BIM-based airport runway pavement intelligent management method and system, and medium

Abstract

The invention discloses a BIM-based airport runway pavement intelligent management method and system, and a medium, and belongs to the technical field of intelligent management. The BIM-based airport runway pavement intelligent management method comprises the steps of controlling an unmanned aerial vehicle to fly along a preset path plan; collecting a surface state image of the airport runway through an airborne camera of the unmanned aerial vehicle, and marking the position of the airport runway to obtain a data set of airport runway surface state data; performing pavement disease identification on each piece of airport runway surface state data in the data set of the airport runway surface state data to obtain a data set of pavement disease information; and performing geospatial registration on the data set of the pavement disease information, and mapping the pavement disease information to corresponding component elements in the airport runway BIM model by utilizing coordinate matching to obtain an updated airport runway BIM model. The airport runway BIM model can be updated, and intelligent management of the airport runway pavement is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of intelligent management, and more specifically, to a method, system, and medium for intelligent management of airport runway pavement based on BIM. Background Art

[0002] In the operation and maintenance of airport infrastructure, real-time monitoring and precise maintenance of runway pavement conditions are directly related to flight safety and operational efficiency. Currently, the mainstream inspection methods still rely primarily on manual visual inspections and handheld devices. These methods suffer from limited coverage, long data collection cycles, and subjective identification of defects. This makes frequent inspections difficult, especially in complex weather conditions.

[0003] The advancement of machine vision technology has opened up new possibilities for intelligent inspection of airport runway pavements. Furthermore, several image recognition algorithms have been developed for identifying pavement conditions. However, airport runway defects can vary greatly, ranging from tiny early cracks to extensive structural damage. The visual characteristics of the same defect can also differ significantly under varying lighting, humidity, and camera angles. Early, subtle defects, in particular, have visual features that are poorly distinguishable from background textures, making them prone to missed or misdetected detections. Furthermore, runway defect identification requires not only determining the defect type but also precisely locating the defect to ensure efficient subsequent maintenance. Traditional defect records are often based on stake numbers or two-dimensional drawings, resulting in abstract and unintuitive information and limited intelligent management. These factors have become key bottlenecks hindering the advancement of intelligent airport runway management. Therefore, an optimized BIM-based intelligent airport runway pavement management method and system is highly desired. Summary of the Invention

[0004] This invention provides a BIM-based intelligent airport runway pavement management method, system, and medium that aims to at least partially or completely address the technical issues of missed or misdetected airport runway pavement defects. Traditional defect records are often based on pile numbers or two-dimensional drawings, resulting in abstract and less intuitive information and insufficient intelligent management. To achieve the objectives of this invention, the technical solutions of this invention are as follows:

[0005] First, a BIM-based intelligent management method for airport runway pavement includes:

[0006] Control the drone to fly along the preset path;

[0007] The surface state image of the airport runway is collected by using an onboard camera of a drone and the position of the airport runway is marked to obtain a data set of the airport runway surface state data, wherein the airport runway surface state data includes the airport runway surface state image and position coordinates;

[0008] Performing pavement disease identification on each airport runway surface state data set in the airport runway surface state data set to obtain a pavement disease information data set, including: performing feature pixel-level saliency enhancement on each airport runway surface state data based on the runway surface state information to obtain visually significant coding features of the airport runway surface; and determining pavement disease information based on the visually significant coding features of the airport runway surface;

[0009] The dataset of pavement damage information is geospatially registered, and the pavement damage information is mapped to the corresponding component elements in the airport runway BIM model using coordinate matching to obtain an updated airport runway BIM model.

[0010] Optionally, performing feature pixel-level saliency enhancement based on the runway surface state information on the surface state data of each airport runway to obtain visually salient coding features of the airport runway surface includes:

[0011] Extracting runway surface visual features from the airport runway surface state image to obtain airport runway surface visual coding features;

[0012] The airport runway surface visual coding features are saliently processed to obtain the airport runway surface visual salient coding features.

[0013] Optionally, extracting runway surface visual features from the airport runway surface state image to obtain airport runway surface visual coding features includes:

[0014] The airport runway surface state image is passed through a runway surface visual feature extractor based on a void convolutional neural network model to obtain an airport runway surface visual feature coding map as the airport runway surface visual coding feature.

[0015] Optionally, performing runway surface visual feature saliency on the airport runway surface visual coding features to obtain airport runway surface visual salient coding features includes:

[0016] Extracting pixel-level airport runway surface visual features from the airport runway surface visual feature coding map to obtain a pixel-level airport runway surface visual feature vector to be enhanced;

[0017] Perform feature distillation on the pixel-level airport runway surface visual feature vector to be enhanced to obtain the pixel-level airport runway surface visual feature vector to be enhanced;

[0018] Based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level distilled airport runway surface visual feature vector to be enhanced, the surface visual feature saliency enhancement is performed on the pixel-level distilled airport runway surface visual feature vector to be enhanced to obtain an enhanced pixel-level airport runway surface visual feature vector and use it as the airport runway surface visual saliency coding feature, wherein the enhanced pixel-level airport runway surface visual feature vector is the channel feature vector of the (i, j)th pixel position of the airport runway surface visual feature saliency coding map.

[0019] Optionally, extracting pixel-level airport runway surface visual features from the airport runway surface visual feature coding map to obtain a pixel-level airport runway surface visual feature vector to be enhanced includes:

[0020] The visual feature encoding map of the airport runway surface is decoupled along the channel dimension to obtain a set of pixel-level airport runway surface visual feature vectors.

[0021] The pixel-level airport runway surface visual feature vector at the (i, j)th pixel position is extracted from the set of pixel-level airport runway surface visual feature vectors as the pixel-level airport runway surface visual feature vector to be enhanced.

[0022] Optionally, based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced, performing surface visual feature significance enhancement on the pixel-level airport runway surface visual feature vector to be enhanced to obtain an enhanced pixel-level airport runway surface visual feature vector, including:

[0023] Based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced and distilled, the size of the feature receptive field of the pixel-level airport runway surface visual feature vector to be enhanced and distilled is determined;

[0024] Based on the size of the feature receptive field, the set of pixel-level airport runway surface visual feature vectors within the local receptive field is filtered out from the set of pixel-level airport runway surface visual feature vectors;

[0025] Based on the set of pixel-level airport runway surface visual feature vectors within the local receptive field, the distilled pixel-level airport runway surface visual feature vector is significantly enhanced to obtain the enhanced pixel-level airport runway surface visual feature vector.

[0026] Optionally, based on the size of the feature receptive field, a set of pixel-level airport runway surface visual feature vectors within the local receptive field is filtered out from the set of pixel-level airport runway surface visual feature vectors, including:

[0027] Based on the size of the feature receptive field, a set of pixel-level airport runway surface visual feature vectors within a preliminary local receptive field is screened from the set of pixel-level airport runway surface visual feature vectors;

[0028] The set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field is subjected to regularized optimization of visual features based on boundary surface-volume space to obtain the set of pixel-level airport runway surface visual feature vectors within the local receptive field.

[0029] Optionally, determining pavement damage information based on visually significant coding features of the airport runway surface includes: performing image semantic segmentation on a visually significant coding map of the airport runway surface to obtain a semantic segmentation result of the airport runway surface state; and determining pavement damage information based on the semantic segmentation result of the airport runway surface state, the pavement damage information including a surface damage type, a surface damage location, geometric parameters of the surface damage, and a preliminary severity of the surface damage.

[0030] And / or, geospatial registration is performed on the dataset of pavement damage information, and coordinate matching is used to map the pavement damage information to corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model, including: extracting the location of surface damage from the pavement damage information, where the location of the surface damage is the pixel coordinates of the center point of the damage; based on the location of the surface damage, calculating the actual geographic coordinates of the surface disaster; performing a spatial query on the actual geographic coordinates of the surface disaster in the BIM database to obtain an identifier of the component element; and based on the identifier of the construction element, mapping the pavement damage information to the corresponding component element in the airport runway BIM model to obtain an updated airport runway BIM model.

[0031] In a second aspect, a BIM-based intelligent airport runway pavement management system is provided, using any of the BIM-based intelligent airport runway pavement management methods described in the first aspect, including:

[0032] The drone control module controls the drone to fly along the preset path;

[0033] A surface state data acquisition module collects surface state images of the airport runway through the drone's onboard camera and marks the location of the airport runway to obtain a dataset of airport runway surface state data. The airport runway surface state data includes the airport runway surface state image and location coordinates.

[0034] A pavement defect recognition module identifies pavement defects on each runway surface condition data set to obtain a pavement defect information data set. This module includes: performing pixel-level saliency enhancement on each runway surface condition data set based on the runway surface condition information to obtain visually significant coding features of the runway surface; and determining pavement defect information based on the visually significant coding features of the runway surface.

[0035] The BIM model update module performs geospatial registration on the pavement damage information dataset and uses coordinate matching to map the pavement damage information to the corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model.

[0036] In a third aspect, a BIM-based intelligent airport runway pavement management and processing device includes a memory and a processor that are communicatively connected, the memory being used to store a computer program, the processor being used to read the computer program, and executing any of the BIM-based airport runway pavement intelligent management methods described in the first aspect above.

[0037] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored on the computer-readable storage medium, and when the instructions are run on a computer, any one of the BIM-based intelligent management methods for airport runway pavements described in the first aspect is executed.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) In the present invention application, the pixel-level recognition technology based on saliency enhancement is fully utilized to detect pavement defects in the surface status images of the airport runway collected by the UAV, accurately capturing various types of defects including minor ones and effectively overcoming the interference of complex background. At the same time, the invention is committed to associating the detailed information of the precisely located defects with the specific components in the airport runway BIM model to construct a dynamically updated digital information model of the runway health status, which is beneficial to the intelligent management of the airport runway pavement.

[0040] (2) In the present invention, the spatial structural characteristics of the distribution of the airport runway surface features are dynamically sensed to autonomously determine the optimal receptive field size for each pixel position. For example, the receptive field is automatically shrunk in the shadow area of ​​the pavement joint to focus on fine cracks, and the receptive field is expanded in the reflective area of ​​the marking to suppress optical artifact interference. Specifically, in this process, redundant information is first stripped away through feature decoupling in the channel dimension, so that the visual feature vector of the airport runway surface at each pixel point is expressed independently, laying the foundation for subsequent refined processing; then, the core features of the disease are extracted through feature distillation, and the noise components irrelevant to the runway surface state are removed; in the dynamic receptive field determination stage, the algorithm adaptively matches the local context perception range based on the feature distribution entropy and spatial correlation of the pixels to be enhanced, and intelligently adjusts the receptive field size and shape according to different morphological characteristics of the disease (such as the continuity of linear cracks and the discreteness of block threshing), ensuring that it can capture the macroscopic extension law of the disease while retaining the microscopic detail characteristics; in particular, finally, combined with the boundary surface-volume space regularization technology, the conformal expression of the feature vector within the local receptive field is constrained to ensure the geometric consistency between the disease boundary and the voxel space.

[0041] (3) In the present invention, the generated semantic segmentation results of the airport runway surface state not only include basic information such as the type of damage and contour location, but also achieve automated measurement of geometric parameters such as crack length and threshing area by integrating the spatial coordinates of BIM components, providing a quantitative basis for maintenance priority assessment. This structured output directly supports the spatial mapping of damage information and BIM models, allowing the cumulative damage associated with facility components such as pavement joints and lighting strips to be visually tracked, significantly improving the spatiotemporal accuracy of preventive maintenance decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and other objects, features, and advantages of the present invention will become more apparent through a more detailed description of the present invention in conjunction with the accompanying drawings. The accompanying drawings are provided to provide a further understanding of the present invention and constitute a part of the specification. Together with the present invention, they are used to explain the present invention and are not intended to limit the present invention. In the drawings, the same reference numerals generally represent the same components or steps.

[0043] Figure 1 This is a flow chart of the BIM-based intelligent airport runway pavement management method applied for by the present invention;

[0044] Figure 2 This is a schematic diagram of the principle of determining pavement damage information of the BIM-based intelligent management method for airport runway pavement applied for by the present invention;

[0045] Figure 3 A schematic flow chart of step S3 of the present invention;

[0046] Figure 4 A schematic flow chart of step S31 of the present invention;

[0047] Figure 5 This is a schematic diagram of the composition structure of the BIM-based airport runway pavement intelligent management system applied for in the present invention. DETAILED DESCRIPTION

[0048] Below, the exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0049] As used in the present application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0050] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0051] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0052] Below, the exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0053] Airport runway pavement diseases

[0054] In the existing field of airport runway pavement technology, airport cement concrete pavements are crucial infrastructure for ensuring safe aircraft takeoff and landing. However, due to long-term impacts from factors such as aircraft loads, climate change, and vehicle dispatch, airport runway pavements often develop various surface defects, such as peeling, roughness, threshing, exposed stones, dusting, and other surface wear and tear, as well as water damage. Concrete is also subject to long-term erosion by factors such as ultraviolet rays and acid rain, leading to surface aging and spalling, cracks, and oil stains. These defects not only affect the performance and service life of airport runway pavements but can also endanger flight safety during takeoff. Timely monitoring of runway pavement conditions is crucial to ensuring flight safety.

[0055] Intelligent management method of airport runway pavement based on BIM

[0056] like Figures 1 to 4 As shown, in the first aspect, a BIM-based intelligent management method for airport runway pavement includes:

[0057] Step S1, controlling the UAV to fly along a preset path;

[0058] Step S2, collecting a surface state image of the airport runway using an onboard camera of the drone and marking the location of the airport runway to obtain a data set of airport runway surface state data, where the airport runway surface state data includes the airport runway surface state image and location coordinates;

[0059] Step S3, performing pavement damage identification on each airport runway surface condition data set in the airport runway surface condition data set to obtain a data set of pavement damage information;

[0060] In step S4, geospatial registration is performed on the dataset of pavement damage information, and coordinate matching is used to map the pavement damage information to corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model.

[0061] In some embodiments, in step S1, the drone is controlled to fly along a preset path plan. A predetermined route can be pre-configured to control the drone to fly along the preset planned path. The drone can be equipped with a camera.

[0062] In some embodiments, due to the presence of dynamically changing elements such as navigation facilities and temporary construction areas around the runway, preferably, the preset path planning integrates the runway geometric parameters, clearance restrictions and real-time airspace status data in the BIM model, and uses an ant colony optimization algorithm under three-dimensional space constraints to generate a globally optimal flight trajectory. At the same time, a real-time obstacle avoidance module is embedded to deal with sudden obstacles, so that the drone can ensure full-section scanning with centimeter-level positioning accuracy in complex operating environments, avoid interference from sensitive areas such as weather radar, and maintain a constant image acquisition resolution through an adaptive elevation adjustment mechanism.

[0063] In some embodiments, the drone's flight path is first designed based on the specific layout of the airport's runway, daily operations, and the key areas to be monitored. This path is not fixed but rather incorporates the airport's real-time operational and environmental conditions, allowing for dynamic adjustments using a pre-set algorithm. In practice, technicians pre-enter the drone's control system with this path information and set relevant parameters, such as speed and altitude. After takeoff, the drone autonomously follows the pre-set path while simultaneously capturing real-time images of the runway surface using its onboard camera. During this process, the drone can adapt to the situation, such as automatically circumventing obstacles when encountering them, ensuring the continuity and integrity of data collection. This approach significantly improves the coverage and frequency of runway surface condition monitoring, providing high-quality data support for subsequent pavement defect identification.

[0064] In some embodiments, in step S2, a surface state image of an airport runway is captured by an onboard camera of a drone and the location of the airport runway is marked to obtain a data set of airport runway surface state data, where the airport runway surface state data includes the airport runway surface state image and location coordinates.

[0065] It should be understood that in order to obtain high-quality, high-resolution images of the airport runway surface status, combined with precise location coordinate information to provide a solid data foundation for subsequent pavement disease identification and analysis, in the technical solution of the present invention, the surface status image of the airport runway can be collected by the drone's onboard camera and the airport runway's location can be marked to obtain a data set of the airport runway surface status data. Here, the high-definition camera carried by the drone can capture every detail of the runway surface, including tiny defects such as cracks and threshing, which are often difficult to detect with human vision. At the same time, recording accurate location coordinates is crucial for accurately mapping disease information into the BIM model, which can ensure the targeted and effective maintenance work, thereby improving the efficiency of the entire operation and maintenance process.

[0066] In some embodiments, after takeoff, the drone autonomously flies along a pre-set path, simultaneously capturing real-time images of the runway surface using its high-resolution camera. Simultaneously, the onboard GPS or other positioning system records the exact coordinates of each captured point. It's worth noting that to ensure image quality and accurate location information, the drone may capture a specific area multiple times or from different angles. All collected images and their corresponding location coordinates are then integrated into a complete dataset. The resulting runway surface condition data dataset includes not only detailed images of the runway surface but also precise location information, laying a solid foundation for subsequent pavement defect identification and geospatial registration. This approach ensures data diversity and richness, providing strong support for subsequent image analysis. Furthermore, thanks to advanced positioning technology, the acquired location coordinates are highly accurate, enabling accurate mapping of defect information into 3D BIM models, further enhancing the scientific and accurate nature of maintenance decisions.

[0067] In some embodiments, in step S3, pavement damage identification is performed on each of the airport runway surface condition data in the data set to obtain a data set of pavement damage information. For example, step S3 may include:

[0068] Step S31, performing feature pixel-level saliency enhancement on the surface state data of each airport runway based on the runway surface state information to obtain visually salient coding features of the airport runway surface;

[0069] Step S32: determining pavement damage information based on visually significant coding features of the airport runway surface.

[0070] In some embodiments, in step S31, feature pixel-level saliency enhancement based on the runway surface state information is performed on each airport runway surface state data to obtain visually significant coding features of the airport runway surface. For example, step S31 may include:

[0071] Step S311, extracting runway surface visual features from the airport runway surface state image to obtain airport runway surface visual coding features;

[0072] Step S312: performing runway surface visual feature saliency on the airport runway surface visual coding features to obtain airport runway surface visual salient coding features.

[0073] In some embodiments, in step S311, it should be understood that traditional convolution operations are limited by local receptive fields and are unable to effectively capture the long-range continuous features of defects such as cracks and threshing. However, by introducing a controllable dilation rate, dilated convolution can construct a multi-scale receptive field without reducing the resolution of the feature map. This feature enables the network to capture macrostructural features such as pavement joints and markings when extracting runway surface texture, while also preserving the continuous morphology of microcracks through skip sampling.

[0074] Therefore, in the technical solution of the present invention, the airport runway surface status image can be passed through a runway surface visual feature extractor based on a dilated convolutional neural network model to extract multi-level airport runway surface visual features of local texture and global semantics, and obtain an airport runway surface visual feature coding map. The airport runway surface visual feature coding map can be used as or can obtain the airport runway surface visual coding features. In this way, the system's ability to understand the airport runway surface status is improved; at the same time, since the dilated convolution can effectively expand the receptive field without reducing the resolution of the airport runway surface status image, the obtained airport runway surface visual feature coding map contains rich detailed information, which is crucial for accurately identifying the type, location and severity of the disease. This high-precision feature extraction lays a solid foundation for subsequent image semantic segmentation and disease positioning, allowing the maintenance team to quickly respond and take necessary repair measures based on the analysis results, thereby ensuring the safety and operational efficiency of the airport runway.

[0075] In some embodiments, in step S312, it should be understood that since traditional algorithms use a static receptive field of fixed size, it is difficult to adapt to the multi-scale characteristics of runway surface defects such as cracks and threshing - long linear cracks require wide-area context association to capture their extension trend, while local threshing relies on fine analysis of micro-textures. Therefore, in the technical solution of the present invention, the visual coding features of the airport runway surface are salient to obtain the visually significant coding features of the airport runway surface. That is, by dynamically sensing the spatial structural characteristics of the distribution of airport runway surface features, the optimal receptive field size for each pixel position is autonomously determined. For example, the receptive field is automatically contracted in the shadow area of ​​the pavement joint to focus on fine cracks, and the receptive field is expanded in the reflective area of ​​the marking to suppress optical artifact interference. Specifically, in this process, redundant information is first stripped away through feature decoupling in the channel dimension, so that the visual feature vector of the airport runway surface at each pixel point is expressed independently, laying the foundation for subsequent refined processing; then, the core features of the disease are extracted through feature distillation, and the noise components irrelevant to the runway surface state are removed; in the dynamic receptive field determination stage, the algorithm adaptively matches the local context perception range based on the feature distribution entropy and spatial correlation of the pixels to be enhanced, and intelligently adjusts the receptive field size and shape according to different morphological characteristics of the disease (such as the continuity of linear cracks and the discreteness of block threshing), ensuring that it can capture the macroscopic extension law of the disease while retaining the microscopic detail characteristics; in particular, finally, combined with the boundary surface-volume space regularization technology, the conformal expression of the feature vector within the local receptive field is constrained to ensure the geometric consistency between the disease boundary and the voxel space. This dynamic feature enhancement mechanism breaks through the feature confusion bottleneck of traditional algorithms under the interference of pavement texture. By establishing a spatial saliency expression mechanism for disease features, it ensures that the morphological integrity and spatial continuity of millimeter-level diseases are accurately preserved. It not only strengthens the saliency expression of the disease area, but also establishes a deep association between the disease morphology and spatial context. The resulting airport runway surface visual feature saliency coding map not only carries semantic information such as disease type and geometric parameters, but also achieves precise separation of disease and normal pavement features through orthogonal decoupling of feature space, providing a high-fidelity data foundation for spatial attribute mapping of BIM models and component-level maintenance decisions.

[0076] Specifically, the method includes: first, performing visual feature decoupling on the airport runway surface visual feature coding map along the channel dimension to obtain a set of pixel-level airport runway surface visual feature vectors.

[0077] In the present application, it should be understood that the traditional multi-channel feature fusion mechanism has defects in its adaptability to complex pavement textures. For example, when runway markings, seams and disease features generate cross-channel coupling in the multi-layer feature map of the convolutional neural network, the regular geometric features of the markings are easily superimposed through the weights between channels to form a pseudo-significant response, which masks the weak feature expression of the real disease. Therefore, in the technical solution of the present application, in order to establish an atomic expression basis for pixel-level features, so that the subsequent processing module can perform a refined analysis of the independent feature vector of each pixel point and avoid the spatial masking effect of redundant information between channels on the disease features, the visual feature coding map of the airport runway surface is visually decoupled along the channel dimension to obtain a set of pixel-level airport runway surface visual feature vectors. That is, by decoupling the visual coding features along the channel dimension, the multi-dimensional channel features of each pixel point are disassembled into independent vectors, breaking the implicit linear correlation constraints between channels in the conventional convolution layer, so that local features such as the gradient mutation at the crack edge and the irregular texture of the threshing area can be separated from the interference caused by channel mixing, forming an independent representation unit of pixel granularity. Furthermore, by stripping away the coupling noise between channels, the decoupled pixel-level visual feature vectors of the runway surface accurately reflect the independent response strength of pixels across different semantic dimensions. For example, this enhances the directional gradient characteristics of crack edges in a specific channel while suppressing the uniform chromaticity characteristics of the marking area in another channel. This decoupling mechanism provides a pure feature space unaffected by channel interference for subsequent dynamic receptive field anchoring. This allows local context-based adaptive enhancement to focus on the morphological characteristics of the defect's essence, rather than the artifacts introduced by channel mixing. Ultimately, this results in a highly discriminative pixel-level defect feature representation system.

[0078] In some embodiments, the visual feature encoding map of the airport runway surface is decoupled along the channel dimension using the following decoupling formula to obtain a set of pixel-level airport runway surface visual feature vectors; wherein the decoupling formula is:

[0079] F∈R H×W×C

[0080]

[0081] Among them, F is the visual feature coding map of the airport runway surface, R is a real number set, H, W, and C are the height, width, and number of channels of F respectively, FeatureDecoupling(F) is the feature decoupling of F, and v 1,1 、v i,j and v H,W are the pixel-level airport runway surface visual feature vectors at the (1,1), (i,j) and (H,W) pixel positions in the set of pixel-level airport runway surface visual feature vectors, i=1,2,...,H; j=1,2,...,W.

[0082] Next, the pixel-level airport runway surface visual feature vector at the (i, j)th pixel position is extracted from the set of pixel-level airport runway surface visual feature vectors as the pixel-level airport runway surface visual feature vector to be enhanced.

[0083] Since traditional image processing methods are insufficient in focusing on local features, for example, when regular textures such as runway surface markings and seams interfere with diseased areas in the global feature space, pixel-based fine processing becomes the key to distinguishing semantic boundaries. Therefore, in the technical solution of the present invention, the pixel-level airport runway surface visual feature vector at the (i, j)th pixel position is extracted from the set of pixel-level airport runway surface visual feature vectors as the pixel-level airport runway surface visual feature vector to be enhanced. That is, by traversing the set of pixel-level airport runway surface visual feature vectors pixel by pixel, and selecting the pixel feature of the specific coordinate (i, j) as the processing anchor point, a local context analysis framework centered on the target pixel is established. In this way, the system can construct an independent processing unit with pixel granularity, so that subsequent feature enhancement can accurately capture local morphological features such as direction and threshing area discreteness according to the uniqueness of each pixel point, and avoid the loss of details caused by global feature pooling. Specifically, during this process, the system traverses the entire set of pixel-level visual feature vectors of the airport runway surface in a sliding window manner, and ensures that each pixel undergoes an independent enhancement process through a coordinate mapping mechanism to form a fine-grained processing network covering the entire map. In this way, local attributes such as the gradient mutation characteristics of the disease edge and the color uniformity characteristics of the marking area can be expressed independently, providing a pure feature base that is not interfered by the neighborhood for the anchoring of the dynamic receptive field. In addition, through the independent enhancement mechanism of the anchored pixels, the algorithm can adaptively adjust the processing strategy according to the feature differences in different regions, such as suppressing regular texture responses in areas with dense markings and amplifying abnormal signals in potential disease areas, ultimately forming a feature enhancement effect with spatial specificity, providing highly discriminative pixel-level input features for subsequent semantic segmentation.

[0084] In some embodiments, the pixel-level airport runway surface visual feature vector at the (i, j)th pixel position is extracted from the set of pixel-level airport runway surface visual feature vectors using the following extraction formula as the pixel-level airport runway surface visual feature vector to be enhanced; wherein the extraction formula is:

[0085] v tbs =v i,j ∈R C

[0086] Among them, v tbs is the visual feature vector of the pixel-level airport runway surface to be enhanced.

[0087] Then, feature distillation is performed on the pixel-level airport runway surface visual feature vector to be enhanced to obtain the pixel-level airport runway surface visual feature vector to be enhanced and distilled.

[0088] In the present invention application, it should be understood that when the pavement image collected by the UAV contains complex backgrounds such as light reflection and pavement markings, the original visual feature vector extracted by the dilated convolution is easily contaminated by redundant information, resulting in key disease features such as crack edge gradient and threshing texture anisotropy being submerged in irrelevant channel responses. Therefore, in order to construct a low-dimensional and compact feature representation space so that the subsequent processing modules can focus on the discriminative features of the disease essence and avoid the misleading of the noise component in the high-dimensional feature space on the adaptive calculation of the receptive field, in the technical solution of the present invention application, the pixel-level airport runway surface visual feature vector to be enhanced is subjected to feature distillation to obtain the pixel-level airport runway surface visual feature vector to be enhanced and distilled. Through the feature importance evaluation in the knowledge distillation process, the algorithm automatically identifies and enhances the weight distribution of key channels such as crack continuity and threshing degree, while suppressing channel responses that are not related to the disease. This distillation mechanism enables the processed feature vector to accurately represent the semantic association strength between pixels and surrounding areas. For example, it enhances the consistency of neighborhood gradients in the direction of crack extension and weakens color mutation interference at the boundaries of the marking line. This provides a denoised feature basis for intelligent anchoring of the dynamic receptive field, effectively improving the accuracy of capturing subtle defects in the subsequent saliency enhancement process.

[0089] In some embodiments, the pixel-level airport runway surface visual feature vector to be enhanced is subjected to feature distillation using the following feature distillation formula to obtain the pixel-level airport runway surface visual feature vector to be enhanced and distilled; wherein the feature distillation formula is:

[0090]

[0091] Among them, ||·|| is the norm of the vector, v s Pixel-level airport runway surface visual feature vector to be enhanced and distilled.

[0092] Then, based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced and distilled, the size of the feature receptive field of the pixel-level airport runway surface visual feature vector to be enhanced and distilled is determined.

[0093] Since the traditional fixed receptive field mechanism is not adaptable enough to the diverse morphologies of diseases, for example, when cracks extend in elongated forms and threshing forms discrete patches, it is difficult for the preset fixed-size convolution kernel to take into account the contextual association features of diseases of different scales. Therefore, in order to establish a pixel-level adaptive semantic perception scale so that the feature enhancement process can capture the global structure of large-scale diseases while retaining the detailed features of minor damage, in the technical solution applied for by the present invention, the size of the feature receptive field of the pixel-level airport runway surface visual feature vector to be enhanced and distilled is determined based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced and distilled. That is, by analyzing the spatial distribution characteristics of the feature vector after distillation, the semantic association range required for each pixel point is intelligently determined.

[0094] In the present application, for linear crack features, the algorithm automatically expands the receptive field based on the continuity of its directional gradient to capture the complete extension path; for point-like threshing features, the receptive field is shrunk to focus on the local texture mutation area. Specifically, through surface visual feature distribution entropy analysis and spatial correlation modeling, the system can identify high-information areas at the edge of the crack, dynamically match the receptive field size that can cover the entire length of the crack, and adjust the receptive field shape to fit the pollution boundary based on the irregular shape of the oil diffusion. This dynamic mechanism enables the regular texture of the marking area to trigger a small receptive field setting due to the low entropy feature, avoiding the introduction of irrelevant contextual noise; while the real diseased area activates a wide range of perception due to the complexity of the features, ensuring the integrity of the morphological features and providing an accurate spatial semantic basis for component-level disease mapping of the BIM model.

[0095] In some embodiments, based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level distilled airport runway surface visual feature vector to be enhanced, the size of the feature receptive field of the pixel-level distilled airport runway surface visual feature vector to be enhanced is determined using the following receptive field anchoring formula; wherein, the receptive field anchoring formula is:

[0096]

[0097] Where log2(·) is the logarithmic function with base 2, and r is the size of the feature receptive field.

[0098] Furthermore, based on the size of the feature receptive field, a set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field is screened out from the set of pixel-level airport runway surface visual feature vectors.

[0099] In the present application, it should be understood that the traditional fixed-area sampling method is not adaptable enough to complex disease morphologies. For example, when pavement cracks extend irregularly or threshing diseases form a discrete distribution, mechanically defining a fixed-size perception range can easily lead to the loss of key features or the introduction of noise. Therefore, in order to establish an initial context information pool that matches the disease morphology and provide basic data support for subsequent regular optimization, in the technical solution of the present application, based on the size of the feature receptive field, a set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field is screened from the set of pixel-level airport runway surface visual feature vectors. That is, by dynamically matching the feature receptive field size, a preliminary candidate area is intelligently delineated in the set of pixel-level feature vectors.

[0100] In the present application, for linear crack features, the algorithm expands the sampling range along the main extension axis based on its directional continuity; for patchy threshing, a radial search is used to capture discrete distribution features. In this process, through the spatial coordinate mapping mechanism, the system quickly locates the candidate feature vector group centered on the target pixel, covering the possible spatial distribution range of the disease. This dynamic screening mechanism allows key information such as the gradient change characteristics of the crack edge and the texture mutation characteristics of the threshing area to be fully retained, while excluding interference signals from irrelevant remote areas. The set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field formed by preliminary screening can not only reflect the local morphological characteristics of the disease, but also have sufficient spatial scalability, providing a solid foundation for the formation of significant feature expressions that conform to the spatial distribution law of pavement diseases.

[0101] In some embodiments, based on the size of the feature receptive field, the following screening formula is used to screen out the set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field from the set of pixel-level airport runway surface visual feature vectors; wherein the screening formula is:

[0102]

[0103] Where W is the set of pixel-level visual feature vectors of the airport runway surface within the initial local receptive field, v i-r,j-r 、v i+r,j-r 、 v m ,n 、v i-r,j+r and v i+r,j+r are the preliminary local receptive field pixel-level airport runway surface visual feature vectors at the (ir, jr), (i+r, jr), (m, n), (ir, j+r) and (i+r, j+r) pixel positions in the set of pixel-level airport runway surface visual feature vectors in the preliminary local receptive field, respectively.

[0104] In particular, it should be understood that in the detection of airport runway surface defects, the set of pixel-level airport runway surface visual feature vectors within the initially screened local receptive field may have problems with blurred boundaries and distorted voxel representations due to interference from pavement marking reflections, seam shadows, and complex textures. In traditional methods, the representation of boundary features (such as crack edges) and volume space features (such as the three-dimensional morphology of threshing areas) is often separated from each other, resulting in the defect features being submerged by environmental noise. Therefore, it is necessary to eliminate the spatial ambiguity of feature representation so that complex morphologies such as the fuzzy boundaries of oil diffusion and the fine lines of crack extension can be accurately analyzed within a unified representation framework.

[0105] In some embodiments, a set of pixel-level airport runway surface visual feature vectors within a preliminary local receptive field is subjected to a visual feature regularization optimization based on a boundary surface-volume space to obtain a set of pixel-level airport runway surface visual feature vectors within a local receptive field.

[0106] In other words, by constructing a commutation constraint between the volume space representation vector and the boundary representation vector, the two are forced to satisfy conformal conditions in the mathematical space. For example, when detecting longitudinal cracks on a runway, the linear boundary features of the cracks must maintain geometric consistency with the crack depth (volume space features) in the vector expression; and for the local threshing area, the voxel distribution of its surface roughness must form an associative mapping with the topological structure of the threshing boundary. Specifically, by dynamically modulating the weighting coefficient, the difference between the two is converged within the regularization threshold range, eliminating pseudo-boundary interference caused by light reflection or texture similarity. This optimization not only strengthens the geometric integrity expression of the diseased area, but also significantly improves the discriminability of key features such as crack direction and threshing range by constraining the spatial consistency of the feature vectors within the local receptive field. The optimized set of pixel-level visual feature vectors of the airport runway surface within the initial local receptive field can support subsequent semantic segmentation to accurately extract the sub-pixel contours of defects, enabling the subsequent saliency enhancement process to accurately distinguish real defects from texture interference based on conformal features, providing spatially consistent defect semantic information for the BIM model, supporting precise spatial positioning of pavement maintenance, and significantly improving the robustness of defect recognition in complex pavement environments.

[0107] In some embodiments, for the pixel-level airport runway surface visual feature vector v to be enhanced and distilled i,j The corresponding set v of pixel-level airport runway surface visual feature vectors within the local receptive field m,n For example, in order to enhance the amplification effect of the characteristic components related to saliency and the suppression / weakening effect of the characteristic components related to non-saliency, it is expected that the set of pixel-level airport runway surface visual feature vectors within the local receptive field v m,nIt can have conformal representation, that is, it is expected that there can be a high correspondence between its volume space representation and boundary representation.

[0108] First, determine the volume space representation vector as:

[0109]

[0110] The boundary representation vector is:

[0111]

[0112] Then, by modulating the weighting coefficients α and β, the boundary surface-volume space tensor has the regularity that satisfies the commutation relation, that is, the volume space representation vector v m,n (3) and the boundary representation vector v m,n (2) The spatial two-norm representation of the difference vector tends to the product of coefficients α and β:

[0113] ||v m,n (3) -v m,n (2) ||2=ω×α×β

[0114] Among them, θ(v m,n ) is v m,n is a significant enhancement weight factor, and ω is a proportional scaling factor.

[0115] That is, through the above formula, α and β are modulated and constrained. In this way, under the condition of appropriately selecting the boundary surface condition and utilizing the conformal commutativity of the volume space, the regularity standard is met, thereby realizing the visual feature vector v of the airport runway surface at the pixel level to be enhanced and distilled within the local receptive field. m,n The highly conformal representation fusion improves the context-aware feature saliency expression of the enhanced pixel-level airport runway surface visual feature vector.

[0116] Subsequently, based on the set of pixel-level airport runway surface visual feature vectors within the local receptive field, the distilled pixel-level airport runway surface visual feature vectors are saliency enhanced to obtain the enhanced pixel-level airport runway surface visual feature vectors and used as the visual salient encoding features of the airport runway surface.

[0117] Considering the deep coupling problem between the features of the disease and the background noise in a complex pavement environment, for example, when the crack edge gradient, the threshing area texture, the pavement markings, and the light reflection are intertwined in the feature space in the high-resolution images collected by drones, conventional feature enhancement methods are difficult to effectively remove the interference signals. Therefore, in order to establish a spatially adaptive feature enhancement mechanism so that the semantic features of the diseased area can break through the masking of the background noise with the support of the local context, in the technical solution applied for by the present invention, based on the set of pixel-level airport runway surface visual feature vectors within the local receptive field, the enhanced pixel-level airport runway surface visual feature vector to be distilled is significantly enhanced to obtain an enhanced pixel-level airport runway surface visual feature vector, wherein the enhanced pixel-level airport runway surface visual feature vector is the channel feature vector of the (i, j)th pixel position of the airport runway surface visual feature saliency coding map.

[0118] In this process, the boundary and volumetric representations of the pixel-level runway surface visual feature vectors within the local receptive field are weighted and aggregated, and regularization constraints are used to balance internal structural consistency with boundary abruptness. For example, in oil spill areas, the volumetric continuity of the color distribution is enhanced while the interference response of the blurred boundary is weakened. This suppresses the regular geometric features at the marking seams due to their low semantic saliency, while the irregular morphological features of the actual defects are multi-dimensionally enhanced through contextual association. The enhanced pixel-level runway surface visual feature vectors form spatially specific activation patterns. Key defect information, such as the continuous fracture characteristics of cracks and the discrete patch characteristics of threshing, is highlighted in the coded image, providing highly discriminative input data for the semantic segmentation module. This pixel-level enhancement mechanism ultimately supports the accurate component-level defect mapping of BIM models, enabling pavement maintenance plans to be formulated based on the defect distribution characteristics in three-dimensional space, significantly improving the scientific and timely nature of maintenance decisions in complex pavement environments.

[0119] In some embodiments, based on a set of pixel-level airport runway surface visual feature vectors within a local receptive field, the enhanced pixel-level airport runway surface visual feature vectors are significantly enhanced using the following saliency enhancement formula to obtain enhanced pixel-level airport runway surface visual feature vectors; wherein the saliency enhancement formula is:

[0120]

[0121] Among them, γ and μ are trainable weighted hyperparameters, θ(v m,n ) is v m,n The significant enhancement weight factor, v sw is the scoring weight vector, obtained by computational derivation or training, softmax(·) is the softmax function, is vector multiplication, v' i,jTo enhance the pixel-level airport runway surface visual feature vector, v' i,j v i,j The corresponding channel feature vector of the (i, j)th pixel position in the airport runway surface visual feature saliency encoding image.

[0122] In some embodiments, the calculated enhanced pixel-level airport runway surface visual feature vector may be used as a visually significant coding feature of the airport runway surface.

[0123] In some embodiments, in step S32, pavement damage information is determined based on visually significant coding features of the airport runway surface. For example, step S32 includes:

[0124] Firstly, image semantic segmentation is performed on the salient coding map of the airport runway surface visual features to obtain the semantic segmentation result of the airport runway surface state.

[0125] In the present application, it should be understood that although the visual feature coding map of the airport runway surface after saliency enhancement has highlighted the characteristic response of the diseased area, it is still necessary to convert the pixel-level feature activation pattern into a structured output with a clear semantic label. Therefore, in order to break through the dependence of the traditional threshold segmentation method on morphological rules and solve the problem of mis-segmentation caused by the similarity of edge features between pavement markings and cracks, in the technical solution of the present application, the airport runway surface visual feature saliency coding map is subjected to image semantic segmentation to obtain the airport runway surface state semantic segmentation result. In this process, the algorithm establishes a spatial association model based on the topological structural characteristics of the disease features. In a specific example of the present application, a directional continuity constraint is constructed for linear cracks to ensure pixel-level coherent labeling of the fracture area; and a regional growing strategy is used to capture the gradient boundary of the oil pollution diffusion area. In this way, the regular geometric features of the pavement markings can be automatically filtered due to the lack of semantic attributes of the disease, while the irregular morphology of the real disease is accurately labeled based on the saliency of the feature. The generated semantic segmentation results for the airport runway surface condition not only include basic information such as damage type and contour location, but also, by integrating the spatial coordinates of BIM components, enable automated measurement of geometric parameters such as crack length and threshing area, providing a quantitative basis for maintenance priority assessment. This structured output directly supports the spatial mapping of damage information with the BIM model, allowing for the visual tracking of cumulative damage associated with facility components such as pavement joints and lighting strips, significantly improving the spatiotemporal accuracy of preventative maintenance decisions.

[0126] Furthermore, based on the semantic segmentation results of the airport runway surface state, pavement damage information is determined, which includes the type of surface damage, the location of the surface damage, the geometric parameters of the surface damage, and the preliminary severity of the surface damage.

[0127] In the present application, it should be understood that although the semantic segmentation network can perform pixel-level annotation of pavement cracks, threshing and other diseased areas, simple area labeling cannot meet the BIM model's spatial mapping requirements for component-level damage. Therefore, in order to break through the limitation of traditional detection results that only provide two-dimensional plane information, a disease data model that conforms to engineering semantics is established, so that features such as the oil diffusion range and crack extension path can be spatially associated with BIM components (such as pavement panel joints and light strip bases). In the technical solution of the present application, by constructing a multi-dimensional feature parsing algorithm, the connected domain pixel clusters in the segmentation results are converted into composite data entities containing type, coordinates, size and degree of damage.

[0128] In one example of this invention application, for crack damage, the algorithm identifies its orientation type (e.g., horizontal, vertical, or mesh) based on the morphology of pixel clusters and calculates its length and average width using skeleton extraction techniques. For threshing areas, edge contour analysis is used to determine the patch area and depth gradient distribution. Structured damage information not only supports the dynamic updating of BIM models but also enables tracking of pavement damage evolution through spatiotemporal data analysis. This can guide the development of preventive maintenance strategies and significantly enhance scientific decision-making capabilities for runway lifecycle management.

[0129] In some embodiments, in step S4, the dataset of pavement damage information is geospatially aligned, and the pavement damage information is mapped to corresponding component elements in the airport runway BIM model using coordinate matching to obtain an updated airport runway BIM model.

[0130] In the present application, it should be understood that although the disease recognition algorithm can accurately locate damaged areas such as pavement cracks and threshing, the heterogeneity between the two-dimensional image coordinate system and the three-dimensional engineering coordinate system of the BIM model makes it difficult to associate the disease information with specific facility components such as road lighting strips. Therefore, in order to break through the information silos of traditional inspection reports and facility management systems, and enable disease information such as crack expansion paths and threshing distribution areas to be attached to BIM component instances in the form of three-dimensional entity attributes, forming a full-element association system of pavement status-facility equipment-spatial position, in the technical solution of the present application, the dataset of pavement disease information is geospatially aligned, and the pavement disease information is mapped to the corresponding component elements in the airport runway BIM model using coordinate matching to obtain an updated airport runway BIM model.

[0131] Specifically, the dataset of pavement damage information can be geospatial registered through the following steps, and the pavement damage information can be mapped to the corresponding component elements in the airport runway BIM model using coordinate matching: the location of surface damage is extracted from the pavement damage information, and the location of the surface damage is the pixel coordinates of the center point of the damage; based on the location of the surface damage, the actual geographic coordinates of the surface disaster are calculated; the actual geographic coordinates of the surface disaster are spatially queried in the BIM database to obtain the identifier of the component element; based on the identifier of the construction element, the pavement damage information is mapped to the corresponding component element in the airport runway BIM model to obtain an updated airport runway BIM model.

[0132] In this invention application, the updated BIM model can present key operational and maintenance indicators such as the cumulative damage value of specific runway panels and the development trend of cracks at joints in real time. This supports preventive maintenance decisions based on spatial topology. For example, this model can optimize blanket construction plans based on the clustering effect of threshing damage around light strips, or predict the risk of subgrade disease based on the extension trajectory of cracks at pavement joints. In this way, the health of the runway can be visually tracked along the time dimension, significantly improving the accuracy and foresight of infrastructure operations and maintenance in complex airport environments.

[0133] In summary, the BIM-based intelligent management method for airport runway pavement applied in accordance with the present invention is clarified, and the surface condition images of the airport runway collected by the drone are used for pavement disease detection by using pixel-level recognition technology based on saliency enhancement, capturing various types of diseases including tiny and occurring diseases, and effectively overcoming complex background interference; at the same time, the detailed information of the disease is committed to being accurately located and associated with the specific components in the airport runway BIM model to construct an updated digital information model of the runway health status, which is conducive to the intelligent management of the airport runway pavement.

[0134] BIM-based airport runway pavement intelligent management system

[0135] like Figure 5 As shown, in a second aspect, a BIM-based airport runway pavement intelligent management system 300 uses any of the BIM-based airport runway pavement intelligent management methods described in the first aspect, including:

[0136] The drone control module 310 controls the drone to fly along a preset path plan;

[0137] a surface state data acquisition module 320 that uses an onboard camera of a drone to capture a surface state image of the runway and annotates the runway location to obtain a data set of runway surface state data, wherein the runway surface state data includes the runway surface state image and location coordinates;

[0138] a pavement damage identification module 330 for performing pavement damage identification on each airport runway surface condition data in the airport runway surface condition data set to obtain a data set of pavement damage information;

[0139] The BIM model updating module 340 performs geospatial registration on the data set of the pavement defect information and uses coordinate matching to map the pavement defect information to corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model.

[0140] It should be understood that the BIM-based airport runway pavement intelligent management system applied for in the present invention is used to implement any of the BIM-based airport runway pavement intelligent management methods in the first aspect, and accordingly may also include: all the technical problems, technical solutions and technical effects recorded in any of the BIM-based airport runway pavement intelligent management methods in the first aspect, and the present invention application will not be repeated here.

[0141] In some embodiments, the BIM-based intelligent airport runway pavement management system 300 according to the present invention can be implemented in various wireless terminals, such as a server with a BIM-based intelligent airport runway pavement management algorithm. In one possible implementation, the BIM-based intelligent airport runway pavement management system 300 according to the present invention can be integrated into a wireless terminal as a software module and / or hardware module. For example, the BIM-based intelligent airport runway pavement management system 300 can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the BIM-based intelligent airport runway pavement management system 300 can also be one of the many hardware modules of the wireless terminal.

[0142] In other embodiments, alternatively, the BIM-based airport runway pavement intelligent management system 300 and the wireless terminal may be separate devices, and the BIM-based airport runway pavement intelligent management system 300 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0143] BIM-based intelligent management and processing device for airport runway pavement

[0144] In the third aspect, the present invention application provides a BIM-based airport runway pavement intelligent management and processing device, including a memory and a processor that are communicatively connected, the memory being used to store a computer program, the processor being used to read the computer program, and executing the BIM-based airport runway pavement intelligent management method as described in any one of the first aspects.

[0145] Those skilled in the art will appreciate that the BIM-based airport runway pavement intelligent management and processing device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any of the steps in the BIM-based airport runway pavement intelligent management method described in the first aspect.

[0146] Computer-readable storage medium

[0147] In a fourth aspect, the present invention application provides a computer-readable storage medium having instructions stored thereon. When the instructions are run on a computer, the BIM-based airport runway pavement intelligent management method as described in any one of the first aspects is executed.

[0148] Those skilled in the art will appreciate that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0149] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function (such as a sound playback function, an image playback function, etc.); the data storage area can store data created based on the use of the mobile phone (such as audio data, a phone book, etc.). In addition, the memory can include a high-speed random access memory and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0150] Wherein, if the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0151] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0152] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A BIM-based intelligent management method for airport runway pavement, characterized in that: include: Control the drone to fly along the preset path; The surface state image of the airport runway is collected by using an onboard camera of a drone and the position of the airport runway is marked to obtain a data set of the airport runway surface state data, wherein the airport runway surface state data includes the airport runway surface state image and position coordinates; Performing pavement disease identification on each airport runway surface state data in the airport runway surface state data dataset to obtain a pavement disease information dataset, including: performing feature pixel-level saliency enhancement on each airport runway surface state data based on the runway surface state information to obtain visually significant coding features of the airport runway surface; Determine pavement damage information based on visually significant coding features of the airport runway surface; The dataset of pavement damage information is geospatially registered, and the pavement damage information is mapped to the corresponding component elements in the airport runway BIM model using coordinate matching to obtain an updated airport runway BIM model.

2. The BIM-based intelligent management method for airport runway pavement according to claim 1 is characterized in that: The runway surface state data of each airport are subjected to feature pixel-level saliency enhancement based on the runway surface state information to obtain the visually salient coding features of the airport runway surface, including: Extracting runway surface visual features from the airport runway surface state image to obtain airport runway surface visual coding features; The airport runway surface visual coding features are saliently processed to obtain the airport runway surface visual salient coding features.

3. The BIM-based intelligent management method for airport runway pavement according to claim 2 is characterized in that: Extracting the runway surface visual features from the airport runway surface state image to obtain the airport runway surface visual coding features, including: The airport runway surface state image is passed through a runway surface visual feature extractor based on a void convolutional neural network model to obtain an airport runway surface visual feature coding map as the airport runway surface visual coding feature.

4. The BIM-based intelligent management method for airport runway pavement according to claim 3 is characterized in that: Performing runway surface visual feature saliency on the airport runway surface visual coding features to obtain the airport runway surface visual salient coding features, including: Extracting pixel-level airport runway surface visual features from the airport runway surface visual feature coding map to obtain a pixel-level airport runway surface visual feature vector to be enhanced; Perform feature distillation on the pixel-level airport runway surface visual feature vector to be enhanced to obtain the pixel-level airport runway surface visual feature vector to be enhanced; Based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level distilled airport runway surface visual feature vector to be enhanced, the surface visual feature saliency enhancement is performed on the pixel-level distilled airport runway surface visual feature vector to be enhanced to obtain an enhanced pixel-level airport runway surface visual feature vector and use it as the airport runway surface visual saliency coding feature, wherein the enhanced pixel-level airport runway surface visual feature vector is the channel feature vector of the (i, j)th pixel position of the airport runway surface visual feature saliency coding map.

5. The BIM-based intelligent management method for airport runway pavement according to claim 4 is characterized in that: Extracting pixel-level airport runway surface visual features from the airport runway surface visual feature coding map to obtain a pixel-level airport runway surface visual feature vector to be enhanced, including: The visual feature encoding map of the airport runway surface is decoupled along the channel dimension to obtain a set of pixel-level airport runway surface visual feature vectors. The pixel-level airport runway surface visual feature vector at the (i, j)th pixel position is extracted from the set of pixel-level airport runway surface visual feature vectors as the pixel-level airport runway surface visual feature vector to be enhanced.

6. The BIM-based intelligent management method for airport runway pavement according to claim 5 is characterized in that: Based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced, the surface visual feature significance enhancement is performed on the pixel-level airport runway surface visual feature vector to be enhanced to obtain an enhanced pixel-level airport runway surface visual feature vector, including: Based on the spatial structure characteristics of the surface visual feature distribution of the pixel-level airport runway surface visual feature vector to be enhanced and distilled, the size of the feature receptive field of the pixel-level airport runway surface visual feature vector to be enhanced and distilled is determined; Based on the size of the feature receptive field, the set of pixel-level airport runway surface visual feature vectors within the local receptive field is filtered out from the set of pixel-level airport runway surface visual feature vectors; Based on the set of pixel-level airport runway surface visual feature vectors within the local receptive field, the distilled pixel-level airport runway surface visual feature vector is significantly enhanced to obtain the enhanced pixel-level airport runway surface visual feature vector.

7. The BIM-based intelligent management method for airport runway pavement according to claim 6 is characterized in that: Based on the size of the feature receptive field, the set of pixel-level airport runway surface visual feature vectors within the local receptive field is screened from the set of pixel-level airport runway surface visual feature vectors, including: Based on the size of the feature receptive field, a set of pixel-level airport runway surface visual feature vectors within a preliminary local receptive field is screened from the set of pixel-level airport runway surface visual feature vectors; The set of pixel-level airport runway surface visual feature vectors within the preliminary local receptive field is subjected to regularized optimization of visual features based on boundary surface-volume space to obtain the set of pixel-level airport runway surface visual feature vectors within the local receptive field.

8. The BIM-based intelligent management method for airport runway pavement according to claim 7 is characterized in that: Determining pavement damage information based on visually significant coding features of the runway surface, including: performing semantic segmentation on a map of visually significant coding features of the runway surface to obtain semantic segmentation results of the runway surface state; determining pavement damage information based on the semantic segmentation results of the runway surface state, the pavement damage information including the type of surface damage, the location of the surface damage, the geometric parameters of the surface damage, and the preliminary severity of the surface damage; And / or, geospatial registration is performed on the dataset of pavement damage information, and coordinate matching is used to map the pavement damage information to corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model, including: extracting the location of surface damage from the pavement damage information, where the location of the surface damage is the pixel coordinates of the center point of the damage; based on the location of the surface damage, calculating the actual geographic coordinates of the surface disaster; performing a spatial query on the actual geographic coordinates of the surface disaster in the BIM database to obtain an identifier of the component element; and based on the identifier of the construction element, mapping the pavement damage information to the corresponding component element in the airport runway BIM model to obtain an updated airport runway BIM model.

9. A BIM-based airport runway pavement intelligent management system, using the BIM-based airport runway pavement intelligent management method according to any one of claims 1 to 8, characterized in that: include: The drone control module controls the drone to fly along the preset path; A surface state data acquisition module collects surface state images of the airport runway through the drone's onboard camera and marks the location of the airport runway to obtain a dataset of airport runway surface state data. The airport runway surface state data includes the airport runway surface state image and location coordinates. A pavement defect recognition module identifies pavement defects on each runway surface condition data set to obtain a pavement defect information data set. This module includes: performing pixel-level saliency enhancement on each runway surface condition data set based on the runway surface condition information to obtain visually significant coding features of the runway surface; and determining pavement defect information based on the visually significant coding features of the runway surface. The BIM model update module performs geospatial registration on the pavement defect information dataset and uses coordinate matching to map the pavement defect information to the corresponding component elements in the airport runway BIM model to obtain an updated airport runway BIM model.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores instructions, and when the instructions are executed on the computer, the BIM-based airport runway pavement intelligent management method described in any one of claims 1 to 8 is executed.

Images