Airport runway disease accurate identification method and system based on sky-ground fusion

Abstract

The invention relates to the technical field of material analysis, in particular to an airport runway disease accurate identification method and system based on sky-ground fusion, and solves the problems that in the prior art, due to the fact that surface clutters and deep abnormal signals are difficult to distinguish, a macro-micro fusion evaluation mechanism is lacked, and the identification accuracy is poor. The false alarm rate of the hidden disease detection of the airport runway is high; and the risk quantification is not accurate. The method comprises the following steps: acquiring deformation monitoring data, and generating foundation curvature distribution data; obtaining runway surface image data, and generating earth surface material marking data; acquiring runway underground echo data, and generating actually measured echo distribution data; according to the foundation curvature distribution data and the earth surface material mark data corresponding to the target area, constructing reference background echo distribution data of the target area; and analyzing an echo characteristic difference value between the reference background echo distribution data and the actually measured echo distribution data of the target area, and determining a structural damage evaluation result of the target area in combination with the foundation curvature distribution data.

Description

Technical Field

[0001] This invention relates to the field of materials analysis technology, specifically to a method and system for accurate identification of airport runway defects based on sky-ground fusion. Background Technology

[0002] As a core infrastructure for air transport, the structural safety of airport runways is of paramount importance. Runways are usually laid on a foundation of rigid concrete panels. During long-term service, the foundation soil is prone to slow and uneven settlement due to changes in hydrogeological conditions. This macroscopic settlement deformation will induce bending and tensile stress inside the overlying concrete panel, making high-stress areas a high-incidence area for structural defects.

[0003] Currently, ground penetrating radar (GPR) is the primary means of detecting hidden defects in runways. In actual engineering, the runway surface is often covered with a large number of asphalt repair blocks and joint fillers due to routine maintenance. When these surface repair materials are detected by radar, they will generate strong interface reflection waves and multiple wave oscillations. Moreover, these strong surface clutter signals are extremely similar in waveform and energy intensity to the abnormal signals caused by deep voids or cavities.

[0004] Furthermore, the existing runway operation and maintenance system is fragmented. Satellite monitoring of foundation settlement and vehicle-mounted radar detection of underground cavities are not interconnected, resulting in the assessment of defects being limited to the amplitude of radar signals. There is a lack of disaster risk assessment based on the foundation mechanical state, and it is impossible to distinguish the risk differences of the same tiny cavity under different foundation conditions. Summary of the Invention

[0005] To address the technical problems in existing technologies, such as the difficulty in distinguishing between surface clutter and deep anomaly signals, and the lack of a macro-micro fusion assessment mechanism, which leads to high false alarm rates and inaccurate risk quantification in the detection of hidden defects in airport runways, the present invention aims to provide a method and system for accurate identification of airport runway defects based on sky-ground fusion. The specific technical solution adopted is as follows: Firstly, a method for accurate identification of airport runway defects based on sky-ground fusion is provided, including: acquiring deformation monitoring data collected by satellite and generating foundation curvature distribution data covering the runway area based on the deformation monitoring data; acquiring runway surface image data collected by UAV and generating surface material marker data characterizing the distribution of surface repair materials based on the image data; acquiring runway subsurface echo data collected by vehicle-mounted ground-penetrating radar and generating measured echo distribution data based on the runway subsurface echo data; constructing benchmark background echo distribution data for the target area based on the foundation curvature distribution data and surface material marker data corresponding to the target area; the benchmark background echo distribution data characterizes the energy form that radar echoes should present under normal background; analyzing the echo characteristic difference values ​​between the benchmark background echo distribution data and the measured echo distribution data of the target area, and combining the foundation curvature distribution data to determine the structural damage assessment results of the target area.

[0006] Based on the above technical solution, the present invention provides a method for accurate identification of airport runway defects based on sky-ground fusion. By fusing three types of data—satellite deformation monitoring, UAV surface imagery, and vehicle-mounted ground-penetrating radar underground echo—the method first uses satellite data to generate curvature distribution data characterizing the complexity of the foundation structure, and then uses UAV imagery to obtain surface repair material marking data. Combining the two types of data, a benchmark background echo distribution that fits the actual environment of the target area is constructed. Then, by analyzing the characteristic differences between the benchmark and the measured echoes and coupling the foundation curvature information, damage assessment is performed. This method effectively solves the problems of difficulty in distinguishing between surface clutter and deep defect signals, and false alarms caused by fragmented operation and maintenance data. It achieves accurate identification of hidden defects in airport runways and provides a scientific basis for runway preventive maintenance.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the method for generating foundation curvature distribution data specifically includes: establishing a local Cartesian coordinate system for the runway area; the local Cartesian coordinate system has the runway entrance center point as the origin, the direction extending along the runway centerline as the longitudinal axis, the direction perpendicular to the centerline pointing to the runway side as the transverse axis, and the direction perpendicular to the ground downwards as the depth axis; projecting the deformation monitoring data collected by satellite from the original geographic coordinate system to the local Cartesian coordinate system to obtain scattered data containing physical coordinates and cumulative settlement; processing the scattered data using a preset interpolation algorithm to generate a cumulative settlement regular grid covering the runway area; performing second-order differential processing on the cumulative settlement regular grid to obtain foundation curvature values ​​characterizing the degree and complexity of the runway's underground structure under stress bending, thereby generating foundation curvature distribution data.

[0008] In conjunction with the first aspect above, in one possible implementation, the method for generating surface material marker data specifically includes: registering runway surface image data collected by UAV to a local Cartesian coordinate system; using a preset semantic segmentation algorithm to classify the surface material of the registered image data, identifying asphalt repair areas and original concrete areas, and generating surface material marker data; the surface material marker data is associated with the longitudinal position of the runway to indicate whether asphalt repair material exists within the lateral coverage area of ​​the corresponding longitudinal position.

[0009] In conjunction with the first aspect above, in one possible implementation, the method for acquiring runway underground echo data collected by vehicle-mounted ground-penetrating radar and generating measured echo distribution data based on the runway underground echo data specifically includes: using a pulse signal from a high-precision odometer as a trigger source to control the vehicle-mounted ground-penetrating radar to collect runway underground echo data at preset intervals; and performing preprocessing on the runway underground echo data, including removing DC components, automatic gain control, and instantaneous amplitude envelope extraction, to obtain measured echo distribution data.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the method for constructing the reference background echo distribution data of the target area specifically includes: extracting the underground echo signal of the exposed concrete section within a preset healthy area, and obtaining the reference attenuation rate by fitting the attenuation curve of the signal amplitude envelope; the reference attenuation rate characterizes the energy loss benchmark of electromagnetic waves in a normal runway medium; determining the regional structural clutter attenuation factor based on the reference attenuation rate, the foundation curvature distribution data and surface material marker data corresponding to the target area; the regional structural clutter attenuation factor is used to characterize the background clutter intensity and attenuation trend of the target area; and constructing reference background echo distribution data with dimensions consistent with the measured echo distribution data based on the regional structural clutter attenuation factor.

[0011] In conjunction with the first aspect mentioned above, in one possible implementation, the method for analyzing the echo characteristic difference between the reference background echo distribution data and the measured echo distribution data of the target area specifically includes: extracting the underground echo signal at the asphalt joint within a preset healthy area, and obtaining a reference spectrum offset threshold through spectrum analysis; the reference spectrum offset threshold is used to distinguish the spectrum characteristics of lossy and lossless media; constructing a transmission cost matrix based on the reference spectrum offset threshold, the measured echo distribution data, and the reference background echo distribution data; the transmission cost matrix consists of a spatial distance term and a medium property penalty term, whereby the spatial distance term characterizes the cost of energy movement in space, and the medium property penalty term is used to distinguish the signal differences corresponding to media with different loss characteristics based on the reference spectrum offset threshold; and determining the echo characteristic difference value characterizing the essential difference between the measured signal and the reference model based on the transmission cost matrix.

[0012] In conjunction with the first aspect above, in one possible implementation, before determining the structural damage assessment results of the target area, the method further includes: extracting the echo characteristic difference values ​​corresponding to all detection segments within a preset healthy area to form a difference value sequence; calculating the arithmetic mean and standard deviation of the difference value sequence, and using the sum of the arithmetic mean and three times the standard deviation as the system noise tolerance; the system noise tolerance represents the maximum statistical residual that normal background clutter can produce.

[0013] In conjunction with the first aspect above, in one possible implementation, the method for determining the structural damage assessment result of the target area specifically includes: if the echo characteristic difference value of the target area is less than or equal to the system noise tolerance, then the target area is determined to be a normal area; if the echo characteristic difference value of the target area is greater than the system noise tolerance, then the target area is determined to be an abnormal candidate segment, and the structural damage assessment value is determined by combining the echo characteristic difference value of the target area, the system noise tolerance, and the foundation curvature value; the structural damage assessment value comprehensively reflects the scale of micro-defects and the vulnerability of the macro-structure.

[0014] In conjunction with the first aspect above, in one possible implementation, the method for determining the structural damage assessment results of the target area further includes: outputting corresponding graded early warning results based on the structural damage assessment values ​​according to preset grading standards, and generating a visualized runway health status distribution map.

[0015] Secondly, a precise identification system for airport runway defects based on sky-ground fusion is provided, including: a macroscopic mechanical field construction module, a multidimensional data perception module, a feature calculation module, and a risk assessment module; the macroscopic mechanical field construction module is used to acquire deformation monitoring data collected by satellite and generate ground curvature distribution data covering the runway area based on the deformation monitoring data; the multidimensional data perception module is used to acquire runway surface image data collected by UAVs and identify the image data to generate surface material marker data characterizing the distribution of surface repair materials; the multidimensional data perception module is also used to acquire runway surface data collected by vehicle-mounted ground penetrating radar. The system generates measured echo distribution data based on the runway subsurface echo data; a feature calculation module is used to construct the baseline background echo distribution data for the target area based on the foundation curvature distribution data and surface material marker data corresponding to the target area; the baseline background echo distribution data characterizes the energy form that radar echoes should present under normal background; the feature calculation module is also used to analyze the echo characteristic difference values ​​between the baseline background echo distribution data and the measured echo distribution data for the target area; and a risk assessment module is used to determine the structural damage assessment results for the target area based on the echo characteristic difference values ​​and the foundation curvature distribution data.

[0016] The present invention has the following beneficial effects: By integrating three types of data—satellite deformation monitoring, UAV surface imagery, and vehicle-mounted ground-penetrating radar underground echoes—the system first uses satellite data to generate curvature distribution data characterizing the complexity of the foundation structure, and then uses UAV imagery to obtain surface repair material marking data. Combining these two types of data, a baseline background echo distribution that closely matches the actual environment of the target area is constructed. Finally, by analyzing the characteristic differences between the baseline and measured echoes and coupling foundation curvature information, damage assessment is performed. This effectively solves the problems of difficulty in distinguishing between surface clutter and deep-seated defects, as well as false alarms caused by fragmented operation and maintenance data. It enables accurate identification of hidden defects in airport runways and provides a scientific basis for runway preventive maintenance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A system structure diagram of an airport runway defect precision identification system based on sky-ground fusion provided in one embodiment of the present invention; Figure 2 A flowchart illustrating a method for accurate identification of airport runway defects based on sky-ground fusion, as provided in one embodiment of the present invention; Figure 3 This is a schematic diagram of the hardware structure of an airport runway defect precision identification device based on sky-ground fusion, provided as an embodiment of the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method and system for accurate identification of airport runway defects based on sky-ground fusion proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the method and system for accurate identification of airport runway defects based on sky-ground fusion provided by the present invention.

[0022] Please see Figure 1 The diagram illustrates a system structure of an airport runway defect identification system based on sky-ground fusion, according to an embodiment of the present invention. The system includes: a macroscopic mechanical field construction module 1, a multidimensional data perception module 2, a feature calculation module 3, and a risk assessment module 4.

[0023] The macroscopic mechanical field construction module 1 is primarily responsible for generating foundation curvature distribution data characterizing the stress-induced bending degree and complexity of the runway's underground structure based on historical satellite deformation monitoring data, providing a macroscopic mechanical environment basis for subsequent modules. This module can be implemented using a high-performance server equipped with professional data processing software, supporting offline batch data processing, and specifically includes four sub-modules: The coordinate system establishment submodule 11 is used to establish a local Cartesian coordinate system for the runway area. This coordinate system takes the center point of the runway entrance as the origin, extends along the runway centerline as the longitudinal axis, points perpendicular to the centerline towards the runway side as the transverse axis, and extends perpendicularly downwards from the ground as the depth axis, providing a unified reference for the spatial alignment of all subsequent data.

[0024] The data projection submodule 12 is responsible for projecting the deformation monitoring data collected by the satellite based on the original geographic coordinate system to the local Cartesian coordinate system through a preset rigid transformation algorithm, and transforming it into scattered data containing physical coordinates and cumulative settlement, so as to achieve the matching of satellite data with the actual spatial location of the runway.

[0025] The settlement grid generation submodule 13 uses a preset interpolation algorithm to process the scattered data and generate a cumulative settlement regular grid covering the entire runway area. This compensates for the discrete and non-uniform defects of the original satellite point cloud data and ensures that the data can be queried and called by subsequent modules.

[0026] The curvature calculation submodule 14 performs second-order differential processing on the cumulative settlement regular grid, calculates the foundation curvature value of each grid node, and integrates it to form foundation curvature distribution data covering the entire runway area. This data will be used by the feature solving module 3 to construct a dynamic benchmark model, and at the same time provide macroscopic structural vulnerability parameters for the risk assessment module 4.

[0027] The multi-dimensional data perception module 2 is responsible for the real-time acquisition, processing, and calibration of two-dimensional data between air and ground. It acquires surface material marker data and measured echo distribution data, providing core input data for the feature calculation module 3. It can be implemented collaboratively through UAV inspection systems, vehicle-mounted detection systems, and data acquisition terminals, and includes four sub-modules: The UAV data processing submodule 21 acquires runway surface image data through a UAV equipped with a high-resolution camera. First, the image data is registered to a local Cartesian coordinate system. Then, a preset semantic segmentation algorithm is used to classify the surface material of the registered image, identifying asphalt repair areas and original concrete areas. Finally, surface material marker data associated with the longitudinal position of the runway is generated. This data is used to indicate whether asphalt repair material exists within the lateral coverage area of ​​the corresponding longitudinal position, providing surface interference source information for the feature calculation module 3 to build a benchmark model.

[0028] The radar data acquisition submodule 22 uses the pulse signal of the high-precision odometer as the trigger source to control the vehicle-mounted multi-channel ground-penetrating radar to collect underground echo data of the runway at preset intervals. The high-precision odometer synchronously records the vehicle's mileage information to ensure the equal spacing and positioning accuracy of the radar data acquisition.

[0029] The data preprocessing submodule 23 performs preprocessing on the collected underground echo data, including removing DC components, automatic gain control, and instantaneous amplitude envelope extraction, to eliminate environmental interference and geometric loss effects in the data and generate measured echo distribution data that can accurately reflect the reflection intensity of the underground medium.

[0030] The data alignment submodule 24 extracts the corresponding surface material marker data fragments according to the mileage coordinate interval of the detection segment, so as to achieve precise spatial alignment between the surface material marker data and the measured echo distribution data, ensuring that the two types of data are called by the feature solving module 3 in the same spatial dimension, thus ensuring the accuracy of the benchmark model construction.

[0031] Feature calculation module 3 is the core algorithm unit of the system. It is responsible for constructing a dynamic benchmark model by combining macroscopic mechanical data and real-time sensing data, and calculating the essential differences between the measured data and the benchmark model. It can be implemented by an industrial computer equipped with embedded core algorithms and includes three sub-modules: The benchmark parameter extraction submodule 31 extracts underground echo signals within a preset healthy area (usually a well-maintained section of road with a preset length before the runway entrance). Among them, the amplitude envelope attenuation curve is fitted from the signal of the exposed concrete section to obtain the benchmark attenuation rate, which characterizes the energy loss benchmark of electromagnetic waves in the normal runway medium. The benchmark spectral offset threshold, which is used to distinguish the spectral characteristics of lossy and lossless media, is obtained from the signal of the asphalt joint through spectral analysis. These two types of parameters provide key basis for dynamic benchmark construction and difference calculation.

[0032] The dynamic benchmark construction submodule 32 calls the foundation curvature distribution data of the macroscopic mechanical field construction module 1 and the surface material marking data of the multidimensional data perception module 2. Combined with the benchmark attenuation rate, it determines the regional structural clutter attenuation factor that characterizes the background clutter intensity and attenuation trend of the target area through a preset calculation method. Then, based on the attenuation factor, it constructs benchmark background echo distribution data with the same dimension as the measured echo distribution data. This data can truly reflect the energy form that the radar echo should present under normal background in the target area.

[0033] The difference calculation submodule 33 first performs probability density normalization on the measured echo distribution data and the reference background echo distribution data, and then maps the pixel spatial coordinates and signal spectrum offset (i.e. frequency shift) to a preset numerical range. Subsequently, it constructs a transmission cost matrix containing a spatial distance term and a medium attribute penalty term (the spatial distance term represents the cost of energy spatial movement, and the medium attribute penalty term distinguishes the signal differences of media with different loss characteristics) by combining the reference spectrum offset threshold. Finally, it obtains the echo feature difference value that represents the essential difference between the measured signal and the reference model through a preset difference calculation method. This difference value will be directly transmitted to the risk assessment module 4 for damage assessment.

[0034] Risk assessment module 4 is responsible for quantifying the risks and outputting decisions based on the feature calculation results, providing a scientific basis for runway operation and maintenance. It can be implemented through a decision terminal and a visualization platform, and includes four sub-modules: The noise tolerance setting submodule 41 extracts the echo characteristic difference values ​​corresponding to all detection segments within the preset healthy area, forms a difference value sequence, calculates the arithmetic mean and standard deviation of the sequence, and uses the sum of the arithmetic mean and three times the standard deviation as the system noise tolerance. This tolerance represents the maximum statistical residual that normal background clutter can produce, providing an objective threshold for anomaly judgment.

[0035] The damage assessment submodule 42 compares the echo characteristic difference value of the target area with the system noise tolerance. If the difference value is less than or equal to the tolerance, the target area is determined to be a normal area; if the difference value is greater than the tolerance, it is determined to be an abnormal candidate segment. At the same time, the foundation curvature value of the target area is called in the macroscopic mechanical field construction module 1. Combined with the difference value and the system noise tolerance, the structural damage assessment value that comprehensively reflects the scale of micro defects and the vulnerability of macro structures is obtained through a preset coupling calculation method.

[0036] The early warning and classification submodule 43 presets multiple levels of disease risk standards. Based on the size of the structural damage assessment value, it outputs the corresponding classification early warning results (usually divided into three levels: high risk, attention, and minor), which clarifies the urgency of diseases in different areas and provides clear handling guidance for operation and maintenance personnel.

[0037] The visualization output submodule 44 maps the graded early warning results to the local Cartesian coordinate system of the runway, generating a visualized distribution map of the runway health status. This visually displays the spatial distribution and risk level of potential defects and hazards, making it easier for maintenance personnel to quickly locate key areas of concern and develop targeted maintenance plans.

[0038] Please see Figure 2 The diagram illustrates a flowchart of a method for accurate identification of airport runway defects based on air-ground fusion, according to an embodiment of the present invention. This method includes: S1. Acquire deformation monitoring data collected by satellite, and generate ground curvature distribution data covering the runway area based on the deformation monitoring data.

[0039] In some implementations, the method of S1 can be specifically implemented through the following S11 to S14, which are explained in detail below: S11. Establish a local Cartesian coordinate system for the runway area.

[0040] The local Cartesian coordinate system takes the center point of the runway entrance as its origin. This origin can be determined by retrieving the positioning parameters from the runway design and construction drawings, or by using a high-precision handheld positioning device to measure the intersection of the runway entrance centerline and the runway width centerline on site. The longitudinal axis (x-axis) extends along the runway centerline, the transverse axis (y-axis, taking the direction pointing to the right side of the runway as an example) is perpendicular to the centerline and points downwards, and the depth axis (z-axis) is perpendicular to the ground and points downwards.

[0041] Constructing a unified spatial reference coordinate system can solve the coordinate incompatibility problem between satellite deformation monitoring data (based on geographic coordinate system) and subsequent vehicle-mounted detection data (based on relative mileage coordinate system), providing a consistent position reference for the spatial alignment and fusion calculation of data from various modules.

[0042] S12. Project the deformation monitoring data collected by the satellite from the original geographic coordinate system to the local Cartesian coordinate system to obtain scatter data containing physical coordinates and cumulative settlement.

[0043] In some implementations, four control points are selected: two corner points at the runway entrance and two corner points at the runway exit. The original geographic coordinates (latitude and longitude) of each control point are measured, as well as its coordinates in the local Cartesian coordinate system (e.g., the right corner point at the entrance is (0, 15, 0) in the local coordinate system, where 15 meters is the runway design width). Based on the coordinates of these control points, a rigid transformation matrix (including rotation and translation parameters) from the original geographic coordinate system to the local Cartesian coordinate system is constructed using a planar affine transformation algorithm (or a four-parameter coordinate transformation algorithm). Since the four corner points of the runway are almost on the same elevation plane (the z-values ​​are approximately equal), using a planar coordinate transformation algorithm can avoid the singularity problem of the equation system in the three-dimensional transformation algorithm, ensuring the stability and accuracy of the transformation parameter solution. Subsequently, each data point in the original geographic coordinate system (including longitude, latitude, and corresponding cumulative vertical subsidence) is substituted into the transformation matrix to convert it into three-dimensional data (x, y, z) in a local Cartesian coordinate system. Here, x and y are the planar physical coordinates in the local coordinate system, z is the spatial coordinate of the underground area located at that location, and the z-axis coordinate with the surface location as the origin can represent the cumulative vertical subsidence of that location relative to the surface. Finally, a scatter dataset containing physical coordinates and cumulative subsidence is obtained.

[0044] S13. The scattered data is processed using a preset interpolation algorithm to generate a cumulative settlement rule grid covering the runway area.

[0045] In some implementations, the universal kriging interpolation algorithm is selected as the preset interpolation algorithm, with a preset grid resolution threshold of 0 m × 1 m (this resolution avoids the curvature value step deviation caused by large grids, ensuring the accuracy of micro-position curvature queries). First, in the xy-plane of the local Cartesian coordinate system, a regular grid is divided at 1 m × 1 m intervals, resulting in several grid nodes (i, j), where i is the grid index along the longitudinal axis and j is the grid index along the transverse axis. For each grid node, the universal kriging interpolation algorithm is called, using the coordinates of scattered data around the node to perform spatial correlation analysis and unbiased estimation, calculating the corresponding cumulative settlement. After traversing all grid nodes, the cumulative settlement values ​​of all nodes are integrated to form a cumulative settlement regular grid (in two-dimensional matrix form) covering the entire runway area, eliminating the spatial discreteness of the original data.

[0046] S14. Perform second-order differential processing on the cumulative settlement regular grid to obtain the foundation curvature value that characterizes the stress bending degree and complexity of the runway underground structure, and generate foundation curvature distribution data.

[0047] In some implementations, for non-boundary nodes (i, j) in a cumulative settlement regular grid, the second-order difference along the longitudinal axis (x-axis) is first calculated: In the formula, The grid resolution is 10 meters. , , These represent the current node (i, j) and the cumulative settlement of the current node along the x-axis to the nodes immediately before and after it.

[0048] Next, calculate the second difference along the horizontal axis (y-axis): In the formula, , , These represent the cumulative settlement of the current node (i, j) and the nodes adjacent to the current node along the y-axis.

[0049] Adding the absolute values ​​of the second-order differences in the two directions yields the foundation curvature value at node (i, j). : In the formula, The physical unit is The larger the value, the greater the degree of bending of the foundation at that location and the stronger the structural complexity.

[0050] For boundary nodes in a cumulative settlement regular grid, it is necessary to first supplement the boundary with extended neighborhood data to eliminate boundary effects, and then use the same curvature calculation algorithm as the internal nodes to complete the calculation. The specific methods for supplementing the boundary with extended neighborhood data include: extending virtual nodes with the same resolution as the internal grid along the boundary normal direction; and using linear interpolation to fit the settlement of the virtual nodes based on the cumulative vertical settlement of the boundary nodes and their adjacent internal nodes, ensuring that the boundary nodes have neighborhood settlement data with the same number as the internal nodes.

[0051] After completing the calculations for all nodes, the foundation curvature values ​​of all nodes are integrated into a two-dimensional matrix. At the same time, a mapping relationship is established between the physical coordinates (x, y) and the grid index (i, j) in the local Cartesian coordinate system of the runway. In the formula, It is a floor function that takes the largest integer not greater than the value in parentheses.

[0052] Finally, the foundation curvature distribution data covering the runway area is generated, and the distribution of cumulative settlement is transformed into curvature parameters that characterize the stress state of the foundation structure, accurately reflecting the degree of bending and complexity of the underground structure of the runway.

[0053] S2. Acquire runway surface image data collected by UAV, and identify the image data to generate surface material marker data that characterizes the distribution of surface repair materials.

[0054] In some implementations, the method of S2 can be specifically implemented through the following S21 to S22, which are explained in detail below: S21. Register the runway surface image data collected by the UAV to the local Cartesian coordinate system.

[0055] In some implementations, high-resolution orthophotos of the runway collected during regular UAV inspections are first acquired (with a preset resolution of 5 cm / pixel to meet the requirements for identifying details of the ground material). Then, spatial control points of the runway are selected as registration references. These control points include two corner points at the runway entrance, two corner points at the exit, and two intersection points at the junction of the runway centerline and the intermediate taxiway, totaling six control points. The physical coordinates of these control points in the local Cartesian coordinate system (e.g., the coordinates of the right corner point at the entrance are (0, 15, 0), where 15 meters represents the runway's design width) have been determined through prior measurements. A corner detection algorithm is used to identify the pixel coordinates corresponding to these control points in the orthophotos. An affine transformation registration algorithm is then used to construct the transformation relationship between the pixel coordinates and the physical coordinates in the local Cartesian coordinate system. Finally, each pixel of the entire orthophoto is substituted into this transformation relationship to complete the spatial registration of the image data to the local Cartesian coordinate system. This ensures that the spatial position of the UAV image is completely aligned with the actual physical space of the runway, eliminating spatial offset between the image and subsequent vehicle-mounted detection data.

[0056] S22. A preset semantic segmentation algorithm is used to classify the surface material of the registered image data, identify the asphalt repair area and the original concrete area, and generate surface material label data.

[0057] The surface material marking data is linked to the longitudinal position of the runway to indicate whether asphalt repair material is present within the lateral coverage area of ​​the corresponding longitudinal position.

[0058] In some implementations, the pre-selected semantic segmentation algorithm is the U-Net deep learning network. The training dataset for this network consists of runway image samples labeled "asphalt repair area" and "original concrete area" (each pixel is labeled with its corresponding material category). The registered orthophoto image is divided into 256×256 pixel image blocks as input units and fed into the U-Net network. The network extracts texture and color features from the image using an encoder-decoder structure and outputs a probability map for each pixel's category. A probability threshold of 0.8 is set (this threshold has been experimentally calibrated to balance classification accuracy and false positive rate). Pixels with a probability ≥ 0.8 are classified as belonging to the corresponding material category, resulting in the material classification result for the entire image; that is, the pixels are classified as either "asphalt repair area" or "original concrete area".

[0059] It should be noted that if the probability corresponding to all output material categories is no greater than 0.8, the following supplementary judgment steps are performed: Extract the maximum probability value among all material categories. If this maximum probability value is greater than or equal to the preset minimum confidence threshold (0.5 in this embodiment), the material category corresponding to the maximum probability value is marked as a suspected material category. In the subsequent construction of the baseline background echo model, the weight of the material parameters in this area is reduced (the corresponding surface material labeling data remains unchanged, but when fusing the material-related parameters of this area (such as the echo attenuation coefficient corresponding to the material), the parameter needs to be multiplied by a preset weighting coefficient, for example, 0.6, to reduce the influence weight of the material parameters of the suspected material area on the baseline background echo model). If the maximum probability value of all material categories is less than 0.5, this area is marked as a material verification area and specially marked in the visualized runway health status distribution map. It is recommended to determine its surface material through manual on-site investigation. The above supplementary judgment rules ensure the full coverage of material classification, avoid missing judgment results due to insufficient model confidence, and guarantee the integrity of the subsequent construction of the baseline background echo model.

[0060] The runway is divided into continuous longitudinal position points according to the acquisition interval of the vehicle-mounted ground-penetrating radar (preset to 5cm). Each longitudinal position point corresponds to the entire lateral coverage area of ​​the runway (i.e., the runway design width). If a pixel in the lateral range corresponding to a certain longitudinal position point is a "asphalt repair area", the corresponding mark value is set to 1; otherwise, it is set to 0. The mark values ​​of all longitudinal positions are integrated into a surface material mark data sequence and stored in association with the runway longitudinal mileage coordinate index.

[0061] S3. Acquire runway underground echo data collected by vehicle-mounted ground-penetrating radar, and generate measured echo distribution data based on the runway underground echo data.

[0062] In some implementations, the method of S3 can be specifically implemented through the following S31 to S33, which are explained in detail below: S31. Using the pulse signal from the high-precision odometer as the trigger source, control the vehicle-mounted ground-penetrating radar to collect runway underground echo data at preset intervals.

[0063] In some implementations, after the vehicle enters the formal testing phase, the mileage data output by the high-precision photoelectric encoder (odometer) is used as the basis, and the preset test segment length is 2 meters (that is, every 2 meters of cumulative driving is recorded as a test segment). This length not only ensures the efficiency of data processing, but also fully covers the distribution range of small defects.

[0064] Meanwhile, the system uses the pulse signal of the photoelectric encoder as the trigger source. Every time it receives a number of pulses corresponding to 5 centimeters of driving distance (for example, each revolution of the photoelectric encoder corresponds to 1 meter of wheel circumference, and each revolution outputs 200 pulses, then every 10 pulses correspond to 5 centimeters of driving distance), it triggers the vehicle-mounted ground-penetrating radar to collect one B-scan raw data, thereby achieving equal-interval acquisition of radar data and ensuring the uniformity of data distribution in the longitudinal mileage direction.

[0065] S32. The underground echo data of the runway is preprocessed by removing the DC component, automatic gain control and instantaneous amplitude envelope extraction to obtain the measured echo distribution data.

[0066] In some implementations, three preprocessing operations are performed sequentially on the collected B-scan raw data: DC component removal: The mean subtraction algorithm is used to calculate the signal mean of each B-scan data, and the original data is subtracted from the mean to eliminate baseline drift interference in the signal, so that the signal baseline returns to near zero, which is convenient for subsequent analysis of the relative intensity of medium reflection. Automatic gain control (AGC) is implemented: an exponential automatic gain control algorithm is used, setting the gain coefficient to increase exponentially with depth (e.g., gain factor G(z) = e^(-z / z)). 0.1z (where z is the radar detection depth), to compensate for the energy loss of radar waves during propagation due to geometric diffusion and medium absorption, so that the signal strength at different depths is within a uniform analyzable range; Extracting the instantaneous amplitude envelope: The Hilbert transform algorithm is used to perform a Hilbert transform on each signal after the first two processes. The magnitude of the transform result is taken as the instantaneous amplitude envelope, highlighting the intensity information of the underground medium reflection interface and filtering out redundant information such as phase.

[0067] By eliminating environmental interference and geometric loss in the raw data, radar signals are converted into characteristic data that only reflect the reflection intensity of the underground medium, thereby improving the accuracy of subsequent analysis.

[0068] Furthermore, the preprocessed instantaneous amplitude envelope data of the current detection segment is arranged according to the dimensions of acquisition channel (longitudinal mileage) and detection depth to form a two-dimensional matrix (rows correspond to longitudinal acquisition channels, and columns correspond to detection depths). This two-dimensional matrix is ​​the measured echo energy distribution data, and the values ​​in the matrix represent the reflection intensity of the underground medium at the corresponding mileage and depth.

[0069] S4. Based on the foundation curvature distribution data and surface material marking data corresponding to the target area, construct the baseline background echo distribution data of the target area.

[0070] The baseline background echo distribution data characterizes the energy pattern that radar echoes should exhibit under normal background conditions.

[0071] In some implementations, method S4 can be specifically implemented using S41 to S43, which are explained in detail below: S41. Extract the underground echo signal of the exposed concrete section within the preset healthy area, and obtain the reference attenuation rate by fitting the attenuation curve of the signal amplitude envelope.

[0072] The reference attenuation rate characterizes the energy loss benchmark of electromagnetic waves in a normal runway medium.

[0073] In some implementations, a pre-defined healthy zone is first identified, such as the section 50 meters before the runway entrance. This area is confirmed by maintenance records to be well-maintained, without asphalt repairs, and composed of original concrete. When a vehicle passes through this area, the system automatically matches pre-stored area material information to identify exposed concrete sections. The amplitude envelopes of all A-scan signals from the ground-penetrating radar within this section are extracted (the amplitude envelopes are extracted using the Hilbert transform in step S3). A least-squares fitting algorithm is then used to fit the amplitude envelope data of each A-scan signal into an exponential decay curve (in the form of...). Where z is the detection depth and A(z) is the amplitude at depth z). The attenuation coefficient α of all fitted curves in the preset healthy area is statistically analyzed (note that, in conjunction with the description in S22, the echo attenuation coefficient corresponding to the suspected material category needs to be multiplied by the preset weighting coefficient), and their arithmetic mean is calculated to obtain the baseline attenuation rate. This provides the basic medium attenuation parameters for the subsequent construction of the dynamic baseline model, avoiding the deviation of the baseline model caused by the differences in the medium characteristics of different runways.

[0074] S42. Determine the regional structural clutter attenuation factor based on the reference attenuation rate, the foundation curvature distribution data corresponding to the target area, and the surface material marking data.

[0075] The regional structural clutter attenuation factor is used to characterize the background clutter intensity and attenuation trend in the target area.

[0076] In some implementations, the center physical coordinates of the target region k are first obtained. Using this coordinate as an index, the foundation curvature value at the corresponding location is retrieved from the foundation curvature grid table generated by the macroscopic mechanical field construction module. .

[0077] Simultaneously, the surface material marker data corresponding to the target area k is retrieved from the data stream of the multi-dimensional data perception module. (1 indicates that asphalt repair exists at this longitudinal location, and 0 indicates that there is no repair). The normalized curvature constant and stress sensitivity coefficient are preset. Then, based on the reference attenuation rate, foundation curvature value, normalized curvature constant, stress sensitivity coefficient, and surface material marker data, the regional structural clutter attenuation factor for the target area k is calculated. : In the formula, Reference attenuation rate (unit: ), providing a benchmark for energy loss under normal conditions; The foundation curvature value of target region k (unit: ); This is the normalized curvature constant, with a value taken from the maximum allowable elastic deformation curvature of a typical runway foundation. For example, a value of [value missing]. (unit: ); Normalize the foundation curvature value to eliminate the influence of dimensions. The larger the value, the greater the degree of foundation bending that exceeds the allowable value. This is the stress sensitivity coefficient, which is an empirical value. It can be obtained by drilling and calibrating typical high-stress sample areas of the runway. It is used to adjust the model's sensitivity to foundation bending, for example, a value of 2.0. The influence of foundation curvature on clutter is amplified by η. The larger η is, the more sensitive the model is to foundation curvature. Only when there is surface repair ( The effect of foundation bending is only taken into account when =1); Ensure that the correction term is not less than 1 to avoid the attenuation rate after correction being lower than the reference value, which conforms to the actual logic that the clutter intensity is not lower than the reference.

[0078] S43. Based on the regional structural clutter attenuation factor, construct a reference background echo distribution data with the same dimension as the measured echo distribution data.

[0079] In some implementations, the average reflection intensity of measured data at the Earth's surface (depth z=0) is first taken as the reference initial amplitude. Obtain the surface material marker corresponding to the target area k. and the calculated regional structural clutter attenuation factor Based on the dimensions of the measured echo distribution data, a two-dimensional matrix of the baseline background echo distribution is constructed. For each point in the matrix Its reference energy value for: In the formula, The depth attenuation term describes the exponential decay of energy with increasing detection depth. Initial amplitude, surface material correction, and depth attenuation are independent factors that act sequentially on the echo energy. Multiplication can reflect the combined effect of these factors, which is consistent with the propagation law of radar echoes.

[0080] After traversing all points, the baseline background echo distribution data is constructed.

[0081] S5. Analyze the echo characteristic differences between the baseline background echo distribution data and the measured echo distribution data of the target area, and combine them with the foundation curvature distribution data to determine the structural damage assessment results of the target area.

[0082] In some implementations, the method for analyzing echo characteristic difference values ​​in S5 can be specifically implemented through the following S51 to S53, which are explained in detail below: S51. Extract the underground echo signal at the asphalt joint within the preset healthy area, and obtain the reference spectrum offset threshold through spectrum analysis.

[0083] The reference spectral offset threshold is used to distinguish the spectral characteristics of lossy and lossless media.

[0084] In some implementations, runway maintenance records are retrieved within a pre-defined healthy area to determine the specific mileage intervals of known asphalt joints. When a vehicle travels through this interval, the A-scan echo signal from the corresponding ground-penetrating radar is extracted. A short-time fourier transform (STFT) algorithm is used to perform spectral analysis on the signal, setting the window function to a Hamming window (window length of 256 sampling points) and an overlap rate of 50%. The frequency shift of the spectrum of each signal segment relative to the center frequency of the ground-penetrating radar transmission is calculated. The frequency shift of the signals at all asphalt joints within the pre-defined healthy area is statistically analyzed, and their arithmetic mean is calculated. Furthermore, considering that the echo frequency shift at asphalt joints is affected by factors such as ambient temperature and asphalt aging, it will fluctuate slightly around the average value (rather than a fixed value) in actual testing. 90% of this average value can be set as the baseline spectral offset threshold, incorporating these normally fluctuating asphalt frequency shifts into the judgment range for lossy media. This avoids misjudging normal asphalt clutter and balances the tolerance for normal fluctuations with the robustness of distinguishing media characteristics.

[0085] S52. Construct a transmission cost matrix based on the reference spectrum offset threshold, measured echo distribution data, and reference background echo distribution data.

[0086] The transmission cost matrix consists of a spatial distance term and a medium property penalty term. The spatial distance term represents the cost of energy moving in space, while the medium property penalty term is used to distinguish the signal differences corresponding to media with different loss characteristics based on a reference spectral offset threshold.

[0087] In some implementations, the total energy of all data points in the measured echo distribution data and the reference background echo distribution data is calculated separately. Then, the energy value of each data point is divided by the total energy of the corresponding data. These two sets of data are converted into a probability distribution in which the sum of the proportions of all data points is 1, so as to avoid the interference of the difference in the total energy of the two sets of data with the subsequent difference calculation results.

[0088] The spatial coordinates corresponding to the data points are mapped to the interval between 0 and 1 using a linear normalization method to obtain normalized spatial location information. At the same time, the frequency shift of the measured signal and the reference spectrum offset threshold are also mapped to the interval between 0 and 1 using the same linear normalization method to obtain normalized frequency shift-related information, thus eliminating the dimensional and order-of-magnitude differences between spatial coordinates and frequency shift.

[0089] Unless otherwise specified, the linear normalization method mentioned in the embodiments of this invention uses maximum and minimum value normalization. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculation result exceeds the interval [0, 1], it is restricted to the range [0, 1] by a truncation function (i.e., if the result is less than 0, it is taken as 0; if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the result.

[0090] Each element in the matrix corresponds to the comprehensive cost of "transferring energy from a pixel in the measured data to a pixel in the reference data", which is composed of a spatial distance term and a medium attribute penalty term. The spatial distance term is calculated based on normalized spatial location information. The closer the spatial locations of the two points, the smaller the value of this term. Its significance is to allow the algorithm to tolerate spatial misalignment interference caused by vehicle vibration and slight positioning drift, avoiding misjudgments caused by such non-disease factors. The medium attribute penalty term is calculated by combining the normalized reference spectrum offset threshold and the measured frequency shift information. When the measured frequency shift is significantly less than the reference threshold (corresponding to the characteristics of lossless media), the value of this term will increase. Its function is to allow the algorithm to identify signals that do not conform to the characteristics of normal lossy media (such as asphalt), and accurately distinguish abnormal signals of underground cavity diseases.

[0091] Finally, by balancing the effects of the two terms through weighting coefficients, the complete transmission cost matrix is ​​obtained: preset attribute weighting balance coefficients. The value is set to 0.5 (experimentally calibrated to balance the effects of space and attributes), defining the u-th pixel in the measured distribution (corresponding to normalized coordinates). Normalized frequency shift ) is transferred to the v-th pixel in the baseline distribution (corresponding to normalized coordinates) Transmission cost : In the formula, : is the square of the Euclidean distance, representing the cost of moving energy through space; : is a linear rectification function used to filter out frequency shift differences that meet certain conditions; The normalized baseline spectral offset threshold; For spatial distance; This is a penalty item for media properties.

[0092] The transmission cost for each u and v is calculated using a weighted summation formula of spatial distance term and medium property penalty term, and then integrated into a transmission cost matrix.

[0093] S53. Based on the transmission cost matrix, determine the echo characteristic difference value that characterizes the essential difference between the measured signal and the benchmark model.

[0094] In some implementations, a total transmission cost function is first constructed. This function is calculated by taking the proportion of energy transmitted from the measured data point to the reference data point as a percentage of the total energy value at that measured data point (i.e., the proportion of energy at that point in the measured echo distribution to the total measured energy), and comparing this proportion with the cost at the corresponding position in the transmission cost matrix. Multiply the results and then sum the products of all pixel pairs. The sum is the total cost of transferring energy from the measured distribution to the reference distribution. In the formula, This represents the proportion of energy transmitted from the measured pixel u to the reference pixel v; the total cost is the quantified distance between the measured distribution and the reference distribution, reflecting the degree of essential difference between the two.

[0095] Then, an iterative algorithm (such as Sinkhorn iteration) is used to solve for the minimum total transmission cost: first, the relevant matrices and vectors are initialized, and then the parameters are updated alternately through loop iteration. The proportion of energy transmission is adjusted in each iteration until the preset number of iterations is completed. The approximate optimal result is obtained by a finite number of fast iterations, avoiding excessive computation time from affecting detection efficiency.

[0096] After iteration, the minimum total cost calculated is the echo characteristic difference value of the current detection segment. This value characterizes the essential difference between the measured echo signal and the normal background echo model: the larger the value, the more difficult it is for the measured signal to be classified as normal background clutter through spatial misalignment adjustment and medium property interpretation, which means that the possibility of real hidden defects in this area is higher.

[0097] In some implementations, before determining the structural damage assessment results for the target area in S5, the following steps are included: extracting the echo characteristic difference values ​​corresponding to all detection segments within a preset healthy area to form a difference value sequence. Then, the arithmetic mean and standard deviation of the difference value sequence are calculated to quantify the central tendency and dispersion of the difference values ​​under normal background conditions. Based on the statistical 3σ criterion, the sum of the arithmetic mean and three times the standard deviation is used as the system noise tolerance to determine the maximum statistical residual that normal background clutter can produce under the current runway structure, surface condition, and equipment noise floor.

[0098] In some implementations, the method for determining the structural damage assessment results of the target area in S5 can be specifically implemented through the following S54 to S55, which are explained in detail below: S54. If the echo characteristic difference value of the target area is less than or equal to the system noise tolerance, the target area is determined to be a normal area.

[0099] If the echo characteristic difference value of the target area is less than or equal to the system noise tolerance, it means that the difference is caused by normal background interference (rather than underground disease), which is consistent with the difference fluctuation range of the healthy area; only when the difference exceeds this tolerance does it mean that there is an abnormal signal that cannot be explained by normal background (i.e., characteristic difference caused by underground disease).

[0100] S55. If the echo characteristic difference value of the target area is greater than the system noise tolerance, the target area is determined as an abnormal candidate segment. The structural damage assessment value is determined by combining the echo characteristic difference value of the target area, the system noise tolerance, and the foundation curvature value.

[0101] The structural damage assessment value comprehensively reflects the scale of micro-defects and the vulnerability of the macro-structure.

[0102] In some implementations, the echo characteristic difference value corresponding to the abnormal candidate segment is retrieved. System noise tolerance Simultaneously, using the physical coordinates of the target area as an index, the foundation curvature value at the corresponding location is retrieved from the foundation curvature grid table generated by the macroscopic mechanical field construction module. Calculate structural damage assessment value : In the formula, The preset smoothing factor is a very small positive number, for example, 1.0 × 10⁻⁶. -6 This is used to avoid the denominator being zero; The preset structural vulnerability coefficient, for example, is 1.5, used to adjust the weight of foundation risk in the overall assessment; This is a preset curvature reference threshold, for example, a value of 2.0 × 10. -4 m -1 , representing the critical curvature of the foundation when it enters the nonlinear deformation stage; The signal-to-noise ratio term for micro-defects represents the signal-to-noise ratio multiple of micro-defects relative to the background noise level, directly reflecting the physical scale of underground micro-defects. This is a macroscopic foundation risk amplification term, characterizing the risk amplification effect caused by macroscopic foundation bending, and reflecting the expansion risk of defects under high stress conditions in the foundation.

[0103] The risk of microscopic defects is amplified by the degree of macroscopic foundation curvature. Multiplication can reflect the physical logic of the integration of defect scale and risk amplification coefficient, which is more in line with the actual development law of disease.

[0104] Furthermore, the method also includes: outputting corresponding graded early warning results based on the structural damage assessment value according to the preset grading standards, and generating a visualized distribution map of the runway health status.

[0105] In some implementation methods, the pre-stored runway defect engineering classification standard is first retrieved. This standard is used to calibrate classic runway defect samples (including borehole-verified voids, cavities, interlayer debonding, etc.), specifically: Level 1 Warning (Red, High Risk): Corresponding structural damage assessment value ≥3.0 indicates large-scale voids in high foundation stress zones or giant cavities in medium stress zones; Level 2 Warning (Yellow, Attention): Corresponds to 1.5≤ <3.0 indicates obvious voids in low-stress areas or minor defects in high-stress areas; Level 3 Warning (Blue, Slight): Corresponds to 1.0 < <1.5 indicates weak signal anomalies in low-stress areas (such as inactive foreign matter left over from construction, or initial interlayer debonding). Normal regions (i.e., regions where the echo characteristic difference value is ≤ the system noise tolerance) are marked in green.

[0106] The warning level (or normal marking) of each detection segment is associated with the corresponding physical coordinates of the runway's local Cartesian coordinate system to clarify the actual spatial location of each risk area. Subsequently, the geographic information visualization module built into the vehicle system maps different warning levels to corresponding colors and marks the longitudinal mileage range of each area; finally, an interactive runway health status distribution map is generated, supporting operations such as spatial positioning and area zooming.

[0107] Based on the above technical solution, by integrating three types of data—satellite deformation monitoring, UAV surface imagery, and vehicle-mounted ground-penetrating radar underground echo—the system first uses satellite data to generate curvature distribution data characterizing the complexity of the foundation structure, and then uses UAV imagery to obtain surface repair material marking data. Combining these two types of data, a baseline background echo distribution that closely matches the actual environment of the target area is constructed. Finally, by analyzing the characteristic differences between the baseline and measured echoes and coupling foundation curvature information, damage assessment is performed. This effectively solves the problems of difficulty in distinguishing between surface clutter and deep-seated defects, as well as false alarms caused by fragmented operation and maintenance data. It enables accurate identification of hidden defects in airport runways and provides a scientific basis for runway preventive maintenance.

[0108] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0109] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0110] In this embodiment of the invention, the air-ground fusion-based airport runway defect precision identification device can be divided into functional units according to the above method example. For example, each function can be divided into its own functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or software. It should be noted that the unit division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0111] This invention also provides a schematic diagram of the hardware structure of a precise airport runway defect identification device based on sky-ground fusion, see [link / reference]. Figure 3 The airport runway defect precision identification device 300 based on sky-ground fusion includes a processor 301, and optionally, a memory 302 connected to the processor 301.

[0112] In the first possible implementation, see Figure 3The airport runway defect precision identification device 300 based on sky-ground fusion also includes a transceiver 303. The processor 301, memory 302, and transceiver 303 are connected via a bus. The transceiver 303 is used to communicate with other devices or communication networks. Optionally, the transceiver 303 may include a transmitter and a receiver. The device in the transceiver 303 that implements the receiving function can be considered as a receiver, which is used to perform the receiving steps in the embodiments of the present invention. The device in the transceiver 303 that implements the transmitting function can be considered as a transmitter, which is used to perform the transmitting steps in the embodiments of the present invention.

[0113] Based on the first possible implementation method Figure 3 The structural diagram shown can be used to illustrate the structure of the airport runway defect precision identification device based on sky-ground fusion involved in the above embodiments.

[0114] in, Figure 3 The diagram also illustrates the system chip in a precise airport runway defect identification device based on sky-ground fusion. In this case, the actions performed by the aforementioned precise airport runway defect identification device based on sky-ground fusion can be implemented by this system chip. The specific actions performed are described above and will not be repeated here.

[0115] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings and the disclosure, will understand and implement other variations of the disclosed embodiments in carrying out the claimed invention. In this invention, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several of the functions listed in this invention.

[0116] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative of the invention and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the invention and its equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A method for accurate identification of airport runway defects based on sky-ground fusion, characterized in that, include: Acquire deformation monitoring data collected by satellite, and generate foundation curvature distribution data covering the runway area based on the deformation monitoring data; Acquire runway surface image data collected by UAV, and identify the image data to generate surface material marker data characterizing the distribution of surface repair materials; Acquire runway underground echo data collected by vehicle-mounted ground-penetrating radar, and generate measured echo distribution data based on the runway underground echo data; Based on the ground curvature distribution data and surface material marking data corresponding to the target area, a baseline background echo distribution data for the target area is constructed; the baseline background echo distribution data characterizes the energy pattern that radar echoes should present under normal background conditions. The difference in echo characteristics between the baseline background echo distribution data and the measured echo distribution data of the target area is analyzed, and combined with the foundation curvature distribution data, the structural damage assessment results of the target area are determined.

2. The method for accurate identification of airport runway defects according to claim 1, characterized in that, Generating the foundation curvature distribution data includes: Establish a local Cartesian coordinate system for the runway area; the local Cartesian coordinate system takes the center point of the runway entrance as the origin, the direction extending along the runway centerline as the longitudinal axis, the direction perpendicular to the centerline pointing to the side of the runway as the transverse axis, and the direction perpendicular to the ground downwards as the depth axis. The deformation monitoring data collected by the satellite is projected from the original geographic coordinate system to the local Cartesian coordinate system to obtain scatter data containing physical coordinates and cumulative settlement. The scattered data is processed using a preset interpolation algorithm to generate a cumulative settlement rule grid covering the runway area; The cumulative settlement regular grid is processed by second-order differential to obtain the foundation curvature value, which characterizes the stress bending degree and complexity of the runway underground structure, and the foundation curvature distribution data is generated.

3. The method for accurate identification of airport runway defects according to claim 2, characterized in that, Generating the surface material marker data includes: Register the runway surface image data collected by the UAV to the local Cartesian coordinate system; A preset semantic segmentation algorithm is used to classify the surface material of the registered image data, identify asphalt repair areas and original concrete areas, and generate surface material marker data. The surface material marker data is associated with the longitudinal position of the runway and is used to indicate whether asphalt repair material exists within the lateral coverage area of ​​the corresponding longitudinal position.

4. The method for accurate identification of airport runway defects according to claim 1, characterized in that, Acquire runway subsurface echo data collected by vehicle-mounted ground-penetrating radar, and generate measured echo distribution data based on the runway subsurface echo data, including: Using the pulse signal from a high-precision odometer as a trigger source, the vehicle-mounted ground-penetrating radar is controlled to collect underground echo data of the runway at preset intervals; The runway underground echo data is preprocessed by removing the DC component, automatic gain control, and instantaneous amplitude envelope extraction to obtain the measured echo distribution data.

5. The method for accurate identification of airport runway defects according to claim 1, characterized in that, Construct baseline background echo distribution data for the target area, including: The underground echo signal of the exposed concrete section within the preset healthy area is extracted, and the reference attenuation rate is obtained by fitting the attenuation curve of the signal amplitude envelope; the reference attenuation rate characterizes the energy loss benchmark of electromagnetic waves in the normal runway medium. Based on the benchmark attenuation rate, the foundation curvature distribution data and surface material marking data corresponding to the target area, the regional structural clutter attenuation factor is determined; the regional structural clutter attenuation factor is used to characterize the background clutter intensity and attenuation trend of the target area. Based on the regional structural clutter attenuation factor, a benchmark background echo distribution data with the same dimension as the measured echo distribution data is constructed.

6. The method for accurate identification of airport runway defects according to claim 5, characterized in that, The differences in echo characteristics between the baseline background echo distribution data and the measured echo distribution data of the target area are analyzed, including: The underground echo signal at the asphalt joint within the preset healthy area is extracted, and a reference spectrum offset threshold is obtained through spectrum analysis; the reference spectrum offset threshold is used to distinguish the spectral characteristics of lossy media and lossless media. Based on the reference spectrum offset threshold, measured echo distribution data, and reference background echo distribution data, a transmission cost matrix is ​​constructed. The transmission cost matrix consists of a spatial distance term and a medium property penalty term. The spatial distance term represents the cost of energy moving in space, and the medium property penalty term is used to distinguish the signal differences corresponding to media with different loss characteristics based on the reference spectrum offset threshold. Based on the transmission cost matrix, the echo characteristic difference value, which characterizes the essential difference between the measured signal and the benchmark model, is determined.

7. The method for accurate identification of airport runway defects according to claim 6, characterized in that, Before determining the structural damage assessment results for the target area, the following steps are also included: Extract the echo feature difference values ​​corresponding to all detection segments within the preset healthy area to form a difference value sequence; The arithmetic mean and standard deviation of the difference value sequence are calculated, and the sum of the arithmetic mean and three times the standard deviation is used as the system noise margin; the system noise margin represents the maximum statistical residual that normal background clutter can produce.

8. The method for accurate identification of airport runway defects according to claim 7, characterized in that, Determine the structural damage assessment results for the target area, including: If the echo characteristic difference value of the target area is less than or equal to the system noise tolerance, the target area is determined to be a normal area. If the echo characteristic difference value of the target area is greater than the system noise tolerance, the target area is determined as an abnormal candidate segment. The structural damage assessment value is determined by combining the echo characteristic difference value of the target area, the system noise tolerance, and the foundation curvature value. The structural damage assessment value comprehensively reflects the scale of micro defects and the vulnerability of macro structures.

9. The method for accurate identification of airport runway defects according to claim 8, characterized in that, Determining the structural damage assessment results for the target area also includes: Based on the preset grading standards, the corresponding grading warning results are output according to the structural damage assessment values, and a visualized runway health status distribution map is generated.

10. A precise identification system for airport runway defects based on sky-ground fusion, characterized in that, include: The module includes a macroscopic mechanical field construction module, a multidimensional data perception module, a feature calculation module, and a risk assessment module. The macroscopic mechanical field construction module is used to acquire deformation monitoring data collected by satellite and generate foundation curvature distribution data covering the runway area based on the deformation monitoring data. The multi-dimensional data perception module is used to acquire runway surface image data collected by the UAV and identify the image data to generate surface material marker data that characterizes the distribution of surface repair materials. The multi-dimensional data sensing module is also used to acquire runway underground echo data collected by vehicle-mounted ground-penetrating radar, and generate measured echo distribution data based on the runway underground echo data. The feature calculation module is used to construct the reference background echo distribution data of the target area based on the foundation curvature distribution data and surface material marking data corresponding to the target area; the reference background echo distribution data represents the energy form that the radar echo should present under normal background. The feature calculation module is also used to analyze the echo feature difference values ​​between the reference background echo distribution data and the measured echo distribution data of the target area; The risk assessment module is used to determine the structural damage assessment results of the target area based on the echo characteristic difference value and the foundation curvature distribution data.

Images