A fine partition evaluation method for hydraulic structure components based on intelligent identification of sea-land-air three domains

Abstract

This invention provides a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains. It relates to the field of intelligent construction and operation and maintenance technology, and includes: collecting full-domain point cloud data and structural defect data of the hydraulic structure; segmenting independent components from the full-domain point cloud data based on geometric features to generate component point cloud subsets; constructing a target BIM model and integrating it into a GIS platform to obtain the component's location information in macro-geographic space; comprehensively evaluating the attribute information of each component based on evaluation dimensions to determine the control level of each component; dividing corresponding control zones based on the control level and location information of each component, and associating the unique identifier of each component with its respective control zone; generating a visual zoning map based on the control zones, and generating a component information list based on the target BIM model and control level, as the assessment result. This significantly improves the accuracy, comprehensiveness, and management efficiency of the assessment.

Description

Technical Field

[0001] This invention relates to the field of intelligent construction and operation and maintenance technology, and in particular to a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea and air domains. Background Technology

[0002] Currently, hydraulic structures, such as port breakwaters, revetments, and wharves, are prone to material aging, structural damage, and functional degradation due to factors such as water erosion, corrosion, load changes, and natural disasters during long-term service. Regular and accurate assessments are essential for ensuring project safety, developing scientific maintenance and renovation plans, and implementing green demolition. However, the assessment of hydraulic structures relies primarily on traditional manual inspections, single-technology testing, and experience-based judgment, which has significant limitations: It is difficult to acquire high-precision, synchronized data across land, water, and air, especially in areas where land and water meet; existing assessments are typically conducted on a segment or large area basis, lacking accurate identification and independent documentation of individual components (such as single wave-breaking blocks or independent piles); the assessment process is highly dependent on the personal experience of technical personnel, lacking a systematic and quantitative comprehensive evaluation system. It is difficult to objectively allocate weights and integrate analyses for the multi-dimensional attributes of components, such as safety status, reuse value, construction risks, and environmental impact, leading to poor consistency in assessment conclusions and difficulty in supporting refined hierarchical management decisions; collected geometric data, defect information, design documents, and macro-geographical information are usually stored in a scattered manner, failing to be effectively integrated into a unified spatial information platform (such as GIS) and information model (such as BIM). This makes it difficult to link and display assessment results between macro-geographic space and micro-component models, failing to provide intuitive and visual decision support for demolition timing planning, equipment selection, and the formulation of safety and environmental protection measures.

[0003] Therefore, in order to overcome the above-mentioned technical problems, the present invention provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea and air domains. Summary of the Invention

[0004] This invention provides a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains. It utilizes collaborative operations involving aerial UAV laser scanning, land-based 3D laser scanning, and underwater ROV robots to achieve intelligent acquisition of high-precision geometric information and structural defect data across all land, air, and water domains for hydraulic structures. Based on the acquired point cloud data, a component identification and segmentation mechanism is constructed, integrating GIS macro-spatial positioning capabilities with the advantages of BIM component-level information management. A linked mapping of component number, spatial coordinates, and attribute information is established to achieve precise positioning and traceable management of hydraulic components. Through a multi-dimensional intelligent scoring model, automated zoning and hierarchical assessments are conducted on hydraulic components and their respective areas, resulting in clearly defined hierarchical and boundary-based control zoning outcomes. This significantly improves the accuracy, comprehensiveness, and management efficiency of the assessment.

[0005] A refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains includes: Step 1: Collect full-domain point cloud data and structural defect data of hydraulic structures based on the three-domain collaborative acquisition mechanism of sea, land and air; Step 2: Identify and segment independent components in the global point cloud data based on geometric features, and generate component point cloud subsets; Step 3: Based on the component point cloud subset and structural defect data, construct a target BIM model with unique identifiers and attribute information. At the same time, integrate the target BIM model into the GIS platform to obtain the component's location information in macro-geographic space. Step 4: Based on the evaluation dimensions, conduct a multi-dimensional comprehensive evaluation of the attribute information of each component in the target BIM model to determine the control level of each component; Step 5: Based on the control level of each component and the component's location information in the macro-geographic space, divide the space into corresponding control zones and associate the component's unique identifier with its respective control zone; Step 6: Generate a visual zoning map based on the control zones, and generate a list of component information based on the target BIM model and control level as the evaluation result.

[0006] Preferably, in a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains, step 1, the land, sea, and air domain collaborative acquisition mechanism includes: Point cloud data and structural defect data of the superstructure of the hydraulic structure were obtained by aerial drone laser scanning. Point cloud data and structural defect data of the structure above the waterline were obtained based on 3D laser scanning of the land area; Point cloud data and structural defect data of underwater structures were obtained by scanning with an underwater ROV robot. By using terrestrial 3D laser and ROV robot to perform overlapping scans of the land-water interface area, point cloud data and structural defect data of the land-water interface area are obtained.

[0007] Preferably, a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains includes: The airspace UAV laser scanning includes flight path planning, setting flight altitude and speed, and simultaneously acquiring laser point clouds and high-resolution images; The land area three-dimensional laser scanning includes setting up multiple scanning stations and stitching point clouds using a target sphere; The underwater ROV robot scanning includes collecting data by approaching the surface of the component in a constant-altitude cruising mode during low tide, and activating sonar and ultrasonic flaw detectors. The airspace, land, and underwater point cloud data are registered and fused using a multi-source data fusion algorithm, specifically including: A coarse registration algorithm based on FPFH features and RANSAC, and a fine registration algorithm based on ICP, are used to unify point cloud data from different domains into the same geographic coordinate system. A fusion algorithm based on Gaussian pyramid downsampling and Kriging interpolation is used to weightedly fuse point cloud data of different resolutions to generate a global point cloud dataset with uniform resolution.

[0008] Preferably, in a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains, step 2, before identifying and segmenting individual components in the full-domain point cloud data based on geometric features, includes: The global point cloud data is preprocessed by denoising and multi-site point cloud stitching, and a target point cloud set is generated based on the preprocessing results; Specifically, based on the target point cloud set, components with independent geometric shapes are identified and segmented according to at least one of geometric feature matching, region growing, or deep learning segmentation algorithms, generating a component point cloud subset.

[0009] Preferably, in a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains, step 3 involves constructing a target BIM model with unique identifiers and attribute information based on a subset of the component's point cloud and structural defect data. Simultaneously, the target BIM model is integrated into a GIS platform to obtain the component's location information in macro-geographic space, including: Generate 3D geometric models of each component in BIM modeling software based on component point cloud subsets; A unique ID number is added to the 3D geometric model of each generated component. The encoding rules for the ID number include: facility type, mileage location, component type, and sequence number information. Structural defect data is used as defect attributes, and the design attributes and material attributes of the pre-collected data are jointly associated with the ID number of the corresponding component to generate complete attribute information of the component; The target BIM model, which integrates the three-dimensional geometric model of the components, unique ID number, and complete attribute information, is imported into the GIS platform; The imported target BIM model is calibrated in the GIS platform based on the coordinate control points obtained from on-site measurements; The target BIM model, after coordinate calibration, is spatially overlaid with the base map and corresponding thematic layers in the GIS platform to obtain the latitude, longitude, and elevation positioning information of each component in the macro-geographic space. The acquisition of the structural defect data includes a structural defect information complementarity fusion algorithm, specifically: U-Net semantic segmentation algorithm is used to identify apparent defects in airspace images, and waveform feature analysis algorithm is used to identify internal defects in underwater ultrasonic data. Based on the registered point cloud coordinates, the apparent defects and internal defects are mapped to the three-dimensional space of the component. The DS evidence theory is used to fuse multi-source defect data to generate a comprehensive defect assessment result of the component as the defect attribute.

[0010] Preferably, in a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains, step 4 involves a multi-dimensional comprehensive evaluation of the attribute information of each component in the target BIM model based on evaluation dimensions to determine the control level of each component, including: The evaluation dimensions are multiple, including: structural functional importance, material reuse value, structural health status, demolition and construction risks, and the degree of impact on the water environment. The target score for each evaluation dimension is calculated based on the attribute information of each component in the target BIM model. Obtain the dimensional weights of each evaluation dimension, and calculate the weighted sum based on the target score of each evaluation dimension and the corresponding dimensional weight to obtain the comprehensive score; The control level classification thresholds are obtained separately, including: a first threshold and a second threshold, wherein the first threshold is greater than the second threshold; When the overall score is equal to or greater than the first threshold, the component is classified as the first control level. When the overall score is less than the first threshold but greater than the second threshold, the component will be classified as the second control level. When the overall score is less than or equal to the second threshold, the component will be classified as the third control level.

[0011] Preferably, a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains includes step 5: based on the control level of each component and its location information in macro-geographic space, dividing the space into corresponding control zones, and associating the unique identifier of each component with its respective control zone, including: Based on the control level of each component, components with the same control level and continuously distributed in space are classified to obtain the component set under each control level. Obtain the macro-geospatial information location of the components, and determine the spatial distribution boundaries of various components based on the macro-geospatial information location of the components; Based on the spatial distribution boundaries, physical areas corresponding to the control levels are divided in the macro-geographic space, and control zones are generated based on the division results; A unique partition number is added within the controlled partition, wherein the partition number code includes the corresponding control level information and the corresponding spatial range information; Associate and bind the unique identifier of all components located within the same control zone with the corresponding control zone number.

[0012] Preferably, in a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains, step 6 involves generating a visual zoning map based on the control zoning, and generating a component information list based on the target BIM model and control level, which serves as the assessment result, including: The control zones and their corresponding control levels are rendered on the base map of the GIS platform, based on different legends, to generate a GIS macro-level map. In the target BIM model, each component is rendered according to its control level to generate a BIM micro-component classification model that can display the classification results of the component level. The key information of the GIS macro-level map and the BIM micro-component level model is processed according to a preset scale to output a paper-based level map on site. The attribute information of each component is extracted based on the target BIM model, and a component information list containing the unique identifier, attribute information, control level and spatial coordinates is generated based on the control level and location information. Based on the control zones and control levels, and according to the component information list, a zone control requirements manual is constructed for each zone, including the dismantling sequence, equipment selection, safety protection, and environmental protection measures.

[0013] Preferably, a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains includes preprocessing the full-domain point cloud data by denoising and stitching together multi-site point cloud data, comprising: Based on point cloud processing software, outlier noise points are identified and removed from the global point cloud data.

[0014] Based on the target spheres deployed at the scanning site or the recorded GPS coordinates, spatial transformation parameters between point cloud data obtained from different scanning stations are calculated.

[0015] Based on the spatial transformation parameters, the point cloud data from multiple stations are aligned and merged into a unified point cloud dataset in a single coordinate system.

[0016] The entire point cloud dataset is used as the target point cloud set for recognition and segmentation operations.

[0017] Preferably, a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains further includes the following before step 1: Based on the results of the on-site survey, and using an AI recognition algorithm based on a CNN and SVM fusion model to analyze historical data, on-site images and terrain data, the scanning boundary of the hydraulic structure is determined, and dangerous areas and safe working areas are intelligently marked based on the scanning boundary to generate a risk heat map of dangerous areas. Record the hydrological conditions of the safe operating area and determine the optimal scanning window based on tidal cycle and wave height data; Collect the original design data, inspection, maintenance and operation data, and port area geographical data including coordinate information of the hydraulic structures; Based on the scanning boundaries, danger zone markers, and geographic data, the operation paths of UAVs and underwater ROVs are planned using reinforcement learning algorithms, and the deployment scheme of land-based three-dimensional laser scanning stations is optimized using genetic algorithms. The equipment deployment and operation paths of the land-sea-air three-domain collaborative acquisition mechanism are determined. A multi-agent collaborative decision-making mechanism for dynamically adjusting the operation plan based on real-time environmental data is also constructed.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: By employing aerial UAV laser scanning, land-based 3D laser scanning, and underwater ROV robot collaborative operations, high-precision geometric information and structural defect data of hydraulic structures across land, air, and water are intelligently collected. Based on the collected point cloud data, a component identification and segmentation mechanism is constructed, integrating the macro-spatial positioning capabilities of GIS with the advantages of BIM component-level information management. A linkage mapping between component number, spatial coordinates, and attribute information is established to achieve precise positioning and traceable management of hydraulic components. Through a multi-dimensional intelligent scoring model, automated zoning and hierarchical assessments are conducted on hydraulic components and their respective areas, resulting in clearly defined hierarchical and boundary-based control zoning outcomes. This significantly improves the accuracy, comprehensiveness, and management efficiency of the assessments.

[0019] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea, and air domains, as described in an embodiment of the present invention. Figure 2 This is a schematic diagram of a drone scanner in a method for refined zoning evaluation of hydraulic structure components based on intelligent identification of land, sea, and air domains, as described in an embodiment of the present invention. Figure 3 This is a schematic diagram of a ground-based three-dimensional laser scanner used in a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea, and air domains, as described in an embodiment of the present invention. Figure 4 This is a schematic diagram of an underwater ROV scanning robot in a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea, and air domains, as described in an embodiment of the present invention. Figure 5 This is a simulation mapping diagram of the refined zoning evaluation of hydraulic structure components in an embodiment of the present invention, based on intelligent identification of hydraulic structure components in three domains: sea, land, and air. Figure 6 This is a technical roadmap for a refined zoning evaluation method for hydraulic structure components based on intelligent identification of land, sea, and air domains, as described in an embodiment of the present invention. Detailed Implementation

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] Example 1: This example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains, such as... Figure 1 As shown, it includes: Step 1: Collect full-domain point cloud data and structural defect data of hydraulic structures based on the three-domain collaborative acquisition mechanism of sea, land and air; Step 2: Identify and segment independent components in the global point cloud data based on geometric features, and generate component point cloud subsets; Step 3: Based on the component point cloud subset and structural defect data, construct a target BIM model with unique identifiers and attribute information. At the same time, integrate the target BIM model into the GIS platform to obtain the component's location information in macro-geographic space. Step 4: Based on the evaluation dimensions, conduct a multi-dimensional comprehensive evaluation of the attribute information of each component in the target BIM model to determine the control level of each component; Step 5: Based on the control level of each component and the component's location information in the macro-geographic space, divide the space into corresponding control zones and associate the component's unique identifier with its respective control zone; Step 6: Generate a visual zoning map based on the control zones, and generate a list of component information based on the target BIM model and control level as the evaluation result.

[0024] In this embodiment, the invention provides a classification and grading assessment method applicable to the demolition or renovation of existing hydraulic structures such as port breakwaters and revetments. This method utilizes aerial UAV laser scanning, land-based 3D laser scanning, and underwater ROV robot collaborative operations to achieve intelligent collection of high-precision geometric information and structural defect data across the land, air, and water areas of hydraulic structures. Based on this, a point cloud data-driven component identification and segmentation mechanism is constructed to obtain uniquely identified component-level digital information. Furthermore, the macro-spatial positioning capabilities of GIS and the component-level information management advantages of BIM are integrated to establish a linkage mapping relationship between component numbers, spatial coordinates, and attribute information, achieving precise positioning and traceable management of hydraulic components. By introducing an intelligent scoring model based on multi-dimensional indicators such as structural function, material value, health status, construction risks, and environmental impact, the components and their respective areas are automatically categorized and graded for assessment, resulting in clearly defined and clearly demarcated control zoning results. Finally, a visualized grading map and component list are output to guide green demolition, risk management, and high-value utilization of resources. The refined zoning assessment simulation mapping map of hydraulic structure components is shown below. Figure 5 As shown, the technology roadmap is as follows: Figure 6 As shown.

[0025] In this embodiment, the three-domain collaborative acquisition mechanism refers to a systematic method for collaboratively collecting point cloud and defect data of hydraulic structures from three dimensions: ocean, land, and air.

[0026] In this embodiment, global point cloud data refers to a three-dimensional point cloud dataset covering the entire area of ​​the hydraulic structure, obtained through collaborative acquisition.

[0027] In this embodiment, structural defect data refers to data that records information on defects such as damage and cracks in hydraulic structure components.

[0028] In this embodiment, the component point cloud subset refers to the subset of point cloud data corresponding to a single component that is identified and segmented from the global point cloud data based on geometric features.

[0029] In this embodiment, the target BIM model refers to the hydraulic structure information model that integrates the unique identifiers of components, attribute information, and defect data.

[0030] In this embodiment, the control level refers to the maintenance and management priority level determined for each component through multi-dimensional comprehensive evaluation.

[0031] In this embodiment, the control zone refers to different management areas divided in the macro-geographic space according to the control level of the components.

[0032] In this embodiment, the visual zoning map refers to the visualization result that displays the information of the control zone and its corresponding components in graphical form.

[0033] In this embodiment, the component information list refers to a detailed list that summarizes the attributes, control levels, and other information of each component, as part of the evaluation results.

[0034] The beneficial effects of the above technical solution are as follows: It utilizes aerial UAV laser scanning, land-based 3D laser scanning, and underwater ROV robot collaborative operations to achieve intelligent collection of high-precision geometric information and structural defect data across the land, air, and water domains of hydraulic structures; it constructs a component identification and segmentation mechanism based on the collected point cloud data, integrating GIS macro-spatial positioning capabilities with the advantages of BIM component-level information management, establishing a linked mapping of component numbers, spatial coordinates, and attribute information to achieve precise positioning and traceable management of hydraulic components; through a multi-dimensional intelligent scoring model, it conducts automated zoning and hierarchical assessments of hydraulic components and their respective areas, resulting in clearly defined hierarchical and boundary-based control zoning outcomes; and it significantly improves the accuracy, comprehensiveness, and management efficiency of the assessment.

[0035] Example 2: Based on Example 1, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across three domains (sea, land, and air). Step 1, the three-domain collaborative data acquisition mechanism, includes: Point cloud data and structural defect data of the superstructure of the hydraulic structure were obtained by aerial drone laser scanning. Point cloud data and structural defect data of the structure above the waterline were obtained based on 3D laser scanning of the land area; Point cloud data and structural defect data of underwater structures were obtained by scanning with an underwater ROV robot. By using terrestrial 3D laser and ROV robot to perform overlapping scans of the land-water interface area, point cloud data and structural defect data of the land-water interface area are obtained.

[0036] In this embodiment, the three-domain collaborative data acquisition mechanism includes three methods: airborne UAV scanning, land-based 3D laser scanning, and underwater ROV robot scanning. In the airborne domain, a multi-rotor UAV equipped with a lidar and visible light sensor (such as the DJI Matrice 300 RTK) is used to acquire 3D spatial information and surface features of the breakwater and its surrounding structure. In the land-based domain, a terrestrial 3D laser scanner (such as the FARO Focus S70) is used to acquire point cloud and defect images of land-based components (top mooring devices, upper part of the breakwater blocks). In the water-based domain, a customized underwater ROV scanning robot is used to acquire point cloud and defect data of underwater components (foundation blocks, underwater sections of pile piers, scour zones). In the water-land interface area, a combination of oblique scanning by a land scanner and close-range scanning by an ROV is used to ensure seamless data integration, achieving full-domain, high-precision, and multi-dimensional geometric and defect data acquisition. A schematic diagram of the UAV scanner is shown below. Figure 2 As shown, the ground-based 3D laser scanner is as follows Figure 3 As shown, the underwater ROV scanning robot is like Figure 4 As shown.

[0037] In this embodiment, airspace UAV laser scanning refers to the technical means of using the lidar equipment carried by the UAV to scan the superstructure of hydraulic structures to obtain point cloud and defect data.

[0038] In this embodiment, land-based three-dimensional laser scanning refers to a method of using a ground-mounted three-dimensional laser scanner to scan the portion of a hydraulic structure above the waterline to obtain data.

[0039] In this embodiment, underwater ROV robot scanning refers to the process of using scanning equipment carried by a remotely controlled underwater robot to detect and collect data on the underwater structure of hydraulic structures.

[0040] In this embodiment, the water-land connection area refers to the key part of the hydraulic structure where the water surface line and the land structure are connected and transitioned.

[0041] The beneficial effects of the above technical solution are: by integrating multiple active scanning technologies, it achieves comprehensive and seamless data collection of hydraulic structures from top to underwater structure, ensuring the integrity and accuracy of data in key connection areas, and laying a reliable data foundation for subsequent refined modeling and evaluation.

[0042] Example 3: Based on Example 2, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains, including: The airspace UAV laser scanning includes flight path planning, setting flight altitude and speed, and simultaneously acquiring laser point clouds and high-resolution images; The land area three-dimensional laser scanning includes setting up multiple scanning stations and stitching point clouds using a target sphere; The underwater ROV robot scanning includes collecting data by approaching the surface of the component in a constant-altitude cruising mode during low tide, and activating sonar and ultrasonic flaw detectors. The airspace, land, and underwater point cloud data are registered and fused using a multi-source data fusion algorithm, specifically including: A coarse registration algorithm based on FPFH features and RANSAC, and a fine registration algorithm based on ICP, are used to unify point cloud data from different domains into the same geographic coordinate system. A fusion algorithm based on Gaussian pyramid downsampling and Kriging interpolation is used to weightedly fuse point cloud data of different resolutions to generate a global point cloud dataset with uniform resolution.

[0043] In this embodiment, the key implementation points of the airspace UAV laser scanning include: using a parallel flight path deployment, with the flight path extending along the breakwater axis and the spacing between flight paths controlled at 10-20 m; controlling the flight altitude at 30-60 m and the flight speed at no more than 3 m / s; ensuring the lidar point cloud density is no less than 200 pts / m², and simultaneously enabling visible light image acquisition, with the overlap between the flight path and the lateral path being no less than 80% and 70%, respectively. The specific operation for land-based scanning is as follows: setting up a scanning station every 20 m along the top surface of the breakwater, using target sphere stitching to ensure point cloud continuity; setting the resolution to 1 mm and the scanning distance to 5-50 m. The specific operation for underwater scanning is as follows: launching into the water during slack tide, using a constant altitude cruise + body-hugging scanning mode, moving 0.5-1 m from the surface of the component at a speed ≤0.3 m / s; simultaneously activating a high-definition underwater camera, sonar, and ultrasonic flaw detector, and relying on the sonar obstacle avoidance system to avoid dangerous areas.

[0044] In this embodiment, route planning refers to a pre-defined three-dimensional spatial flight path for the UAV flight scan.

[0045] In this embodiment, flight altitude and speed refer to the relative flight altitude and travel speed parameters when controlling the UAV to perform scanning operations.

[0046] In this embodiment, laser point cloud and high-resolution image refer to the set of three-dimensional coordinates of the object surface obtained by lidar and high-resolution two-dimensional image data obtained by optical camera.

[0047] In this embodiment, a scanning station refers to each specific location where the instrument is set up and data is collected during a terrestrial 3D laser scanning operation.

[0048] In this embodiment, the target ball refers to a spherical marker with specific geometric features used in ground scanning to assist in the alignment and stitching of scanning data from different sites.

[0049] In this embodiment, point cloud stitching refers to the process of merging point cloud datasets obtained from multiple scans at different locations or from different perspectives into a complete and unified 3D point cloud model by matching common features.

[0050] In this embodiment, the slack tide period refers to the period when the water level is relatively stable and the water flow is relatively slow.

[0051] In this embodiment, the constant altitude cruise mode refers to the working mode in which the remotely controlled underwater robot maintains a constant distance from the surface of the component when performing scanning tasks.

[0052] In this embodiment, sonar refers to a device that uses the property of sound waves propagating in water to detect the shape and location of underwater objects.

[0053] In this embodiment, the ultrasonic flaw detector refers to a non-destructive testing device that uses ultrasonic waves to detect internal defects (such as cracks and voids) in materials.

[0054] The beneficial effects of the above technical solution are: by optimizing the specific operating parameters and collaborative modes of various scanning methods, it ensures the efficiency, high accuracy and completeness of data acquisition from the air, land to underwater, and in particular improves the detection capability of concealed areas and key defects, providing a solid and reliable technical guarantee for building accurate digital models.

[0055] In this embodiment, a multi-source data coordinate unified registration algorithm is used: The algorithm is used to unify point cloud data of airspace, land area and underwater area to the same geographic coordinate system. It is divided into two stages: coarse registration and fine registration. The specific description is as follows: (1) Coarse registration: The FPFH (Fast Point Feature Histogram) registration algorithm based on global features is adopted. The steps are as follows: ① Extract the FPFH feature descriptors of the point cloud of each domain to characterize the local geometric features of the point cloud. ② Using the land area three-dimensional laser scanning data as the reference coordinate system, find the corresponding feature point pairs between the point cloud of airspace and underwater area and the reference point cloud through feature matching. ③ Use the Random Sample Consensus (RANSAC) algorithm to remove erroneous matching points, calculate the initial spatial transformation matrix, complete the coarse registration, and control the coordinate deviation of the point cloud of each domain to the decimeter level. (2) Fine registration: The fine registration algorithm based on Iterative Closest Point (ICP) is adopted. The steps are as follows: ① Using the coarse registration result as the initial value, traverse each point of the point cloud of airspace and underwater area and search for the nearest point in the reference point cloud. ② Minimize the Euclidean distance error between point pairs and iteratively optimize the spatial transformation matrix until convergence. ③ For the land-water interface region, use the common target sphere of the overlapping scan data as control points to forcefully constrain the registration results, ensuring that there are no data gaps in the interface region, and finally achieving a registration accuracy of ±5mm.

[0056] To address the resolution differences among UAV point clouds (low resolution), land-based laser point clouds (high resolution), and underwater ROV point clouds (medium resolution), a Gaussian pyramid downsampling and interpolation fusion algorithm was adopted. (1) Constructing a Gaussian pyramid for point clouds: Multi-scale downsampling of high-resolution land-based point clouds was performed to generate point cloud pyramids at different resolution levels. (2) Resolution matching: The point clouds in the airspace and underwater domains were matched with the land-based point clouds at the corresponding levels of the pyramids. Kriging interpolation was used to supplement details in low-resolution areas. (3) Weighted fusion: Weights were assigned based on the confidence levels of point clouds in different domains. The weight of land-based point clouds was set to 0.5, and the weights of airspace and underwater point clouds were each set to 0.25. A unified resolution global point cloud dataset was generated by weighted averaging.

[0057] In this embodiment, the code deployment and verification process of the multi-source data fusion algorithm is as follows: The code deployment and verification process of the multi-source data fusion algorithm is as follows: (1) Data Input: Airspace, land, and underwater point cloud files, UAV imagery, and ROV ultrasonic data are required. (2) Accuracy Verification: This part is divided into point cloud registration accuracy test and defect information fusion accuracy test. ① The evaluation index for point cloud accuracy test requires extracting the coordinates of 20 target spheres in the fused point cloud, comparing them with the true values ​​from the total station, and calculating the mean square error (RMSE). After registration, the RMSE of the overlapping area of ​​the point cloud should be ≤5mm. The RMSE is calculated as follows: ;in, The root mean square error (RMSE) represents the average deviation between the registered point cloud data and the actual spatial location. The smaller the RMSE value, the higher the registration accuracy and the better the data fusion effect. This indicates the number of sampling points, n=20; Indicates the sampling point number value; This indicates the first element in the fused global point cloud dataset. The x-coordinate value of each sampling point; This indicates the first element in the fused global point cloud dataset. The ordinate values ​​of each sampling point; This indicates the first element in the fused global point cloud dataset. The vertical coordinate values ​​of each sampling point; This indicates the first measurement obtained based on the total station's field measurements. The true value of the x-coordinate of each sampling point; This indicates the first measurement obtained based on the total station's field measurements. The true value of the ordinate of each sampling point; This indicates the first measurement obtained based on the total station's field measurements. The true value of the vertical coordinates of each sampling point; and the water-land connection area is required to have a certain integrity, that is, the proportion of the point cloud without faults in the water-land connection area is ≥98% through visual inspection + algorithm automatic detection. ② Defect information fusion accuracy test. Run the U-Net semantic segmentation + DS evidence theory fusion algorithm to output the defect list. Then compare the fusion result with the true value of manual drilling; where, the defect detection rate = number of detected defects / number of true defects × 100%, and the qualified standard is required to be ≥95%. The defect location error calculation method is the same as RMSE, and the qualified standard is ≤10cm. (3) Output results: full-domain fused point cloud, defect vector file (can be directly imported into GIS / BIM platform).

[0058] Example 4: Based on Example 1, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains. Step 2, before identifying and segmenting independent components from the full-domain point cloud data based on geometric features, includes: The global point cloud data is preprocessed by denoising and multi-site point cloud stitching, and a target point cloud set is generated based on the preprocessing results; Specifically, based on the target point cloud set, components with independent geometric shapes are identified and segmented according to at least one of geometric feature matching, region growing, or deep learning segmentation algorithms, generating a component point cloud subset.

[0059] In this embodiment, the original scan data needs to be preprocessed before component identification and segmentation. Specifically, this includes: using CloudCompare software to remove noise points (such as water flow or marine life shadows) from the point cloud; stitching together multi-station scan data using a target sphere and GPS coordinates; and controlling the overall point cloud accuracy within ±5mm. The processed point cloud data forms a target point cloud set in a unified coordinate system, providing a high-quality data foundation for subsequent automatic component identification and segmentation based on geometric features (shape, size, elevation).

[0060] In this embodiment, multi-site cloud stitching refers to the operation of fusing multiple sets of point cloud data collected from different scanning locations or different devices into a unified and complete point cloud dataset through coordinate registration.

[0061] In this embodiment, preprocessing refers to the preparatory work of cleaning, integrating and optimizing the raw data before the main data analysis and processing steps.

[0062] In this embodiment, the target point cloud set refers to the point cloud dataset that has been preprocessed, meets the quality requirements, and is used for subsequent component identification and segmentation.

[0063] The beneficial effects of the above technical solution are: by performing preprocessing such as denoising and stitching on the original point cloud data, the integrity and quality of the data are significantly improved, laying a reliable data foundation for subsequent accurate identification and segmentation of components.

[0064] Example 5: Based on Example 1, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains. In step 3, a target BIM model with unique identifiers and attribute information is constructed based on a subset of component point clouds and structural defect data. Simultaneously, the target BIM model is integrated into a GIS platform to obtain the component's location information in macro-geographic space, including: Generate 3D geometric models of each component in BIM modeling software based on component point cloud subsets; A unique ID number is added to the 3D geometric model of each generated component. The encoding rules for the ID number include: facility type, mileage location, component type, and sequence number information. Structural defect data is used as defect attributes, and the design attributes and material attributes of the pre-collected data are jointly associated with the ID number of the corresponding component to generate complete attribute information of the component; The target BIM model, which integrates the three-dimensional geometric model of the components, unique ID number, and complete attribute information, is imported into the GIS platform; The imported target BIM model is calibrated in the GIS platform based on the coordinate control points obtained from on-site measurements; The target BIM model, after coordinate calibration, is spatially overlaid with the base map and corresponding thematic layers in the GIS platform to obtain the latitude, longitude, and elevation positioning information of each component in the macro-geographic space. The acquisition of the structural defect data includes a structural defect information complementarity fusion algorithm, specifically: U-Net semantic segmentation algorithm is used to identify apparent defects in airspace images, and waveform feature analysis algorithm is used to identify internal defects in underwater ultrasonic data. Based on the registered point cloud coordinates, the apparent defects and internal defects are mapped to the three-dimensional space of the component. The DS evidence theory is used to fuse multi-source defect data to generate a comprehensive defect assessment result of the component as the defect attribute.

[0065] In this embodiment, the unique ID number refers to a unique identifier string assigned to each component, encoded according to specific rules, and not repeated.

[0066] In this embodiment, the mileage position refers to the longitudinal position distance calculated along the axis of the hydraulic structure or a specific reference line.

[0067] In this embodiment, design attributes refer to the original design parameter information of components derived from engineering design drawings and documents.

[0068] In this embodiment, material properties refer to information describing the type, specifications, strength, and other characteristics of the materials used in the component.

[0069] In this embodiment, coordinate control points refer to reference points that are pre-determined on-site and have known precise geodetic coordinates, used to align the model to the real-world coordinate system.

[0070] In this embodiment, coordinate calibration refers to the process of adjusting the spatial position and orientation of the model using coordinate control points to make it consistent with the actual geographic coordinate system.

[0071] In this embodiment, the base map refers to the basic background map in the geographic information system, which is used to provide reference information such as terrain, water system, and roads.

[0072] In this embodiment, the thematic layer refers to a data layer that is overlaid on the base map in a geographic information system and represents a specific thematic element (such as administrative division or geological information).

[0073] In this embodiment, the BIM model construction specifically involves: importing the segmented component point cloud into Revit software to generate a parametric BIM model; assigning a unique ID number to each component (numbering rule: facility type-work area-component type-serial number, such as: FB-005-FLK-01 (FB represents breakwater, 005 represents the work area number to which the component belongs, FLK represents wave-breaking block, and 01 represents the sequential number of this type of component under this work area); and binding basic attributes such as material, size, and construction time; adding defect attributes (crack location / length, corrosion area, scour depth) and environmental attributes (tide level of the water area, whether it is in an ecologically sensitive area) collected by scanning. GIS integration specifically involves: importing the BIM model into the ArcGIS platform; using RTK coordinate calibration; accurately overlaying the model onto the Black Point Harbour GIS map; associating macro-geographic layers such as waterways, ecologically sensitive areas, and tide gauge stations; and achieving three-dimensional spatial positioning of the components using latitude, longitude, and elevation, forming a complete data chain of "component ID - spatial coordinates - attribute information".

[0074] In this embodiment, the structural defect information complementary fusion algorithm is used to integrate defect data collected from various domains (such as surface cracks in aerial images and internal defects of underwater ROVs) to form a complete defect file of the component. (1) Defect feature extraction: Deep learning semantic segmentation algorithms (such as U-Net) are used to identify surface cracks, corrosion and other apparent defects in high-resolution aerial images; waveform feature analysis algorithms are used to extract hidden defects such as internal voids and concrete spalling from ultrasonic flaw detection data of underwater ROVs. (2) Defect spatial association: Based on the registered point cloud coordinates, the defect information identified from each domain is mapped to the three-dimensional spatial position of the corresponding component to establish the association relationship of "defect type-location coordinates-severity". (3) Defect information fusion: Evidence theory (DS theory) is used to fuse multi-source defect data to solve the problem of conflicting results of different detection methods at the same location and output a comprehensive defect assessment result.

[0075] The beneficial effects of the above technical solution are: by accurately converting point cloud data into a three-dimensional model with unique identifiers and rich attribute information, and deeply integrating it with the geographic information system, the organic unity from micro-component attributes to macro-geographical location is achieved. This not only ensures the spatial accuracy of the model, but also endows each component with a traceable and complete information archive, greatly enhancing the spatial visibility, information integration and decision support capabilities of the refined management of hydraulic structures.

[0076] Example 6: Based on Example 1, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains. In step 4, the attribute information of each component in the target BIM model is comprehensively evaluated from multiple dimensions based on evaluation dimensions to determine the control level of each component, including: The evaluation dimensions are multiple, including: structural functional importance, material reuse value, structural health status, demolition and construction risks, and the degree of impact on the water environment. The target score for each evaluation dimension is calculated based on the attribute information of each component in the target BIM model. Obtain the dimensional weights of each evaluation dimension, and calculate the weighted sum based on the target score of each evaluation dimension and the corresponding dimensional weight to obtain the comprehensive score; The control level classification thresholds are obtained separately, including: a first threshold and a second threshold, wherein the first threshold is greater than the second threshold; When the overall score is equal to or greater than the first threshold, the component is classified as the first control level. When the overall score is less than the first threshold but greater than the second threshold, the component will be classified as the second control level. When the overall score is less than or equal to the second threshold, the component will be classified as the third control level.

[0077] In this embodiment, the evaluation dimensions and their weights are: structural functional importance (weight 30%), material reuse value (weight 25%), structural health status (weight 25%), demolition construction risk (weight 15%), and water environment impact (weight 5%). Based on the component attributes in the BIM model, the platform automatically extracts the classification indicators for each component (such as whether it is a core load-bearing component, concrete strength grade, defect grade, and construction risk of the area), calculates the score according to the scoring standards of each dimension in the "Classification and Grading Table for Demolition of Black Point Harbor Breakwater" (as shown in Appendix 1), and then calculates the comprehensive score by weighting. The classification thresholds are set as follows: Level 1 control ≥ 80 points (first threshold), Level 2 control 60 (second threshold) - 79 points (since the score is rounded to the nearest integer, a score less than the first threshold is 79 points), and Level 3 control... 60 points (second threshold) (as shown in Appendix 2). The component control level is automatically determined based on the comprehensive score.

[0078] Appendix 1: Grading Standards Table Appendix 2 Control Level Table In this embodiment, structural functional importance refers to the degree of structural role and safety criticality that the component plays in the entire hydraulic structure.

[0079] In this embodiment, the value of material reuse refers to the economic or resource value of the component material that can be recycled and reused after dismantling.

[0080] In this embodiment, structural health status refers to the current physical state and integrity of a component as assessed based on defect data.

[0081] In this embodiment, the demolition construction risk refers to the safety risks and technical difficulties that may exist during the removal or handling of the component.

[0082] In this embodiment, the degree of water environment impact refers to the magnitude of the potential impact of the component or its dismantling operation on the ecological environment of the surrounding waters.

[0083] In this embodiment, the dimension weight refers to the importance ratio coefficient assigned to each evaluation dimension in the comprehensive evaluation calculation.

[0084] In this embodiment, the control level classification threshold refers to a pre-set specific score limit value used to classify the comprehensive score into different control levels.

[0085] In this embodiment, the first control level refers to the component level with the highest overall score that requires the highest priority or strictest control.

[0086] In this embodiment, the second control level refers to the component level with a moderate overall score that requires general control.

[0087] In this embodiment, the third control level refers to the component level with the lowest overall score and relatively low control priority.

[0088] In this embodiment, the multi-dimensional comprehensive evaluation system establishes five fixed evaluation dimensions: structural functional importance, material reuse value, structural health status, demolition and construction risks, and the degree of impact on the water environment. Quantitative scoring standards and weights are assigned to each dimension, ensuring the objectivity of the evaluation results and reducing the influence of subjective human experience. The evaluation targets the specific state of hydraulic structures after long-term service. Its dynamic nature is reflected in decision support in response to critical dynamic load events such as strong winds and waves, rather than continuous real-time monitoring of the structure. This aligns with the long-term service nature of hydraulic structures.

[0089] The beneficial effects of the above technical solution are: by establishing a multi-dimensional evaluation system covering structure, materials, health, risk and environment, and by adopting a weighted calculation and threshold grading mechanism, a scientific and quantitative assessment of component management priorities is achieved, making subsequent maintenance decisions and resource allocation more objective and targeted, and effectively improving the level of refinement and intelligence in facility management.

[0090] Example 7: Based on Example 1, this example provides a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains. Step 5: Based on the control level of each component and its location information in macro-geographic space, corresponding control zones are divided in space, and the unique identifier of each component is associated with its respective control zone, including: Based on the control level of each component, components with the same control level and continuously distributed in space are classified to obtain the component set under each control level. Obtain the macro-geospatial information location of the components, and determine the spatial distribution boundaries of various components based on the macro-geospatial information location of the components; Based on the spatial distribution boundaries, physical areas corresponding to the control levels are divided in the macro-geographic space, and control zones are generated based on the division results; A unique partition number is added within the controlled partition, wherein the partition number code includes the corresponding control level information and the corresponding spatial range information; Associate and bind the unique identifier of all components located within the same control zone with the corresponding control zone number.

[0091] In this embodiment, the specific implementation of the partitioning and numbering includes: ①Zoning: Control zones are divided according to control level and spatial location, and marked with different colors (Level 1 zone: red, Level 2 zone: yellow, Level 3 zone: green). Each zone is assigned a zone number.

[0092] ② Numbering rules: The coding rules for zone numbering include the corresponding control level information and the corresponding spatial range information, for example: YJ-001-050, where YJ = Level 1 control, 001-050 = mileage 1m-50m.

[0093] ③ Component and partition binding: Each component's ID number is associated with its partition number, enabling bidirectional traceability between "partition and component". For example, component FB-005-FLK-01 within partition YJ-001-050 can be quickly queried for partition control requirements and component details.

[0094] In this embodiment, the component set refers to a subset formed by grouping and classifying all components according to the same control level and spatial continuity principle.

[0095] In this embodiment, the spatial distribution boundary refers to the outer contour line used to define the area occupied by a set of components with the same control level in the macro-geographic space.

[0096] In this embodiment, a physical region refers to a geographical area with a defined range that is delineated in the real world or on a map based on spatial distribution boundaries.

[0097] In this embodiment, the partition number refers to a unique code identifier assigned to each designated control partition, which contains level and spatial location information.

[0098] The beneficial effects of the above technical solution are as follows: by clustering components with the same management priority in physical space and delineating clear boundary areas based on their actual geographical distribution range, the management of discrete components has been elevated to continuous spatial area management. At the same time, by establishing a two-way association between components and zones, a refined spatial management system has been formed, which greatly facilitates the subsequent formulation of differentiated inspection, maintenance and control strategies according to regions, and improves the organizational efficiency of management work and the rationality of resource allocation.

[0099] Example 8: Based on Example 1, this example provides a refined zoning assessment method for hydraulic structure components based on intelligent identification across land, sea, and air domains. In step 6, a visual zoning map is generated based on the control zoning, and a component information list is generated based on the target BIM model and control level. As the assessment result, it includes: The control zones and their corresponding control levels are rendered on the base map of the GIS platform, based on different legends, to generate a GIS macro-level map. In the target BIM model, each component is rendered according to its control level to generate a BIM micro-component classification model that can display the classification results of the component level. The key information of the GIS macro-level map and the BIM micro-component level model is processed according to a preset scale to output a paper-based level map on site. The attribute information of each component is extracted based on the target BIM model, and a component information list containing the unique identifier, attribute information, control level and spatial coordinates is generated based on the control level and location information. Based on the control zones and control levels, and according to the component information list, a zone control requirements manual is constructed for each zone, including the dismantling sequence, equipment selection, safety protection, and environmental protection measures.

[0100] In this embodiment, the visualization results include: ① GIS macro-level classification map: Control zones (red / yellow / green) are overlaid on the Noi Pointe-Ha port area map, with zone numbers, mileage ranges, and core control requirements marked (e.g., blasting is prohibited in the first-level zone, and double-layer anti-fouling curtains are required); ② BIM micro-level component classification model: The classification results of each component are displayed in three dimensions, and detailed attributes (defects, scores, and disposal suggestions) can be viewed by clicking on the component; ③ On-site paper classification map: The zone classification map is printed at a scale of 1:200 and posted at the on-site construction command center. The data list results include: ① "List of Classified and Graded Components for Demolition of Hydraulic Facilities in Noi Pointe-Ha Port": Includes information such as component ID, zone number, structural type, material value, health status, construction risks, disposal suggestions, and spatial coordinates; ② "Classification and Zoning Control Requirements Manual": Clearly defines the specific requirements for demolition sequence, equipment selection, safety protection, and environmental protection measures for each control level.

[0101] In this embodiment, the GIS macro-level map refers to a visual map on a geographic information system platform that uses a map as a background and different legend colors or styles to display various spatial control zones and their levels.

[0102] In this embodiment, the BIM micro-component hierarchical model refers to a three-dimensional model in the building information model that distinguishes each component by color or label according to its control level, thereby displaying a detailed hierarchical distribution.

[0103] In this embodiment, the on-site paper grading map refers to a paper drawing that is printed out after processing the key information in the GIS macro grading map and the BIM micro component grading model according to a scale and size suitable for on-site use.

[0104] In this embodiment, the component information list refers to a detailed data file that summarizes all key attributes, control levels, and spatial location information of each component in the form of a table or list.

[0105] In this embodiment, the zone control requirements manual refers to a special work guidance document formulated for each control zone based on the overall situation of its internal components, which includes specific dismantling steps, required equipment, safety measures and environmental protection requirements.

[0106] The beneficial effects of the above technical solution are as follows: by generating multi-level visualization results from macro-geographic zoning to micro-component models, and combining detailed component lists and special management manuals, the assessment data is transformed into intuitive decision charts and operable management documents, which significantly improves the presentation effect and practical application value of the assessment results, and provides a systematic and standardized direct basis for subsequent engineering management, maintenance and demolition and environmental protection.

[0107] Example 9: Based on Example 4, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across land, sea, and air domains. The method involves preprocessing the full-domain point cloud data by denoising and stitching together multi-site point cloud data, including: Based on point cloud processing software, outlier noise points are identified and removed from the global point cloud data.

[0108] Based on the target spheres deployed at the scanning site or the recorded GPS coordinates, spatial transformation parameters between point cloud data obtained from different scanning stations are calculated.

[0109] Based on the spatial transformation parameters, the point cloud data from multiple stations are aligned and merged into a unified point cloud dataset in a single coordinate system.

[0110] The entire point cloud dataset is used as the target point cloud set for recognition and segmentation operations.

[0111] In this embodiment, the specific operations of the point cloud data preprocessing are as follows: using CloudCompare software to denoise and remove noise points (such as water flow and marine organism shadows) from the point cloud; calculating the spatial transformation parameters between each scanning station using pre-deployed target spheres or recorded GPS coordinates; using these parameters to align and stitch the point clouds scanned by multiple stations into a unified coordinate system overall point cloud dataset, with the overall point cloud accuracy controlled within ±5mm, forming the target point cloud set used for subsequent component recognition and segmentation.

[0112] In this embodiment, outlier noise refers to isolated or abnormal data points in point cloud data that are far from the surface of the real object due to scanning errors or environmental interference.

[0113] In this embodiment, spatial transformation parameters refer to the mathematical parameters required to align point cloud data acquired by one scanning station to the coordinate system of another scanning station or a unified coordinate system through operations such as rotation and translation.

[0114] In this embodiment, the overall point cloud dataset refers to the complete point cloud data set covering the entire hydraulic structure under a unified coordinate system, formed by aligning and merging point cloud data acquired from multiple scanning stations.

[0115] The beneficial effects of the above technical solution are: by removing noise from point cloud data and accurately stitching it together based on on-site reference points, the accuracy and overall consistency of the data are effectively improved, providing a high-quality and complete data foundation for the accurate identification and segmentation of subsequent components.

[0116] Example 10: Based on Example 1, this example provides a refined zoning evaluation method for hydraulic structure components based on intelligent identification across three domains (sea, land, and air). Before step 1, it further includes: Based on the results of the on-site survey, and using an AI recognition algorithm based on a CNN and SVM fusion model to analyze historical data, on-site images and terrain data, the scanning boundary of the hydraulic structure is determined, and dangerous areas and safe working areas are intelligently marked based on the scanning boundary to generate a risk heat map of dangerous areas. Record the hydrological conditions of the safe operating area and determine the optimal scanning window based on tidal cycle and wave height data; Collect the original design data, inspection, maintenance and operation data, and port area geographical data including coordinate information of the hydraulic structures; Based on the scanning boundaries, danger zone markers, and geographic data, the operation paths of UAVs and underwater ROVs are planned using reinforcement learning algorithms, and the deployment scheme of land-based three-dimensional laser scanning stations is optimized using genetic algorithms. The equipment deployment and operation paths of the land-sea-air three-domain collaborative acquisition mechanism are determined. A multi-agent collaborative decision-making mechanism for dynamically adjusting the operation plan based on real-time environmental data is also constructed.

[0117] In this embodiment, preliminary preparations are required before scanning and data acquisition. These include: organizing an expert team to conduct on-site surveys, delineating scanning boundaries, defining the start and end points of breakwaters and revetments, marking dangerous areas such as navigation channels, ecologically sensitive areas, and underwater shoals / reefs, and establishing prohibited scanning zones and safe operating zones; recording hydrological conditions, acquiring data such as tidal cycles, wave heights, and current speeds, and determining the optimal scanning window (prioritizing periods of slack tide and wave heights <0.5m); collecting design data (original design drawings, component dimensions, material specifications), operational data (previous inspection reports, maintenance records), and geographical data (port area GIS map, navigation channel planning map, and ecologically sensitive area distribution map); and planning UAV flight routes, ground scanning stations, and ROV operation paths based on the above information to ensure safe, efficient, and blind-spot-free scanning operations.

[0118] In this embodiment, the scanning boundary refers to the spatial limit of the hydraulic structure and its surrounding area that needs to be covered for this data acquisition operation.

[0119] In this embodiment, the danger zone refers to a local area within the scanning boundary that may pose a safety threat to personnel or equipment (such as structural instability, high voltage, etc.).

[0120] In this embodiment, the safe working area refers to the area within the scanning boundary that has been assessed and confirmed as a safe place for personnel and equipment to carry out data acquisition operations.

[0121] In this embodiment, the optimal scanning window period refers to the specific time period that is most conducive to safe and efficient underwater and waterline data collection, based on hydrological conditions such as tides and waves.

[0122] In this embodiment, the historical inspection, maintenance and operation data refers to all records and documents related to the inspection, testing, maintenance and repair, and daily operation of the hydraulic structure throughout its history.

[0123] In this embodiment, port area geographic data refers to spatial data containing information such as port area topography, coordinate control points, and existing surveying results.

[0124] In this embodiment, equipment deployment and operation path refer to the installation / deployment locations and mobile scanning routes determined for various types of scanning equipment in the air, on land, and underwater, based on prior planning.

[0125] The beneficial effects of the above technical solution are: by conducting systematic on-site surveys, safety planning, environmental analysis, and historical data integration before data collection, a scientific and safe implementation plan is formulated for subsequent collaborative operations, effectively avoiding operational risks, and making full use of environmental windows and existing information, ensuring the efficiency, smoothness, and reliability of data collection work.

[0126] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for fine partition evaluation of hydraulic structure components based on intelligent identification of sea-land-air three domains, characterized in that, include: Step 1: Collect full-domain point cloud data and structural defect data of hydraulic structures based on the three-domain collaborative acquisition mechanism of sea, land and air; Step 2: Identify and segment independent components in the global point cloud data based on geometric features, and generate component point cloud subsets; Step 3: Based on the component point cloud subset and structural defect data, construct a target BIM model with unique identifiers and attribute information. At the same time, integrate the target BIM model into the GIS platform to obtain the component's location information in macro-geographic space. Step 4: Based on the evaluation dimensions, conduct a multi-dimensional comprehensive evaluation of the attribute information of each component in the target BIM model to determine the control level of each component; Step 5: Based on the control level of each component and the component's location information in the macro-geographic space, divide the space into corresponding control zones and associate the component's unique identifier with its respective control zone; Step 6: Generate a visual zoning map based on the control zones, and generate a list of component information based on the target BIM model and control level as the evaluation result.

2. The method according to claim 1, characterized in that, In step 1, the three-domain (sea, land, and air) collaborative data collection mechanism includes: Point cloud data and structural defect data of the superstructure of the hydraulic structure were obtained by aerial drone laser scanning. Point cloud data and structural defect data of the structure above the waterline were obtained based on 3D laser scanning of the land area; Point cloud data and structural defect data of underwater structures were obtained by scanning with an underwater ROV robot. By using terrestrial 3D laser and ROV robot to perform overlapping scans of the land-water interface area, point cloud data and structural defect data of the land-water interface area are obtained.

3. The method according to claim 2, wherein, include: The airspace UAV laser scanning includes flight path planning, setting flight altitude and speed, and simultaneously acquiring laser point clouds and high-resolution images; The land area three-dimensional laser scanning includes setting up multiple scanning stations and stitching point clouds using a target sphere; The underwater ROV robot scanning includes collecting data by approaching the surface of the component in a constant-altitude cruising mode during low tide, and activating sonar and ultrasonic flaw detectors. The airspace, land, and underwater point cloud data are registered and fused using a multi-source data fusion algorithm, specifically including: A coarse registration algorithm based on FPFH features and RANSAC, and a fine registration algorithm based on ICP, are used to unify point cloud data from different domains into the same geographic coordinate system. A fusion algorithm based on Gaussian pyramid downsampling and Kriging interpolation is used to weightedly fuse point cloud data of different resolutions to generate a global point cloud dataset with uniform resolution.

4. The method according to claim 1, characterized in that, In step 2, before identifying and segmenting independent components in the global point cloud data based on geometric features, the following steps are included: The global point cloud data is preprocessed by denoising and multi-site point cloud stitching, and a target point cloud set is generated based on the preprocessing results; Specifically, based on the target point cloud set, components with independent geometric shapes are identified and segmented according to at least one of geometric feature matching, region growing, or deep learning segmentation algorithms, generating a component point cloud subset.

5. The method according to claim 1, characterized in that, In step 3, based on the component point cloud subset and structural defect data, a target BIM model with unique identifiers and attribute information is constructed. Simultaneously, the target BIM model is integrated into a GIS platform to obtain the component's location information in macro-geographic space, including: Generate 3D geometric models of each component in BIM modeling software based on component point cloud subsets; A unique ID number is added to the 3D geometric model of each generated component. The encoding rules for the ID number include: facility type, mileage location, component type, and sequence number information. Structural defect data is used as defect attributes, and the design attributes and material attributes of the pre-collected data are jointly associated with the ID number of the corresponding component to generate complete attribute information of the component; The target BIM model, which integrates the three-dimensional geometric model of the components, unique ID number, and complete attribute information, is imported into the GIS platform; The imported target BIM model is calibrated in the GIS platform based on the coordinate control points obtained from on-site measurements; The target BIM model, after coordinate calibration, is spatially overlaid with the base map and corresponding thematic layers in the GIS platform to obtain the latitude, longitude, and elevation positioning information of each component in the macro-geographic space. The acquisition of the structural defect data includes a structural defect information complementarity fusion algorithm, specifically: U-Net semantic segmentation algorithm is used to identify apparent defects in airspace images, and waveform feature analysis algorithm is used to identify internal defects in underwater ultrasonic data. Based on the registered point cloud coordinates, the apparent defects and internal defects are mapped to the three-dimensional space of the component. The DS evidence theory is used to fuse multi-source defect data to generate a comprehensive defect assessment result of the component as the defect attribute.

6. The method according to claim 1, wherein, In step 4, the attribute information of each component in the target BIM model is comprehensively evaluated from multiple dimensions based on the evaluation dimensions to determine the control level of each component, including: The evaluation dimensions are multiple, including: structural functional importance, material reuse value, structural health status, demolition and construction risks, and the degree of impact on the water environment. The target score for each evaluation dimension is calculated based on the attribute information of each component in the target BIM model. Obtain the dimensional weights of each evaluation dimension, and calculate the weighted sum based on the target score of each evaluation dimension and the corresponding dimensional weight to obtain the comprehensive score; The control level classification thresholds are obtained separately, including: a first threshold and a second threshold, wherein the first threshold is greater than the second threshold; When the overall score is equal to or greater than the first threshold, the component is classified as the first control level. When the overall score is less than the first threshold but greater than the second threshold, the component will be classified as the second control level. When the overall score is less than or equal to the second threshold, the component will be classified as the third control level.

7. The method according to claim 1, characterized in that, Step 5: Based on the control level of each component and its location information in macro-geographic space, divide the space into corresponding control zones, and associate the unique identifier of each component with its respective control zone, including: Based on the control level of each component, components with the same control level and continuously distributed in space are classified to obtain the component set under each control level. Obtain the macro-geospatial information location of the components, and determine the spatial distribution boundaries of various components based on the macro-geospatial information location of the components; Based on the spatial distribution boundaries, physical areas corresponding to the control levels are divided in the macro-geographic space, and control zones are generated based on the division results; A unique partition number is added within the controlled partition, wherein the partition number code includes the corresponding control level information and the corresponding spatial range information; Associate and bind the unique identifier of all components located within the same control zone with the corresponding control zone number.

8. The method according to claim 1, characterized in that, In step 6, a visual zoning map is generated based on the control zones, and a component information list is generated based on the target BIM model and the control level. These components serve as the evaluation outcome and include: The control zones and their corresponding control levels are rendered on the base map of the GIS platform, based on different legends, to generate a GIS macro-level map. In the target BIM model, each component is rendered according to its control level to generate a BIM micro-component classification model that can display the classification results of the component level. The key information of the GIS macro-level map and the BIM micro-component level model is processed according to a preset scale to output a paper-based level map on site. The attribute information of each component is extracted based on the target BIM model, and a component information list containing the unique identifier, attribute information, control level and spatial coordinates is generated based on the control level and location information. Based on the control zones and control levels, and according to the component information list, a zone control requirements manual is constructed for each zone, including the dismantling sequence, equipment selection, safety protection, and environmental protection measures.

9. The method according to claim 4, characterized in that, The preprocessing of the global point cloud data, including denoising and multi-site point cloud stitching, includes: Based on point cloud processing software, outlier noise points are identified and removed from the global point cloud data; Based on the target spheres deployed at the scanning site or the recorded GPS coordinates, calculate the spatial transformation parameters between point cloud data obtained from different scanning stations; Based on the spatial transformation parameters, the point cloud data from multiple stations are aligned and merged into a unified point cloud dataset in a unified coordinate system. The entire point cloud dataset is used as the target point cloud set for recognition and segmentation operations.

10. The method according to claim 1, characterized in that, Before step 1, the following is also included: Based on the results of the on-site survey, and using an AI recognition algorithm based on a CNN and SVM fusion model to analyze historical data, on-site images and terrain data, the scanning boundary of the hydraulic structure is determined, and dangerous areas and safe working areas are intelligently marked based on the scanning boundary to generate a risk heat map of dangerous areas. Record the hydrological conditions of the safe operating area and determine the optimal scanning window based on tidal cycle and wave height data; Collect the original design data, inspection, maintenance and operation data, and port area geographical data including coordinate information of the hydraulic structures; Based on the scanning boundaries, danger zone markers, and geographic data, the operation paths of UAVs and underwater ROVs are planned using reinforcement learning algorithms, and the deployment scheme of land-based three-dimensional laser scanning stations is optimized using genetic algorithms. The equipment deployment and operation paths of the land-sea-air three-domain collaborative acquisition mechanism are determined. A multi-agent collaborative decision-making mechanism for dynamically adjusting the operation plan based on real-time environmental data is constructed.

Images