Heterogeneous road loose disease modeling and detecting method, device, equipment and medium

Abstract

The invention discloses a heterogeneous road loose disease modeling and detection method, device, equipment and medium, and the method comprises the steps: building a road structure three-dimensional matrix model based on the actual geometric dimension of a road, setting a generation region in the road structure three-dimensional matrix model, initializing a material identifier, generating an initial heterogeneous road structure model, and carrying out the modeling and detection of the heterogeneous road loose disease. Generating a self-adaptive irregular loose disease in a target horizon of the initial heterogeneous road structure model, forming a target heterogeneous road structure model containing the loose disease, and generating radar echo simulation data of the target heterogeneous road structure model, inputting the radar echo simulation data into a deep learning disease detection model for loose disease detection; according to the method, refined and real modeling is carried out on hidden loose diseases, radar forward simulation data containing complex heterogeneous backgrounds and random-shaped loose diseases can be generated, real data samples are provided for disease detection, and the recognition precision of the loose diseases can be improved.

Description

Technical Field

[0001] This invention relates to the field of road engineering, and in particular to a method, apparatus, equipment and medium for modeling and detecting loose defects in heterogeneous roads. Background Technology

[0002] As a core component of transportation infrastructure, road engineering's structural integrity directly impacts traffic safety and operational lifespan. Accurate detection and early warning of hidden defects within roads are crucial aspects of road maintenance. Among these, loosening defects are common hidden defects caused by material aging, load-bearing effects, and environmental erosion during road operation. These defects manifest as material debonding within the road structural layers and loosening of aggregate bonds. If not detected and addressed promptly, they can gradually lead to serious defects such as potholes and subsidence, causing road safety accidents. With the integration of non-destructive testing technology and artificial intelligence, deep learning-based intelligent road defect identification technology using ground-penetrating radar has become a mainstream development direction. However, real-world radar image data of road defects is scarce, especially for common but complex defects like "loosening." Loosening defects often present as irregular, branching, sheet-like, or network-like structures, significantly different from traditional regular voids.

[0003] Currently, existing methods cannot meet the actual needs of loose road defects detection. Firstly, existing modeling methods can only simulate regular or simple irregular cavities, failing to accurately reproduce the natural irregular morphology and spatial distribution characteristics of loose defects, resulting in significant deviations from the actual morphology of loose defects in real roads. Secondly, due to the lack of real and effective simulation data of loose defects, deep learning defect detection models trained on existing datasets cannot effectively learn and distinguish the unique characteristics of loose defects in real roads, such as disordered reflection signals, enhanced energy scattering, and discontinuous phase axes, in ground-penetrating radar images. If only cavity defect data is used to train the model, it will lead to serious omissions and misjudgments of loose defects in actual engineering inspections, restricting the comprehensiveness and accuracy of intelligent identification systems for internal road defects and failing to provide reliable technical support for maintenance decisions regarding loose road defects. Summary of the Invention

[0004] The main objective of this invention is to provide a method, apparatus, equipment, and medium for modeling and detecting loose defects on heterogeneous roads. This invention aims to solve the technical problems of existing technologies being unable to realistically simulate the geometric shape and spatial distribution of loose defects, unable to accurately generate targeted simulation data, and having insufficient training data for loose defects, resulting in low accuracy in defect identification.

[0005] To achieve the above objectives, this invention provides a method for modeling and detecting loose road defects in heterogeneous roads, the method comprising the following steps: A three-dimensional matrix model of the road structure is constructed based on the actual geometric dimensions of the road. In the three-dimensional matrix model of the road structure, a generation region is set and material identifiers are initialized to generate an initial heterogeneous road structure model, which includes aggregates, binders and voids. Adaptive irregular loose defects are generated within the target layer of the initial heterogeneous road structure model to form a target heterogeneous road structure model containing loose defects. Radar echo simulation data of the target heterogeneous road structure model is generated by using a radar echo simulation model of heterogeneous road structure, and the radar echo simulation data is input into a pre-trained deep learning disease detection model for loose disease detection.

[0006] Optionally, the step of generating adaptive irregular loose defects within the target layer of the initial heterogeneous road structure model to form a target heterogeneous road structure model containing loose defects includes: One or more smooth critical line paths are randomly generated within the target layer of the initial heterogeneous road structure model; Random radius expansion is performed along the smooth keyline path to generate an irregular loose region contour. The outline of the irregular loose region is rasterized; The identifiers representing the binder within the rasterized irregular loose region outline are updated to identifiers representing the loose material, thus forming a target heterogeneous road structure model containing loose defects.

[0007] Optionally, the step of randomly generating one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model includes: The starting point of loose disease is randomly selected within the target layer of the initial heterogeneous road structure model; Based on the starting point, a sequence of keyline control points is generated using a stepping method with random directional perturbation; The extension step size and curvature of the key line control point sequence are adjusted based on the preset step size and the maximum value of the preset direction disturbance to obtain the adjusted key line control point sequence. The adjusted key line control point sequence is used to characterize the main path of disease in different forms. The adjusted keyline control point sequence is smoothed using the B-spline interpolation algorithm to obtain a smoothed keyline path.

[0008] Optionally, the step of randomly expanding the radius along the smooth keyline path to generate an irregular loose region contour includes: Dense sampling is performed along the smooth keyline path to obtain a set of sampling points; The difference in the expansion radius of each sampling point in the sampling point set is calculated based on random numbers, and the minimum expansion radius of the loose disease is added to the difference in the expansion radius to obtain the initial random expansion radius of each sampling point; Based on the material properties of the target layer, the initial random expansion radius is adjusted to obtain the random expansion radius corresponding to each sampling point in the sampling point set; A circular region is generated with each sampling point as the center and the corresponding random expansion radius as the radius. The union of all circular regions is used as the initial contour of the loose region. The initial contour of the loose region is smoothed by morphological closing operation to obtain the contour of the irregular loose region.

[0009] Optionally, before rasterizing the irregular loose region outline, the method further includes: Verify whether the geometric features of the irregular loose region outline meet the geometric constraints of the layer. The geometric features include area, aspect ratio and / or concavity coefficient. The geometric constraints include area constraints and / or morphological constraints. If the geometric features of the irregular loose region outline satisfy the geometric constraints of the layer, then the step of rasterizing the irregular loose region outline is performed. If the geometric features of the irregular loose region contour do not meet the geometric constraints of the layer, then return to the step of randomly generating one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model.

[0010] Optionally, the step of rasterizing the outline of the irregular loose region includes: Obtain the raster resolution parameters corresponding to the three-dimensional matrix model of the road structure; The target layer is divided into two-dimensional matrix raster units according to the raster resolution parameters; Map the coordinates of the irregular loose region outline to the two-dimensional matrix grid cell; Mark the grid cells containing the outline coordinates of the irregular loose region to complete the rasterization of the irregular loose region outline.

[0011] Optionally, the step of generating radar echo simulation data of the target heterogeneous road structure model through a radar echo simulation model of a heterogeneous road structure, and inputting the radar echo simulation data into a pre-trained deep learning disease detection model for loose disease detection, includes: Electromagnetic property parameters are configured for the loose materials in the target heterogeneous road structure model. Electromagnetic simulation calculations are performed on the configured target heterogeneous road structure model using a radar echo simulation model of heterogeneous road structure to generate radar echo simulation data. The radar echo simulation data is input into a pre-trained deep learning disease detection model for feature extraction to obtain disease feature data. Based on the disease characteristic data, detect whether there are loose diseases in the target heterogeneous road structure model, and output the loose disease detection results.

[0012] Furthermore, to achieve the above objectives, this invention also proposes a heterogeneous road loosening disease modeling and detection device that applies the heterogeneous road loosening disease modeling and detection method described above, the heterogeneous road loosening disease modeling and detection device comprising: The structural model building module is used to construct a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. The heterogeneous model construction module is used to set the generation region and initialize material identifiers in the three-dimensional matrix model of the road structure to generate an initial heterogeneous road structure model, wherein the initial heterogeneous road structure model contains aggregates, binders and voids; The loose disease model construction module is used to generate adaptive irregular loose diseases within the target layer of the initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose diseases. The defect detection module is used to generate radar echo simulation data of the target heterogeneous road structure model through the radar echo simulation model of the heterogeneous road structure, and input the radar echo simulation data into the pre-trained deep learning defect detection model for loose defect detection.

[0013] Furthermore, to achieve the above objectives, this application also proposes a non-homogeneous road loose disease modeling and detection device, the device comprising: a memory, a processor, and a non-homogeneous road loose disease modeling and detection program stored in the memory, the processor being used to run the non-homogeneous road loose disease modeling and detection program, the computer program being configured to implement the steps of the non-homogeneous road loose disease modeling and detection method as described above.

[0014] In addition, to achieve the above objectives, this application also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the non-homogeneous road loose disease modeling and detection method described above.

[0015] This invention constructs a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. A generation region is set within this model, and material identifiers are initialized to generate an initial heterogeneous road structure model. This initial heterogeneous road structure model includes aggregates, binders, and voids. Adaptive irregular loose defects are generated within the target layer of this initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose defects. Radar echo simulation data of the target heterogeneous road structure model is generated using a heterogeneous road structure radar echo simulation model, and this radar echo simulation data is input into a pre-programmed... This invention utilizes a trained deep learning-based disease detection model to detect loose road defects. By effectively restoring the heterogeneity of road structures and the true characteristics of loose defects, it overcomes the limitations of traditional homogeneous models and regular defect simulations, improving the realism of heterogeneous road structure models and achieving realistic modeling of loose defects. Radar echo simulation significantly reduces the cost and improves detection efficiency. The deep learning-based disease detection model enables accurate identification of loose defects, effectively uncovering their characteristic patterns and providing technical support for early warning and precise maintenance of road defects, thus ensuring the safety, stability, and service life of road structures. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of the heterogeneous road loose disease modeling and detection equipment involved in the hardware operating environment of the embodiment of the present invention; Figure 2 This is a flowchart illustrating the first embodiment of the method for modeling and detecting loose defects in heterogeneous roads according to the present invention. Figure 3 This is a flowchart illustrating the second embodiment of the method for modeling and detecting loose defects in heterogeneous roads according to the present invention. Figure 4 This is a structural block diagram of the first embodiment of the non-homogeneous road loose disease modeling and detection device of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0020] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of the heterogeneous road loose disease modeling and detection equipment involved in the hardware operating environment of the embodiment of the present invention.

[0021] like Figure 1 As shown, the non-homogeneous road loose disease modeling and detection device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen and an input unit such as a keyboard; the user interface 1003 may also include standard wired and wireless interfaces. The network interface 1004 may optionally include standard wired and wireless interfaces (such as Wireless-Fidelity (Wi-Fi) interfaces). The memory 1005 may be high-speed random access memory (RAM) or stable non-volatile memory (NVM), such as a disk storage device. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0022] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the modeling and detection equipment for loose defects in heterogeneous roads. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0023] like Figure 1 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a non-homogeneous road loose disease modeling and detection program.

[0024] exist Figure 1 In the heterogeneous road loose disease modeling and detection device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and memory 1005 in the heterogeneous road loose disease modeling and detection device of the present invention can be set in the heterogeneous road loose disease modeling and detection device. The heterogeneous road loose disease modeling and detection device calls the heterogeneous road loose disease modeling and detection program stored in the memory 1005 through the processor 1001 and executes the heterogeneous road loose disease modeling and detection method provided in the embodiment of the present invention.

[0025] This invention provides a method for modeling and detecting loose defects in heterogeneous roads, referring to... Figure 2 , Figure 2 This is a flowchart illustrating the first embodiment of the method for modeling and detecting loose defects in heterogeneous roads according to the present invention.

[0026] In this embodiment, the method for modeling and detecting loose defects in heterogeneous roads includes the following steps: Step S10: Construct a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road.

[0027] It should be understood that the executing entity of this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a terminal electronic device capable of performing the above functions. The following description uses a non-homogeneous road loose disease modeling and detection device (hereinafter referred to as the detection device) as an example to illustrate this embodiment and the following embodiments.

[0028] It should be noted that the actual geometric dimensions of a road refer to its real physical dimensions in a real-world scenario, including the road's total length, width, thickness of each structural layer (such as surface layer, base layer, and subbase layer), pavement slope, cross slope, and other geometric parameters that can be obtained through on-site measurements.

[0029] A three-dimensional matrix model of a road structure can be a three-dimensional mesh matrix model constructed using numerical discretization methods based on the actual geometric dimensions of the road. Each mesh cell in the matrix corresponds to a tiny region in the actual space of the road and can be used to characterize the geometric location and subsequent material properties of that region.

[0030] Step S20: Set the generation region and initialize the material identifier in the three-dimensional matrix model of the road structure to generate the initial heterogeneous road structure model.

[0031] It should be noted that the generation area can be a specific three-dimensional spatial range pre-defined in the three-dimensional matrix model of the road structure for generating road materials (aggregates, binders, voids), usually covering the main structural layers of the road (such as surface layer and base layer), for material distribution simulation.

[0032] Material identifiers are unique codes (such as numerical codes or character codes) used to distinguish different road materials. Each identifier corresponds to a material and can be embedded in the grid cells of a three-dimensional matrix to achieve rapid identification and differentiation of the material properties of the grid cells.

[0033] The initial heterogeneous road structure model can be a three-dimensional model that contains the main components of the real road (aggregates, binders, and voids) after the material identifiers are initialized and the material distribution is assigned based on the three-dimensional matrix model of the road structure. It can simulate the heterogeneity of the road material distribution (i.e., the non-uniform distribution of materials in space).

[0034] It should be noted that the initial heterogeneous road structure model includes aggregates, binders, and voids. Aggregates are the skeleton material in the road structure, mainly composed of granular materials such as crushed stone and gravel, bearing the load of the road and being a core component of the road structure's strength. Binders are adhesive materials (such as asphalt and cement) used to bind the aggregates, connecting the dispersed aggregates into a whole, filling the gaps between them, and improving the overall integrity and stability of the road structure. Voids are naturally formed blank areas between road materials that are not filled by aggregates and binders; they are an important manifestation of road heterogeneity and a significant cause of loosening defects.

[0035] In some embodiments, the detection equipment delineates the material generation area in a three-dimensional matrix model based on road structure design drawings and actual detection results, clarifies the three-dimensional coordinate range of the area, ensures coverage of structural layers prone to loosening defects in the road, and excludes non-target areas such as road edges and subgrades; assigns unique material identifiers to aggregates, binders, and voids respectively, establishes a correspondence between identifiers and material types, embeds the identifiers into each grid cell of the three-dimensional matrix, and completes the global initialization of the identifiers; based on the proportion and distribution law of real road materials, a random generation algorithm (such as the Monte Carlo algorithm) is used to assign corresponding materials to each grid cell in the generation area, simulates the uneven distribution of aggregates, binders, and voids, and finally forms an initial heterogeneous road structure model, thereby achieving accurate simulation of road heterogeneity, restoring the material composition and distribution characteristics of real roads, and breaking the limitations of traditional homogeneous models.

[0036] Step S30: Generate adaptive irregular loose defects within the target layer of the initial heterogeneous road structure model to form a target heterogeneous road structure model containing loose defects.

[0037] It should be noted that the target layer can be a specific structural layer (such as asphalt surface layer or base course) in the initial heterogeneous road structure model that is prone to loosening defects. Due to long-term exposure to vehicle loads and environmental factors, the target layer is prone to aggregate and binder bonding failure, resulting in loosening defects.

[0038] Adaptive irregular loose defects are loose defects that adaptively adjust their shape, size, distribution location, and density based on parameters such as the material distribution characteristics, thickness, and structural strength of the target layer. Their morphology is highly consistent with the irregularity (no fixed shape, blurred boundaries) of loose defects in real roads, rather than artificially preset regular shapes.

[0039] It should be noted that the target heterogeneous road structure model containing loose defects refers to the complete three-dimensional model obtained by embedding adaptive irregular loose defects within the target layer of the initial heterogeneous road structure model. This model retains the heterogeneous material distribution characteristics of the road while also including realistically simulated loose defects. Loose defects refer to defects in road structures caused by factors such as aging, fatigue, and water damage of the binder, leading to the failure of the bond between aggregates and binder, resulting in loosening and detachment of aggregates, increased voids, and a decrease in the overall structural integrity.

[0040] In some embodiments, the detection device locates the target layer in the initial heterogeneous road structure model according to the road structure design and the occurrence pattern of defects, and clarifies the three-dimensional range, thickness and material distribution characteristics of the target layer; based on the statistical data of real loose defects, it sets the core parameters of the defects (e.g., minimum / maximum size, distribution density, degree of looseness, etc.), and the parameter settings are adaptively matched with the material distribution and structural strength of the target layer (e.g., in areas with dense material voids, the probability of defect occurrence is higher and the degree of looseness is greater); using an adaptive generation algorithm, combined with the material properties of the grid cells of the target layer, it randomly generates irregularly shaped loose defect areas within the target layer, and simulates loose defects by modifying the material identifiers of the corresponding grid cells (e.g., changing the binder identifier to the void identifier to simulate bond failure).

[0041] It is understandable that this embodiment solves the problem of the traditional rule-based disease simulation not matching reality by accurately simulating the irregularity and randomness of loose diseases in real roads; through an adaptive generation method, the distribution and morphology of diseases are matched with the material characteristics and structural state of the target layer, thereby improving the realism and pertinence of disease simulation.

[0042] Step S40: Generate radar echo simulation data of the target heterogeneous road structure model through the radar echo simulation model of the heterogeneous road structure, and input the radar echo simulation data into the pre-trained deep learning disease detection model for loose disease detection.

[0043] It should be noted that the mathematical model built on electromagnetic propagation theory to simulate the propagation, reflection, and scattering of radar waves in road structures can generate corresponding radar echo signals based on the material properties (dielectric constant, conductivity) and geometric characteristics of the road structure.

[0044] Radar echo simulation data can be data that simulates the radar detection process through a radar echo simulation model of a non-homogeneous road structure, and outputs data with the same format as the real radar detection signal. It includes the reflection signal characteristics of radar waves at different material interfaces (such as aggregates and binders, binders and voids, loose areas and normal areas), and is a data source for disease detection.

[0045] It should be noted that the pre-trained deep learning disease detection model can be a deep learning model (e.g., based on CNN, Transformer, etc.) trained using real road radar echo data and disease labels, which has the ability to extract, identify, and locate loose disease features. After training, it can be used for disease detection using radar echo data without retraining.

[0046] In some embodiments, the detection device can preprocess the generated radar echo simulation data by noise reduction, normalization, and signal enhancement to remove interference signals and improve data quality. The preprocessed radar echo simulation data is then input into a pre-trained deep learning disease detection model. The model extracts disease features (such as echo amplitude and phase changes) from the data to identify the location, extent, and degree of looseness of the disease and outputs the detection results. The detection results output by the model are compared with the actual location and extent of the loose disease in the target model to verify the detection accuracy. If the accuracy is not up to standard, the deep learning model can be fine-tuned or the simulation data can be regenerated for supplementary training.

[0047] It is understandable that this embodiment constructs a heterogeneous road structure model containing loose defects, thus generating a large amount of realistic radar echo data without the need for extensive on-site radar detection, reducing the cost and workload of on-site detection. By utilizing a pre-trained deep learning model, it achieves rapid and accurate identification and location of loose defects, significantly improving detection efficiency and reducing human error compared to traditional manual detection. At the same time, by combining simulation data with real data, the detection performance of the deep learning model can be further optimized, improving the reliability of defect detection.

[0048] Furthermore, to improve the accuracy of disease feature discovery, step S40 above may include: Step S401: Configure electromagnetic property parameters for the loose materials in the target heterogeneous road structure model, and perform electromagnetic simulation calculations on the configured target heterogeneous road structure model through the heterogeneous road structure radar echo simulation model to generate radar echo simulation data.

[0049] It should be noted that loose material can be the material in the area of ​​loose disease in the target heterogeneous road structure model. It is mainly manifested as a mixture of loosely distributed aggregate, increased voids and a small amount of residual binder after the bonding between aggregate and binder fails. Its physical state is significantly different from that of normal road material.

[0050] In practical implementation, the electromagnetic property parameters of the loose material (which may include relative permittivity and conductivity) should be set to values ​​between those of a well-bonded material and air to simulate its partial debonding and increased porosity. This property value can be defined in the material configuration file for subsequent radar forward modeling.

[0051] Step S402: Input the radar echo simulation data into the pre-trained deep learning disease detection model for feature extraction to obtain disease feature data.

[0052] It should be noted that the disease characteristic data can be quantitative data that can characterize loose diseases after feature extraction through deep learning models, including feature vectors, signal feature maps, etc., which can intuitively reflect the location, range and degree of looseness of loose diseases.

[0053] In some embodiments, the detection device can batch input preprocessed radar echo simulation data into a deep learning model. The model extracts basic signal features (such as echo amplitude and propagation time) through shallow networks and extracts higher-order defect features (such as amplitude mutation patterns and phase anomaly features) through deep networks, thus completing feature selection and quantification.

[0054] Step S403: Based on the disease feature data, detect whether there are loose diseases in the target heterogeneous road structure model, and output the loose disease detection results.

[0055] In some embodiments, the processed disease feature data is input into the classification and recognition layer of a pre-trained deep learning disease detection model. The classification and recognition layer is configured to infer and judge the disease feature data based on the trained recognition rules, analyze the matching degree between the feature data and loose disease features, determine whether loose diseases exist in the target model based on the matching degree of the feature data, calculate the three-dimensional coordinate range of the disease area (corresponding to the grid cell coordinates of the target model), quantify the looseness level of the disease (e.g., mild, moderate, severe) based on the intensity of the feature data, calculate the confidence level of the detection result, and output the loose disease detection result.

[0056] This embodiment constructs a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. A generation region is set within this model, and material identifiers are initialized to generate an initial heterogeneous road structure model. This initial heterogeneous road structure model includes aggregates, binders, and voids. Adaptive irregular loose defects are generated within the target layer of this initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose defects. Radar echo simulation data of the target heterogeneous road structure model is generated using a heterogeneous road structure radar echo simulation model, and this radar echo simulation data is input into a pre-programmed... The trained deep learning-based disease detection model is used to detect loose road defects. This embodiment effectively restores the heterogeneity of road structures and the true characteristics of loose defects, breaking through the limitations of traditional homogeneous models and regular defect simulations, improving the realism of heterogeneous road structure models, and achieving realistic modeling of loose defects. Radar echo simulation significantly reduces the cost of defect detection and improves detection efficiency. The deep learning-based disease detection model enables accurate identification of defects, effectively uncovering the characteristic patterns of loose defects, providing technical support for early warning and precise maintenance of road defects, and ensuring the safety, stability, and service life of road structures.

[0057] refer to Figure 3 , Figure 3 This is a flowchart illustrating the second embodiment of the method for modeling and detecting loose defects in heterogeneous roads according to the present invention.

[0058] Based on the first embodiment described above, in this embodiment, step S30 further includes: Step S301: Randomly generate one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model.

[0059] It should be noted that the smooth critical line path can be a continuous curve or broken line path that is randomly generated within the target layer and has no obvious inflection point. It serves as the core baseline for the generation of loose disease areas. The direction, length, and curvature of the smooth critical line path are random and conform to the material distribution characteristics of the target layer, and are used to characterize the overall extension direction of loose disease.

[0060] In some embodiments, the detection device may employ a random generation algorithm to randomly select the start and end points of each path within the three-dimensional range of the target layer. Priority may be given to areas with dense material voids and low binder content as the starting point (aligning with the characteristics of areas prone to actual defects). Based on the start and end points, a continuous and smooth critical line path is generated using a curve generation algorithm (e.g., Bézier curve, spline curve, etc.) to avoid sharp inflection points. The generated path is then optimized to ensure it lies entirely within the target layer and is compatible with the boundary and material distribution characteristics of the target layer.

[0061] Furthermore, in order to accurately reproduce the morphology of the actual loose disease trunk path and ensure that the smooth path precisely matches the target layer boundary, the above step S301 may include: Step S3011: Randomly select the starting point of loose disease within the target layer of the initial heterogeneous road structure model; Step S3012: Based on the starting point, generate a sequence of keyline control points using a stepping method with random directional perturbation; Step S3013: Adjust the extension step size and curvature of the key line control point sequence based on the preset step size and the maximum value of the preset direction disturbance to obtain the adjusted key line control point sequence. The adjusted key line control point sequence is used to characterize the main path of disease in different forms. Step S3014: Use the B-spline interpolation algorithm to smooth the adjusted keyline control point sequence to obtain a smoothed keyline path.

[0062] Understandably, loose material defects typically develop along weak zones or stress concentration lines. To simulate this characteristic, this embodiment generates a critical line path that characterizes the "trunk" of defect development.

[0063] 1. Starting point selection: within the effective spatial range of the target layer (e.g., the base layer). Within, a starting point is randomly and uniformly selected. .

[0064] 2. Control point sequence generation: From Starting from this point, a random walk is performed to generate a sequence of keyline control points. .

[0065] The first keyline control point sequence Each control point satisfies: in, To preset the step size, Let k be the direction angle at step k. The maximum value of the directional disturbance; parameters "Extension step size" is used to control the path. These two parameters are used to control the "curvature" of the path. By adjusting these two parameters, different forms of "disease trunks" can be generated, such as straight, meandering, or branching (which can be achieved by starting a new walking sequence at a specific point).

[0066] 3. Path Smoothing: The B-spline interpolation algorithm is used to smooth the control point sequence, resulting in a continuous and smooth critical line path. This step ensures that the generated disease outline is natural and smooth, avoiding harsh angles.

[0067] Step S302: Perform random radius expansion along the smooth keyline path to generate an irregular loose region contour.

[0068] It should be noted that the irregular loose area outline can be a closed area outline without a fixed shape and with blurred boundaries formed by expanding it with a random radius, corresponding to the shape characteristics of the actual loose disease.

[0069] It is understandable that random radius expansion can be a process of randomly generating expansion radii of different sizes with each point on the smooth keyline path as the center, and expanding the range along both sides or around the path. The size of the radius is adaptively adjusted according to the path position and the material distribution characteristics of the target layer to ensure that the expanded area has an irregular shape.

[0070] Furthermore, in order to accurately construct the outline area of ​​loose disease, step S302 above may include: Step S3021: Perform dense sampling along the smooth keyline path to obtain a set of sampling points; Step S3022: Calculate the difference in the expansion radius of each sampling point in the sampling point set based on random numbers, and add the preset minimum expansion radius of loose disease to the difference in the expansion radius to obtain the initial random expansion radius of each sampling point; Step S3023: Adjust the initial random expansion radius based on the material properties of the target layer to obtain the random expansion radius corresponding to each sampling point in the sampling point set; Step S3024: Generate a circular region with each sampling point as the center and the corresponding random expansion radius as the radius, and use the union of all circular regions as the initial contour of the loose region; Step S3025: Perform morphological closing operation smoothing on the initial contour of the loose region to obtain the contour of the irregular loose region.

[0071] It is understandable that loose diseases spread unevenly around the main trunk line. This example simulates this diffusion process: 1. Random radius assignment: along the smoothed path Perform dense sampling to obtain a set of sampling points. For each sampling point Independently generate a random expansion radius The generation method is as follows: , in, and These are the preset minimum and maximum expansion radii of loose disease. and The setting can be adjusted according to the characteristics of the layer materials. For example, a larger setting can be used for layers with poor adhesion. To simulate a more severe loosening.

[0072] 2. Initial contour generation: (Based on each...) With the center of the circle, Draw a circle with radius . The union of all these circles constitutes the initial outline area of ​​the loose disease. This area naturally has varying widths and rough edges.

[0073] 3. Contour Smoothing Optimization: Perform morphological closing operations (e.g., using circular structuring elements) on the initial contour. This operation can: a) smooth the contour edges, eliminating jagged edges caused by the stacking of discrete disks; b) fill small voids and narrow grooves within the contour that may be caused by random radius fluctuations, making the contour closer to a continuous "sheet" or "band". The final result is an irregular, loosely bound region contour. .

[0074] Step S303: Rasterize the outline of the irregular loose region.

[0075] In some embodiments, the detection device can decompose the generated irregular loose region contour into several grid units with the same size as the matrix grid, according to the three-dimensional matrix grid accuracy of the initial heterogeneous road structure model, so that the loose region contour is completely adapted to the grid system of the three-dimensional matrix model.

[0076] Furthermore, to ensure that the generated loose defects conform to the actual engineering conditions and stratum constraints, the following may be included before step S303: S313: Verify whether the geometric features of the irregular loose region outline meet the geometric constraints of the layer, wherein the geometric features include area, aspect ratio and / or concavity coefficient, and the geometric constraints include area constraints and / or morphological constraints. S323: If the geometric features of the irregular loose region outline satisfy the geometric constraints of the layer, then the step of rasterizing the irregular loose region outline is executed. S333: If the geometric features of the irregular loose region contour do not meet the geometric constraints of the layer, then return to the step of randomly generating one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model.

[0077] Understandably, the detection equipment can verify whether the geometric features of the contour of an irregular, loose region meet the constraints of its layer; these geometric features include area A and aspect ratio. Concavity coefficient At least one of them.

[0078] In practical implementation, to ensure that the generated loose defects conform to engineering realities and stratigraphic constraints, the outline of irregular loose areas can be modified. Validation should be conducted. Validation criteria can be based on layer geometry constraints, including: Area constraint: Total area of ​​loose regions It should be smaller than the maximum allowable disease area for that layer. .

[0079] Shape constraints: The aspect ratio of its circumscribed rectangle can be calculated. Or calculate its concavity coefficient. (defined as the ratio of the actual contour perimeter to the convex hull perimeter), and it is required to be within a certain range to avoid generating overly extreme or unreasonable shapes.

[0080] If the verification fails, the starting point of the loose disease can be randomly selected again within the target layer, or the key line control point sequence can be regenerated by stepping with random directional perturbation, and the irregular loose area outline can be generated again.

[0081] Furthermore, in order to accurately transform the abstract outline into quantifiable and operable grid cells, step S303 above may include: Step S3031: Obtain the raster resolution parameters corresponding to the three-dimensional matrix model of the road structure; Step S3032: Divide the target layer into two-dimensional matrix grid units according to the grid resolution parameters; Step S3033: Map the coordinates of the irregular loose region outline to the two-dimensional matrix grid cell; Step S3034: Mark the grid cells containing the outline coordinates of the irregular loose region to complete the rasterization of the irregular loose region outline.

[0082] In practice, the detection equipment will verify the outline of the loose region. Rasterization is performed and mapped onto the corresponding layer slice (two-dimensional matrix) of the three-dimensional matrix model M of the road structure. All cells on this slice are traversed; if the cell coordinates are located within... If the current identifier is a value (e.g., 2 or 3) that characterizes a binder (such as asphalt or cement), then update it to a specific identifier (e.g., 5) that characterizes a "loose material".

[0083] Step S304: Update the identifiers representing the binder within the rasterized irregular loose region outline to identifiers representing the loose material, thus forming a target heterogeneous road structure model containing loose defects.

[0084] It should be noted that the binder identifier is a unique code used to characterize the binder (such as asphalt or cement) in the initial heterogeneous road structure model. It is associated with the material properties of the binder and is used to distinguish the binder from other materials (aggregates, voids).

[0085] A loose material identifier is a unique code used to characterize loose materials (mixtures formed after the bonding between aggregates and binders fails). It is associated with the material properties of the loose material and is used to distinguish loose areas from normal areas.

[0086] In some embodiments, the detection device can filter out all grid cells located in irregular loose areas based on the rasterized marking results, extract the coordinates of these grid cells in the three-dimensional matrix model and their current material identifiers; using a batch processing algorithm, all identifiers representing cementitious materials in the filtered grid cells are uniformly updated to identifiers representing loose materials; during the update process, the original identifiers representing aggregates and voids in the grid cells are retained and not modified; the updated three-dimensional matrix model is fully verified to check whether all cementitious material identifiers in the loose areas have been updated to loose material identifiers and whether the identifiers in the non-loose areas remain unchanged, ensuring no omissions or errors in updating; the verified model is optimized to adjust the transition of identifiers between the loose areas and the surrounding normal areas to avoid abrupt boundaries; after optimization, a target heterogeneous road structure model containing loose defects is obtained.

[0087] This embodiment simulates the natural extension characteristics of real loose diseases by randomly generating smooth paths, avoiding artificially preset regular shapes for disease outlines. Through random radius expansion, it accurately generates the outlines of irregular loose areas, perfectly restoring the characteristics of real loose diseases—no fixed shape and blurred boundaries—solving the problem of traditional regular outlines not matching actual diseases. Rasterization achieves precise adaptation between the irregular loose area outlines and the initial 3D matrix model, transforming abstract outlines into quantifiable and operable raster units. Batch updating of identifiers enables digital representation of loose disease areas, accurately simulating the process of aggregate and binder bonding failure, allowing the target model to truly reflect the material state of loose diseases. Ultimately, it forms a realistic model containing loose diseases, providing high-quality sample support for radar simulation and disease detection.

[0088] Furthermore, this embodiment of the invention also proposes a computer-readable storage medium storing a non-homogeneous road loose disease modeling and detection program. When the non-homogeneous road loose disease modeling and detection program is executed by a processor, it implements the steps of the non-homogeneous road loose disease modeling and detection method described above.

[0089] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0090] The aforementioned computer-readable storage medium may be included in the non-homogeneous road loose disease modeling and detection equipment; or it may exist independently and not assembled into the non-homogeneous road loose disease modeling and detection equipment.

[0091] Furthermore, this invention also proposes a computer program product, including a non-homogeneous road loose disease modeling and detection program, which, when executed by a processor, implements the steps of the non-homogeneous road loose disease modeling and detection method described above.

[0092] The specific implementation of the computer program product of this invention is basically the same as the embodiments of the above-mentioned method for modeling and detecting loose defects in heterogeneous roads, and will not be repeated here.

[0093] Reference Figure 4 , Figure 4 This is a structural block diagram of the first embodiment of the non-homogeneous road loose disease modeling and detection device of the present invention.

[0094] like Figure 4As shown, the non-homogeneous road loose disease modeling and detection device proposed in this embodiment of the invention includes: The structural model building module 10 is used to build a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. The heterogeneous model construction module 20 is used to set the generation region and initialize the material identifier in the three-dimensional matrix model of the road structure to generate an initial heterogeneous road structure model, wherein the initial heterogeneous road structure model includes aggregates, binders and voids; The loose disease model construction module 30 is used to generate adaptive irregular loose diseases within the target layer of the initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose diseases. The defect detection module 40 is used to generate radar echo simulation data of the target heterogeneous road structure model through the radar echo simulation model of the heterogeneous road structure, and input the radar echo simulation data into the pre-trained deep learning defect detection model for loose defect detection.

[0095] This embodiment constructs a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. A generation region is set within this model, and material identifiers are initialized to generate an initial heterogeneous road structure model. This initial heterogeneous road structure model includes aggregates, binders, and voids. Adaptive irregular loose defects are generated within the target layer of this initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose defects. Radar echo simulation data of the target heterogeneous road structure model is generated using a heterogeneous road structure radar echo simulation model, and this radar echo simulation data is input into a pre-programmed... The trained deep learning-based disease detection model is used to detect loose road defects. This embodiment effectively restores the heterogeneity of road structures and the true characteristics of loose defects, breaking through the limitations of traditional homogeneous models and regular defect simulations, improving the realism of heterogeneous road structure models, and achieving realistic modeling of loose defects. Radar echo simulation significantly reduces the cost of defect detection and improves detection efficiency. The deep learning-based disease detection model enables accurate identification of defects, effectively uncovering the characteristic patterns of loose defects, providing technical support for early warning and precise maintenance of road defects, and ensuring the safety, stability, and service life of road structures.

[0096] The heterogeneous road loose-walled disease modeling and detection device provided in this application, employing the heterogeneous road loose-walled disease modeling and detection method described in the above embodiments, can solve the technical problems of heterogeneous road loose-walled disease modeling and detection. Compared with the prior art, the beneficial effects of the heterogeneous road loose-walled disease modeling and detection device provided in this application are the same as those of the heterogeneous road loose-walled disease modeling and detection method described in the above embodiments, and other technical features in the heterogeneous road loose-walled disease modeling and detection device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0097] It should be understood that the above are merely illustrative examples and do not constitute any limitation on the technical solutions of the present invention. In specific applications, those skilled in the art can make settings as needed, and the present invention does not impose any restrictions on this.

[0098] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of this invention. In practical applications, those skilled in the art can select some or all of the workflow to achieve the purpose of this embodiment according to actual needs, and no restrictions are imposed here.

[0099] In addition, for technical details not described in detail in this embodiment, please refer to the method for modeling and detecting non-homogeneous road loose disease provided in any embodiment of the present invention, which will not be repeated here.

[0100] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0101] It should be noted that the user information (including but not limited to user device information, user personal information, user location information, user behavior information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0102] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0103] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0104] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for modeling and detecting loose defects in heterogeneous roads, characterized in that, The method includes: Based on the actual geometric dimensions of the road, a three-dimensional matrix model of the road structure is constructed; In the three-dimensional matrix model of the road structure, a generation region is set and material identifiers are initialized to generate an initial heterogeneous road structure model, which includes aggregates, binders and voids. Adaptive irregular loose defects are generated within the target layer of the initial heterogeneous road structure model to form a target heterogeneous road structure model containing loose defects. Radar echo simulation data of the target heterogeneous road structure model is generated by using a radar echo simulation model of heterogeneous road structure, and the radar echo simulation data is input into a pre-trained deep learning disease detection model for loose disease detection.

2. The method for modeling and detecting loose defects in heterogeneous roads as described in claim 1, characterized in that, The process of generating adaptive irregular loose defects within the target layer of the initial heterogeneous road structure model to form a target heterogeneous road structure model containing loose defects includes: One or more smooth critical line paths are randomly generated within the target layer of the initial heterogeneous road structure model; Random radius expansion is performed along the smooth keyline path to generate an irregular loose region contour. The outline of the irregular loose region is rasterized; The identifiers representing the binder within the rasterized irregular loose region outline are updated to identifiers representing the loose material, thus forming a target heterogeneous road structure model containing loose defects.

3. The method for modeling and detecting loose defects in heterogeneous roads as described in claim 2, characterized in that, The process of randomly generating one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model includes: The starting point of loose disease is randomly selected within the target layer of the initial heterogeneous road structure model; Based on the starting point, a sequence of keyline control points is generated using a stepping method with random directional perturbation; The extension step size and curvature of the key line control point sequence are adjusted based on the preset step size and the maximum value of the preset direction disturbance to obtain the adjusted key line control point sequence. The adjusted key line control point sequence is used to characterize the main path of disease in different forms. The adjusted keyline control point sequence is smoothed using the B-spline interpolation algorithm to obtain a smoothed keyline path.

4. The method for modeling and detecting loose defects in heterogeneous roads as described in claim 3, characterized in that, The step of expanding the random radius along the smooth keyline path to generate an irregular loose region contour includes: Dense sampling is performed along the smooth keyline path to obtain a set of sampling points; The difference in the expansion radius of each sampling point in the sampling point set is calculated based on random numbers, and the minimum expansion radius of the loose disease is added to the difference in the expansion radius to obtain the initial random expansion radius of each sampling point; Based on the material properties of the target layer, the initial random expansion radius is adjusted to obtain the random expansion radius corresponding to each sampling point in the sampling point set; A circular region is generated with each sampling point as the center and the corresponding random expansion radius as the radius. The union of all circular regions is used as the initial contour of the loose region. The initial contour of the loose region is smoothed by morphological closing operation to obtain the contour of the irregular loose region.

5. The method for modeling and detecting loose defects in heterogeneous roads as described in claim 4, characterized in that, Before rasterizing the irregular loose region outline, the method further includes: Verify whether the geometric features of the irregular loose region outline meet the geometric constraints of the layer. The geometric features include area, aspect ratio and / or concavity coefficient. The geometric constraints include area constraints and / or morphological constraints. If the geometric features of the irregular loose region outline satisfy the geometric constraints of the layer, then the step of rasterizing the irregular loose region outline is performed. If the geometric features of the irregular loose region contour do not meet the geometric constraints of the layer, then return to the step of randomly generating one or more smooth critical line paths within the target layer of the initial heterogeneous road structure model.

6. The method for modeling and detecting loose defects in heterogeneous roads as described in claim 2, characterized in that, The step of rasterizing the outline of the irregular loose region includes: Obtain the raster resolution parameters corresponding to the three-dimensional matrix model of the road structure; The target layer is divided into two-dimensional matrix raster units according to the raster resolution parameters; Map the coordinates of the irregular loose region outline to the two-dimensional matrix grid cell; Mark the grid cells containing the outline coordinates of the irregular loose region to complete the rasterization of the irregular loose region outline.

7. The method for modeling and detecting loose defects in heterogeneous roads as described in any one of claims 1 to 6, characterized in that, The process of generating radar echo simulation data of the target heterogeneous road structure model through a radar echo simulation model of heterogeneous road structure, and inputting the radar echo simulation data into a pre-trained deep learning disease detection model for loose disease detection, includes: Electromagnetic property parameters are configured for the loose materials in the target heterogeneous road structure model. Electromagnetic simulation calculations are performed on the configured target heterogeneous road structure model using a radar echo simulation model of heterogeneous road structure to generate radar echo simulation data. The radar echo simulation data is input into a pre-trained deep learning disease detection model for feature extraction to obtain disease feature data. Based on the disease characteristic data, detect whether there are loose diseases in the target heterogeneous road structure model, and output the loose disease detection results.

8. A device for modeling and detecting loose defects in heterogeneous roads, characterized in that, The device includes: The structural model building module is used to construct a three-dimensional matrix model of the road structure based on the actual geometric dimensions of the road. The heterogeneous model construction module is used to set the generation region and initialize material identifiers in the three-dimensional matrix model of the road structure to generate an initial heterogeneous road structure model, wherein the initial heterogeneous road structure model contains aggregates, binders and voids; The loose disease model construction module is used to generate adaptive irregular loose diseases within the target layer of the initial heterogeneous road structure model, forming a target heterogeneous road structure model containing loose diseases. The defect detection module is used to generate radar echo simulation data of the target heterogeneous road structure model through the radar echo simulation model of the heterogeneous road structure, and input the radar echo simulation data into the pre-trained deep learning defect detection model for loose defect detection.

9. A modeling and detection device for loose road defects in heterogeneous roads, characterized in that, The heterogeneous road loose disease modeling and detection device includes: a memory, a processor, and a heterogeneous road loose disease modeling and detection program stored in the memory. The processor is used to run the heterogeneous road loose disease modeling and detection program, which is configured to implement the heterogeneous road loose disease modeling and detection method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a non-homogeneous road loose disease modeling and detection program, which, when executed by a processor, implements the non-homogeneous road loose disease modeling and detection method as described in any one of claims 1 to 7.

Images