are as follows:

[0062] H 航 =f·P·R 航 / C 航

[0063] V 航 =K·n·V 车 / L 车

[0064] L 基 = (2-A)·P·R 航

[0065] L 间 =2·VD·V / 2

[0066] Wherein: H 航 R represents the flight altitude of the drone during inspection, f represents the focal length of the camera mounted on the drone, and R represents the focal length of the camera mounted on the drone. 航 For the camera's resolution, C 航 V represents the sensor size of the camera. 航 Where K is the maximum flight speed, n is the forward image frame length, and L is the camera scale. 车 L is the length of the vehicle body. 基 Let A be the baseline length for photography, and L be the forward overlap. 间 D represents the route spacing and D represents the lateral overlap.

[0067] Specifically, in this embodiment, the basic flight path of the UAV is set as the center line of the lane; in order to eliminate the occlusion effect of the vehicle on the crack image, the car is set to travel in the opposite direction of the traffic flow, and the vehicle length and driving speed are used as key constraints to limit the maximum flight speed of the UAV; the flight altitude, maximum flight speed, photography baseline length and flight path interval of the UAV during inspection are calculated using formulas (1) to (4), and the inspection cycle is 2 days.

[0068] Step S22: Based on the set drone cruise route, take aerial photos to obtain aerial images, and perform edge detection on the road boundary lines of the aerial photos to obtain feature images containing boundary line features. Perform Hough transform on the feature images and extract boundary line features with geometric features such as straight line length, inclination rate, and parallelism as constraints. Traverse the pixels of the aerial photos, assign the value of pixels outside the boundary line to 0 according to the boundary line features, and keep the pixels inside the boundary line unchanged. Segment the road surface area in the image and generate a detection image based on the road surface area.

[0069] Specifically, in this embodiment, the Prewitt operator is used to perform edge detection of road boundary lines on the aerial image to obtain a feature image containing boundary line features; then, the image is subjected to Hough transform, and boundary line features are extracted using geometric features such as line length, slope, and parallelism as constraints; the pixels of the initial image are traversed, and the pixels outside the boundary line are assigned a value of 0 according to the boundary line features, while the pixels inside the boundary line remain unchanged, removing features irrelevant to the road surface area, segmenting the road surface area in the image, and finally arraying the segmented road surface image into 50×50 sub-block images;

[0070] Step S23: Input the detected image into the trained crack recognition model to obtain the crack recognition result.

[0071] Specifically, in this embodiment, the crack recognition model uses a multi-layer neural network with 2500 input nodes, two output nodes A and B, and 71 hidden layer nodes. When nodes A and B output 1 and 0 respectively, it indicates that the sub-block image contains a crack; when nodes A and B output 0 and 1 respectively, it indicates that the sub-block image does not contain a crack. The hidden layer has 71 nodes.

[0072] The process of training a neural network is as follows: Initial values ​​are assigned to the weights *w* and biases of the connection channels between layers in the neural network; samples from the CrackForest open-source dataset are arrayed into 50×50 sub-blocks and sequentially input into the input layer, then processed through the hidden layers, reaching the output layer to complete one forward propagation, obtaining one output value of the neural network; the signal error is calculated based on the neural network output value and the expected value, and this error is used as the input signal for backpropagation back to the neural network, calculating the error of each layer sequentially; each layer of the neural network calculates the weight error Δw and bias error Δb through its own error and learning rate calculation; finally, the weights and biases between layers are updated once, completing one backpropagation; this process is repeated iteratively, feeding all arrayed images and their corresponding expectations from the dataset into the neural network for training, continuously updating the weights and biases of each layer, until a fully trained neural network is obtained;

[0073] This also includes establishing a real-time crack monitoring database for drones: A trained neural network model is used to analyze a list of sub-image blocks. When a sub-image block containing cracks is identified, a geodetic Gaussian projection forward and inverse calculation algorithm is employed. Based on the drone's focal length, the spatial coordinates of the camera center, and the angle, the true spatial physical coordinates of all pixels in the original crack image are obtained. The original crack image and the true spatial coordinates of all its pixels are then uploaded to the cloud, and a drone-monitored crack image database is established using MySQL software.

[0074] Step S3: Once a crack is detected, locate the crack position and use ground-penetrating radar to detect the crack position and obtain the crack's morphological and material data.

[0075] Then, all cracks on the road section are visualized on a digital platform. In this embodiment, the steps for visualizing cracks on a digital platform are as follows:

[0076] (1) Ground penetrating radar non-destructive testing of road surface, and establishment of depth-based crack dataset: A vehicle-mounted ground penetrating radar is used to perform non-destructive testing of the road surface. Each time the ground penetrating radar transmits and receives electromagnetic waves, it can generate an A-scan image. The A-scan images are arranged along the road direction. Secondly, the mysql-connector-python data connection library and OpenCV image processing library are installed in Python to obtain crack image coordinate information from the cloud crack image database. The A-scan images are selected and labeled to extract the A-scan image set corresponding to each crack. Finally, each image set is filled with black and white based on the amplitude. After interpolating adjacent images in pairs, the depth slice is obtained to obtain the C-scan image set corresponding to each crack.

[0077] (2) Establishing a crack model: The C-scan image set format is converted to HSV format. The weighted average method is used to assign different weights to the three components of each pixel in the crack image. The image is then grayscaled. Median filtering is used to remove noise from the image and grayscale enhancement is performed. Then, different thresholding transformation methods, such as taking one non-zero values, are used to transform the grayscale image into a black and white binary image set. Next, the Canny edge detection algorithm in the OpenCV image processing library is called to read the contour of the C-scan image. The contour points of adjacent C-scan images are connected based on the principle of minimizing distance. The connection is converted into contour point form using the principle of equal distance to obtain a three-dimensional contour model. Contour points that are not connected to the road surface in the three-dimensional contour model are deleted to obtain the crack contour coordinate point set based on the C-scan image set. Finally, the crack model is reconstructed using the triangulation algorithm.

[0078] (3) Automatic visualization of cracks on the digital platform: The planar information of the road surface contour points of the reconstructed model is matched with the planar information of the crack coordinate points in the aerial image to determine the model position; then, the crack model is embedded into the 3D road model, and the overlapping models are superimposed to obtain the complete crack model; finally, Revit secondary development is used to automatically place crack models in batches, and the construction of the 3D road crack digital model is completed by traversing the crack images, realizing the automatic visualization of cracks on the digital platform.

[0079] Step S4: Based on the morphological and material data of the crack location, and combined with a three-dimensional static digital twin platform, the optimal frequency and power are determined through microwave thermally induced asphalt self-healing process simulation analysis. Specifically, this includes:

[0080] Step S41: Based on the material data at the crack location, read the gradation composition, thermal conductivity, specific heat capacity, and thermal diffusivity of the asphalt mixture at the crack location;

[0081] Step S42: Determine the three-dimensional spatial information of the crack contour points based on the morphological data of the crack location, and construct a physical model of the cracked Marshall specimen. The construction process of the physical model of the Marshall specimen specifically includes:

[0082] Step S421: Using the volume of a standard cylindrical Marshall specimen as the total volume, randomly generate aggregates of different particle sizes until the cumulative volume of aggregates in each interval reaches the corresponding target value;

[0083] Step S422: Based on the obtained aggregate particle size, with the constraints of not overlapping with the generated spheres and not overlapping with the specimen boundary, uniformly generate spheres in the standard Marshall specimen space to obtain the physical model of the Marshall specimen.

[0084] The particle size of the aggregate is:

[0085] d = d z +ηk (d z+1 -d z )

[0086] Where: d is the particle size of the aggregate, η k Let d be a random number with a given probability distribution. z and d z+1 These represent the upper and lower limits of the aggregate particle size within the interval, respectively.

[0087] Step S43: By configuring different frequencies and input power, simulation experiments are conducted to obtain the heating rate and uniform temperature distribution at the crack under different frequencies and input power.

[0088] Step S44: Based on the heating rate and temperature distribution uniformity at the crack, output the optimal frequency and optimal input power for thermal induction microwave thermal induction.

[0089] In this embodiment, the step includes two parts. First, based on the secondary development of Dynamo for Revit using Python, data synchronization and interaction between the digital platform and each module are realized to construct a three-dimensional dynamic digital twin platform, including:

[0090] (1) Automatically model cracks from crack images obtained from cloud crack databases to digital platforms;

[0091] (2) The spatial three-dimensional information of the crack profile, the asphalt mixture gradation composition at the crack, and the thermal conductivity, specific heat capacity and thermal diffusivity information of aggregates and functional fillers are imported into the three-dimensional aggregate model established by ABAQUS and the microwave heating conduction model established by COMSOL Multiphysics; and the obtained optimal frequency and input power are interacted with the digital platform.

[0092] (3) Visualize the repaired crack status on the digital platform and provide crack early warning information.

[0093] Then, using ABAQUS finite element software and COMSOL Multiphysics simulation software, a microwave thermal induction simulation model was established to obtain the optimal microwave thermal induction frequency and input power, including:

[0094] (1) Based on the secondary development of Revit, the data interaction between the digital model and the simulation model is carried out, and the gradation composition of the asphalt mixture at the location of the crack, the physical parameters of the corresponding aggregate and thermally induced functional filler, and the three-dimensional spatial information of the crack profile are read.

[0095] (2) Establishing a physical model of a three-dimensional aggregate crack Marshall specimen for asphalt concrete: ① Capturing aggregate particle size for each particle size range: Using the volume of a standard cylindrical Marshall specimen as the total volume, calculate the required volume for each particle size range based on the asphalt mixture gradation composition; use the Monte Carlo method to randomly generate particle sizes within the range above and below the range according to formula (5), and then calculate the volume of spherical aggregates; repeat this step and accumulate the volume of the generated spherical aggregates; stop when the total volume generated in the range reaches the required volume for the current range; ② Establishing a geometric model: Based on the particle size information generated in the first step, iterate randomly at multiple points in the space of the standard cylindrical Marshall specimen. Determine whether the aggregate overlaps with the existing aggregate and specimen boundaries; if not, proceed with subsequent aggregation placement; if overlapping, randomly move to a new position for aggregation until the requirements are met. When the aggregate movement exceeds the upper limit, the placement point is regenerated; the three-dimensional spatial information of the crack from the digital platform to the simulation software is transformed into relative coordinates and input into the Marshall specimen space; aggregate spheres within the crack space and overlapping with the crack boundary are removed to obtain the geometric model of the cracked Marshall specimen; ③ Establish the physical model: output the geometric information of the spherical aggregates of the geometric model of the cracked Marshall specimen with spatial coordinates and diameter; then select the spherical geometric information of the interval corresponding to the particle size of each aggregate and thermally induced functional filler, and input its corresponding physical parameters during the generation of the physical model; generate the physical model of the three-dimensional aggregate cracked Marshall specimen of asphalt concrete in ABAQUS through the parametric Python script;

[0096] (3) Establishing a microwave heat conduction model: A microwave model was established in the COMSOL Multiphysics application library. Microwaves were emitted into the cavity through a waveguide port placed on the side of the microwave oven. The microwave oven and waveguide were filled with air, and the oven walls and waveguide walls were coated with copper. The sample was placed on a turntable at the bottom of the microwave oven. The microwave oven was 300×300×300mm in size, the waveguide was 50×78×18mm in size, the glass plate radius was 80mm, the initial temperature of the cracked Marshall sample model and the air was 25℃, and the sample was heated for 100s.

[0097] (4) Microwave thermal induction simulation to obtain the optimal frequency: The spatial information of all spheres in the cracked Marshall specimen physical model obtained in ABAQUS, as well as the physical parameters such as thermal conductivity, specific heat capacity and thermal diffusivity of the corresponding aggregates and functional fillers, are imported into COMSOL Multiphysics and combined with the microwave heating model to perform microwave thermal conduction simulation. The heating rate and temperature distribution uniformity at the crack are used as evaluation indicators to obtain the optimal frequency and input power of microwave thermal induction. Finally, the optimal thermal induction data are returned to the digital platform.

[0098] Step S5: Repair the cracks in the road using the optimal frequency and power determined by the simulation analysis process.

[0099] In this embodiment, the microwave thermal induction method for repairing cracks is as follows:

[0100] Based on the crack location information from the digital platform, a warning sign is placed 50 meters from the target repair point. Upon reaching the target repair point, the lane crack is within the microwave's effective range. The microwave regulator is activated to repair the crack using the optimal frequency and power imported by the digital platform. The crack is detected every 5 seconds using ground-penetrating radar (GPR) in S3 mode. When crack healing is detected, microwave heating is stopped and the crack model is deleted from the digital platform. If the heating time exceeds 100 seconds, and the crack repair volume reaches 70%, heating is stopped and a repair model is created to replace the crack model. If the heating time reaches 150 seconds, heating is stopped for 3000 seconds, followed by a second heating cycle. If the crack volume reaches 70%, heating is stopped and a repair model is created to replace the crack model. If the second heating time reaches 50 seconds, heating is stopped and a warning is sent to the digital platform, transmitting images of crack changes during the heating process. After heating stops, the warning sign is retrieved, and the repair continues at the next target crack repair point, achieving real-time repair of road cracks.

[0101] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins, characterized in that, include: Step S1: Establish a three-dimensional static digital twin platform for the road; Step S2: Obtain aerial photographs of the road taken by the drone to monitor for the presence of cracks; Step S3: Once a crack is detected, locate the crack position and use ground-penetrating radar to detect the crack position and obtain crack morphology and material data. Step S4: Based on the morphological and material data of the crack location, and combined with the three-dimensional static digital twin platform, the optimal frequency and optimal power are determined by microwave thermally induced asphalt self-healing process simulation analysis. Step S5: Repair the cracks in the road using the optimal frequency and power determined by the simulation analysis process; Step S1 specifically includes: Step S11: Use a LiDAR mobile scanning system to scan the road and obtain vehicle point cloud data in the road surface coordinate system. The LiDAR mobile scanning system integrates a global navigation satellite system, an inertial navigation system, a LiDAR, and a time synchronization unit. It obtains trajectory position information through combined navigation of the global navigation satellite system and the inertial navigation system, collects point cloud data through LiDAR, and processes the trajectory data and point cloud data based on the principles of time registration and spatial registration to obtain vehicle point cloud data in the road surface coordinate system. Step S12: By using elevation filtering, retain the point cloud data within the target elevation range, and then by using local normal vector filtering, cluster the road surface points to obtain the road surface points. Extract the road feature information based on the road surface points, wherein the road feature information includes position parameters and vector information. Step S13: Based on the extracted road feature information, establish a three-dimensional static digital twin digital platform; Step S13 specifically includes: Step S131: Based on the extracted road feature information, select regular spacing to encrypt the three-dimensional feature points of the road, construct a continuous regular triangular network using the triangular network algorithm, and generate a three-dimensional road surface model through texture mapping and texture parameter recognition. Step S132: Set control parameters according to the category and geometric characteristics of the feature model, build a standard feature model library, read the coordinates, geometry and category information of the features and road markings, and construct a three-dimensional static digital twin platform; Step S4 specifically includes: Step S41: Based on the material data at the crack location, read the gradation composition, thermal conductivity, specific heat capacity, and thermal diffusivity of the asphalt mixture at the crack location; Step S42: Determine the three-dimensional spatial information of the crack contour points based on the morphological data of the crack location, and construct the physical model of the cracked Marshall specimen; Step S43: Simulation experiments were conducted by configuring different frequencies and input powers to obtain the heating rate and uniform temperature distribution at the crack under different frequencies and input powers; Step S44: Based on the heating rate and temperature distribution uniformity at the crack, output the optimal frequency and optimal input power for thermal induction microwave thermal induction; The construction process of the Marshall sample physical model specifically includes: Step S421: Using the volume of a standard cylindrical Marshall specimen as the total volume, randomly generate aggregates of different particle sizes until the cumulative volume of aggregates in each interval reaches the corresponding target value; Step S422: Based on the obtained aggregate particle size, with the constraint that it does not overlap with the already generated spheres and does not overlap with the specimen boundary, spheres are uniformly generated in the standard Marshall specimen space to obtain the Marshall specimen physical model. The constraint that it does not overlap with the already generated spheres and does not overlap with the specimen boundary includes removing aggregate spheres that are in the crack space or overlap with the crack boundary.

2. The method for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins according to claim 1, characterized in that, Step S2 specifically includes: Step S21: Set the drone's cruise route; Step S22: Based on the set drone cruise route, take aerial photos to obtain aerial images, and perform edge detection on the road boundary lines of the aerial photos to obtain feature images containing boundary line features. Perform Hough transform on the feature images and extract boundary line features with geometric morphological features including straight line length, inclination rate, and parallelism as constraints. Traverse the pixels of the aerial photos, assign the value of pixels outside the boundary line to 0 according to the boundary line features, and keep the pixels inside the boundary line unchanged. Segment the road surface area in the image and generate a detection image based on the road surface area. Step S23: Input the detected image into the trained crack recognition model to obtain the crack recognition result.

3. The method for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins according to claim 2, characterized in that, The crack recognition model is a multi-layer neural network with 2,500 input nodes, two output nodes (A and B), and 71 hidden layer nodes.

4. The method for self-healing of microwave thermally induced cracks in asphalt pavement based on digital twins according to claim 1, characterized in that, The particle size of the aggregate is: d = d z + η k ( d z+1 - d z ) in: d The particle size of the aggregate, η k For a random number with a given probability distribution, d z and d z+1 These represent the upper and lower limits of the aggregate particle size within the interval, respectively.

5. A self-healing device for microwave thermally induced cracks in asphalt pavement based on digital twins, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-4.

Images