An intelligent monitoring method and system for post-stage management and protection of ecological restoration engineering

Abstract

The application discloses an intelligent monitoring method and system for post-period management and maintenance of an ecological restoration project, and belongs to the technical field of post-period management and maintenance of ecological restoration projects, which is used to solve the technical problem that the current post-period management and maintenance of ecological restoration projects has high cost and great risk in invasive monitoring, and it is difficult to accurately determine the long-term stability of the ecological restoration body, which is not conducive to realizing intelligent monitoring. The method comprises the following steps: periodically collecting and processing the trace change characteristics of each accessory based on the geometric center point of the current ecological restoration project; detecting the trace change of the anchor rod trace change amount, the anchor rod end angle change amount, the protective net node displacement amount and the drainage ditch scouring change amount to obtain the significant change feature point; matching the change vector field and the adjacent relationship graph under the association rule to output the association analysis result; structuring the problem association data to determine the inversion state, and obtaining the inversion state judgment table of the current ecological restoration project; and generating a risk response strategy.

Description

Technical Field

[0001] This application relates to the field of post-construction management technology for ecological restoration projects, and in particular to an intelligent monitoring method and system for post-construction management of ecological restoration projects. Background Technology

[0002] In ecological restoration projects on mine slopes, steep rock walls, and decommissioned sites, numerous auxiliary structures are typically installed to ensure the stability of the restored structure. These include: anchoring systems (anchors, cables, and nails), protective systems (active and passive protective nets, and lattice beams), drainage systems (intercepting ditches, drainage ditches, and drop troughs), and vegetation support structures (vegetation bag fixing nets and climbing plant supports). After the restoration project is completed and accepted, these structures are usually only used as reference points for later maintenance and inspection, and the wealth of condition information they contain is often overlooked.

[0003] However, existing post-restoration management technologies still have certain technical shortcomings. For example, the rust morphology of the exposed ends of anchor bolts, the local deformation patterns of protective nets, and the scouring marks on the walls of drainage ditches objectively record key information such as stress changes, displacement trends, and seepage paths within the restored body. However, existing technologies do not utilize these surface traces of engineering accessories as monitoring sources, relying solely on human experience for post-restoration management.

[0004] Secondly, traditional invasive monitoring is costly and risky. Re-deploying sensors on completed restorations requires drilling, burying, and wiring, which is not only expensive but also poses a risk of damage during construction. This is especially true in areas where personnel cannot easily access, such as steep rock walls and radioactive sites, where the deployment and maintenance of monitoring equipment are extremely difficult.

[0005] Therefore, there is an urgent need for a non-invasive, low-cost, all-time diagnostic method for the post-construction management of ecological restoration projects, so as to achieve intelligent monitoring of the post-construction management of ecological restoration projects. Summary of the Invention

[0006] This application provides an intelligent monitoring method and system for the post-construction management of ecological restoration projects, which addresses the following technical problems: In existing post-construction management of ecological restoration projects, invasive monitoring is costly and risky, and it is difficult to accurately determine the long-term stability of the ecological restoration body, which is not conducive to achieving intelligent monitoring.

[0007] The embodiments of this application adopt the following technical solutions: On the one hand, this application provides an intelligent monitoring method for the post-construction management of ecological restoration projects, including: periodically collecting and processing trace change characteristics of each appendage based on the spatial information coding of appendages in the ecological restoration project, and generating an appendage trace information database related to the ecological restoration project; detecting trace changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour through the appendage trace information database to obtain significant change feature points; matching the change vector field with the proximity relationship diagram under association rules according to the spatial location of the appendages corresponding to the significant change feature points, and outputting association analysis results; performing structured inversion state determination on the problem association data in the association analysis results to obtain an inversion state determination table for the current ecological restoration project; matching the corresponding management measures for the problem area state according to the spatial influence range of the problem area in the inversion state determination table, and generating a risk response strategy for the current ecological restoration project based on a graded response mechanism.

[0008] This application's embodiments utilize an intelligent monitoring method to automatically collect and process changes in traces of appendages in ecological restoration projects, thereby improving monitoring efficiency and reducing labor costs. Periodic data collection and processing enables real-time monitoring of ecological restoration projects, allowing for timely detection of potential problems and preventing damage to the ecological environment. Furthermore, establishing an appendage trace information database allows for the accumulation of long-term data, facilitating long-term tracking and research of ecological restoration projects. Moreover, trace change detection and association rule matching enable more precise analysis of the ecological restoration project's status, improving the accuracy of analysis results. Based on an inversion state determination table and a tiered response mechanism, management measures can be quickly formulated and implemented, increasing the response speed to problem areas. Simultaneously, risk assessment and response strategy development effectively reduce the risks of ecological restoration projects, ensuring ecological stability. Finally, it also contributes to improving the management level of ecological restoration projects, achieving scientific and refined management.

[0009] In one feasible implementation, before periodically collecting and processing the trace change characteristics of each appendage based on the geometric center point of the current ecological restoration project according to the spatial information encoding of the appendages in the ecological restoration project, and generating a database of appendage trace information related to the ecological restoration project, the method further includes: using a real-time dynamic differential positioning device to mark all visible engineering appendages in the restoration area corresponding to the current ecological restoration project one by one to obtain appendage category information; wherein, the visible engineering appendages include at least: anchor bolt ends, protective net nodes, drainage ditch feature points, and lattice beam nodes; and marking each appendage in the marked appendage location information with exclusive marking information to obtain appendage marking information, including: performing color recognition on the anchor bolt ends based on image registration points. The process involves: processing the anchor bolt marking information; performing hardware-based QR code information appending processing on the protective net nodes to obtain net node QR code positioning information; calculating the cross-sectional positioning benchmark for the drainage ditch feature points to obtain drainage ditch cross-sectional positioning information; wherein, the appendage marking information includes: the anchor bolt marking information, the net node QR code positioning information, and the drainage ditch cross-sectional positioning information; based on the appendage category information and the appendage marking information, spatial information encoding processing is performed on the three-dimensional spatial coordinates of each appendage to obtain an appendage spatial index table; and based on the three-dimensional spatial coordinates, the spatial distance between each type of appendage is calculated to obtain a spatial adjacency table; wherein, the appendage spatial index table includes the unique code, type, spatial coordinates, marking method, and initial acquisition time of each appendage.

[0010] This application's embodiments establish a traceable and reproducible monitoring benchmark by uniformly coding and spatially calibrating the appendages, avoiding target confusion or positioning deviations during subsequent comparisons. Furthermore, the use of identification methods provides stable manual control points for subsequent multi-year, multi-batch image registration, significantly improving the accuracy of cross-period comparisons.

[0011] In one feasible implementation, based on the spatial information encoding of appendages in the ecological restoration project, the trace change characteristics of the current ecological restoration project are periodically collected and processed based on the geometric center point of each appendage, and a database of appendage trace information related to the ecological restoration project is generated, specifically including: Using drones, the current ecological restoration project is automatically photographed at each waypoint along a preset flight path, resulting in a sequence of images covering the entire restoration area. This sequence includes position data, attitude data, and timestamp data for each photo. Based on the position data and image feature points in the sequence, point cloud computing is performed on the position and attitude of the appendages to obtain a reconstructed sparse point cloud. Based on the reconstructed sparse point cloud and the corresponding high-quality dense point cloud, a digital surface model and orthophoto of the appendages are generated. According to the geometric center point of each appendage, the orthophoto is cropped into local image blocks. The pixel area of ​​the rust region in the local image block at the anchor bolt end is calculated, and the rust region is fitted with an ellipse to obtain the main direction of the rust patch based on the angle of the ellipse's major axis. The location of the rust patch is then identified. The image blocks at the ends of anchor bolts are used to determine the inclination angle of the anchor bolt ends. The offset vector between the center of the protective netting nodes and the reference point is calculated to obtain the node offset information. The side lengths and angles of the four mesh openings around the nodes are also obtained to determine the mesh deformation rate. The elevation change curve of the ditch wall profile is calculated for the local image blocks of the drainage ditch to determine the maximum scour depth. Based on the original design cross-section, the deposition thickness at each point on the bottom of the ditch in the local image blocks is determined. Based on the data acquisition cycle, the main direction of the rust patches, the inclination angle of the anchor bolt ends, the node offset information, the mesh deformation rate, the maximum scour depth, and the deposition thickness are periodically integrated to obtain evolutionary trace features. A database is constructed by combining the spatial index table of appendages, the evolutionary trace features of each appendage, and the original image blocks to generate an information database of appendage traces related to ecological restoration projects.

[0012] This application's embodiments utilize a data acquisition method combining UAVs and ground-based laser scanning, achieving both comprehensive coverage of large-scale slopes and detailed modeling of local attachments, resulting in a ground resolution of 0.5 cm. Furthermore, by quantitatively extracting trace features such as anchor bolt corrosion area, tilt angle, protective net displacement, and drainage ditch scour depth, qualitative surface phenomena are transformed into calculable and comparable numerical indicators, providing a data foundation for subsequent change detection.

[0013] In one feasible implementation, the trace change detection is performed on the changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour using the appendage trace information database to obtain significant change feature points. Specifically, this includes: comparing the currently stored aerial photography data with the original reference aerial photography data using the appendage trace information database; if the comparison result indicates the presence of change features, calculating the homography transformation matrix under relevant invariant feature transformation points for each appendage's local image block and the reference image block, and projecting the current appendage image onto the coordinate system of the appendage reference image using the homography transformation matrix to generate a difference image; based on the anchor bolt end state, performing RGB color threshold segmentation on the difference image, re-identifying the area of ​​newly added rusted areas and the area of ​​fading rusted areas to obtain the rust diameter change; and calculating the difference between the current rust patch's principal direction and the reference principal direction angle to obtain the rust principal direction change angle; based on the anchor bolt end inclination angle state, analyzing the difference image... The current point cloud data is used to calculate the change in tilt angle between the current end tilt angle and the reference end tilt angle; and based on the normal vector projection, the tilt direction is calculated for the change in tilt angle to obtain the tilt direction quantity; wherein, the tilt direction quantity includes: inside the slope and outside the slope; based on the displacement state of the protective net nodes, the QR code information in the difference image is identified and processed to obtain the current pixel coordinates; and the current pixel coordinates are compared with the reference pixel coordinates to obtain the converted net node displacement vector; the elevation change between the current drainage ditch cross-section positioning information and the reference drainage ditch cross-section positioning information in the difference image is identified, the scour deepening area and the deposition thickening area are identified, and the maximum scour depth increment and deposition thickness increment are calculated; the rust diameter change, the rust main direction change angle, the tilt angle change, the tilt direction quantity, the net node displacement vector, and the maximum scour depth increment and deposition thickness increment are normalized, and the appendages exceeding the preset quantization threshold range are identified as significant change feature points.

[0014] This application's embodiments achieve pixel-level alignment of the same appendage at different time points through periodic repeated acquisition and image registration techniques, enabling the detection of millimeter-level displacement and square-centimeter-level corrosion changes. Furthermore, the normalization processing of the change amounts eliminates the dimensional differences between different indicators, allowing cross-type data (such as tilt angle changes and displacement changes) to be correlated and analyzed at the same scale. Simultaneously, by utilizing a set "significant change point" threshold, environmental noise and measurement errors are automatically filtered out, reducing interference from invalid data.

[0015] In one feasible implementation, based on the spatial location of the appendages corresponding to the significant change feature points, the change vector field and the proximity graph are matched under association rules to output the association analysis results. Specifically, this includes: projecting the appendages corresponding to the significant change feature points onto a two-dimensional plane graph and marking the direction of change for each change amount to obtain a visualized change vector field; traversing all significant change feature points and, based on the derived spatial adjacency table, identifying the spatial neighbors between the appendages corresponding to the significant change feature points to establish the proximity graph; wherein, the nodes in the proximity graph are significant change feature points, and the edges are lines connecting spatial neighbors; based on the change vector field and the proximity graph, identifying each connected component in the proximity graph structure, and matching the node types and change features contained in the connected components with predefined association rules. The system performs a one-by-one comparison of data in the association rule library. The association rule library includes several pieces of information on the appearance of appendages and corresponding risk issues. If the node set of a connected component completely matches all node types in the association rule library and the change characteristics of each node conform to the direction or type specified by the rule, then the corresponding association rule is determined as the issue determination result of the connected component. The issue determination result includes at least: shallow sliding trend, loosening of protective netting, change in water catchment path, regional drying, and local loosening of anchor bolts. If the node set of a connected component matches multiple association rules, then the association rule with the highest priority is determined as the issue determination result of the connected component. Based on the issue determination result of each connected component, an association analysis report is generated, and the association analysis results are output. The association analysis results include: problem area number, a list of related appendage codes, a judgment conclusion, and a confidence level.

[0016] This application's embodiments transform isolated single-point changes into a spatially continuous change field by constructing an adjacency graph and a change vector field graph, revealing the inherent mechanical connections between points. Furthermore, predefined engineering logic rules (e.g., anchor bolt inclination angle change + lattice beam displacement → shallow sliding trend) replace complex machine learning models, achieving interpretable, low-computing-power, and easily maintainable intelligent diagnosis. Simultaneously, the rule matching algorithm performs pattern recognition based on connected components, automatically delineating problem region boundaries and avoiding the subjectivity of manual delineation.

[0017] In one feasible implementation, the problem-related data in the correlation analysis results are subjected to structured inversion state determination to obtain an inversion state determination table for the current ecological restoration project. Specifically, this includes: performing inversion state processing on each connected component of the problem-related data in the correlation analysis results to obtain the inversion state determination table, including: statistically analyzing the distribution of problem appendage types contained in the connected components; statistically summarizing the change indicators of each connected component to generate a quantitative summary result; calculating the spatial impact range of the problem area for each connected component; determining the risk level of the problem determination result based on the matching corresponding correlation rule number; and generating the structured inversion state determination table based on the distribution of problem appendage types, the quantitative summary result, the spatial impact range, and the risk level. The inversion state determination table is a set of multiple problem analysis results associated with the significant change feature points.

[0018] This application's embodiments avoid false alarms from single indicators and improve the reliability of diagnostic results by verifying and calculating confidence levels for each matching rule. Furthermore, it automatically assigns risk levels based on the judgment conclusions, enabling differentiated allocation of management resources and prioritizing high-risk areas. By generating a structured judgment table and storing historical states, it supports long-term tracking and trend analysis of problem areas, facilitating the evaluation of the effectiveness of management measures.

[0019] In one feasible implementation, based on the spatial influence range of the problem area in the inversion state determination table, the corresponding matching processing of management and maintenance measures for the problem area status is performed. Specifically, this includes: obtaining the spatial coordinates of all attachments within the spatial influence range of the problem area; calculating the minimum convex hull of the spatial coordinates; determining the vertex coordinates of the minimum convex hull as the regional boundary coordinates of the spatial influence range of the problem area; obtaining the regional centroid coordinates of the spatial influence range of the problem area based on the regional boundary coordinates; generating navigation target points for management and maintenance personnel based on the regional boundary coordinates and the regional centroid coordinates; performing risk type determination processing on the problem analysis results in the inversion state determination table to obtain risk determination conclusion data; querying a preset solution correspondence table based on the risk determination conclusion data to generate specific management and maintenance instructions; and performing line node calibration processing on the management and maintenance route for each risk type information corresponding to the attachments within the problem area based on the navigation target points to generate a structured management and maintenance instruction sheet. The structured management and maintenance instruction sheet includes: instruction number, generation time, problem area number, regional centroid coordinates, regional boundary coordinate list, problem determination conclusion, risk level, management and maintenance measure list, and response time limit.

[0020] This application's embodiments pinpoint the spatial extent of the problem area to meter-level polygons and output the centroid coordinates, providing maintenance personnel with a precise, navigable location and significantly reducing the on-site investigation scope. Furthermore, the maintenance measures employ a fixed format of "action + location + parameters," with specific and clear instructions, avoiding execution deviations caused by vague descriptions. Utilizing the characteristic that structured instruction sheets can be directly parsed by mobile terminals, automated connection from diagnosis to task assignment is achieved.

[0021] In one feasible implementation, a risk response strategy for the current ecological restoration project is generated based on a tiered response mechanism. Specifically, this includes: if the structured maintenance instruction sheet is an emergency maintenance instruction sheet, an emergency push message is generated via the application programming interface of a real-time communication tool; and the emergency push message is repeatedly pushed according to a countdown mechanism; if the structured maintenance instruction sheet is a high-risk maintenance instruction sheet, the corresponding structured maintenance instruction sheet is imported into the maintenance work order system, and a work order generation notification is issued; if the structured maintenance instruction sheet is a low-risk maintenance instruction sheet, the corresponding structured maintenance instruction sheet is packaged into a quarterly maintenance monitoring report, and the quarterly maintenance monitoring report is identified as a key monitoring target for the next data collection cycle.

[0022] In one feasible implementation, after the maintenance work is completed, new evolutionary trace features are extracted again; if the new evolutionary trace features do not reach the baseline evolutionary trace features and belong to the state of permanent engineering modification, then the new evolutionary trace features are used as the current baseline evolutionary trace features; the current baseline evolutionary trace features are uploaded to the appendage trace information database and updated to the new appendage trace information database.

[0023] On the other hand, this application also provides an intelligent monitoring system for the post-construction management of ecological restoration projects. The system is characterized in that it has instructions that can be executed by at least one processing module, so that the at least one processing module can execute the intelligent monitoring method for the post-construction management of ecological restoration projects described in any of the above embodiments.

[0024] This application provides an intelligent monitoring method and system for the post-construction management of ecological restoration projects. Compared with the prior art, the embodiments of this application have the following beneficial technical effects: 1. Intelligent monitoring methods can automatically collect and process changes in traces of ecological restoration projects, thereby improving monitoring efficiency and reducing labor costs.

[0025] 2. Periodic data collection and processing can enable real-time monitoring of ecological restoration projects, timely detection of potential problems, and prevention of damage to the ecological environment.

[0026] 3. Establishing a database of traces of attachments helps to accumulate long-term data, facilitating long-term tracking and research of ecological restoration projects.

[0027] 4. By detecting trace changes and matching association rules, the status of ecological restoration projects can be analyzed more accurately, thus improving the accuracy of the analysis results.

[0028] 5. Based on the inversion state determination table and hierarchical response mechanism, management and maintenance measures can be quickly formulated and implemented, improving the response speed to problem areas.

[0029] 6. By conducting risk assessments and developing response strategies, the risks of ecological restoration projects can be effectively reduced, ensuring the stability of the ecological environment.

[0030] 7. It helps to improve the management level of ecological restoration projects and achieve scientific and refined management. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart illustrating an intelligent monitoring method for post-construction management of ecological restoration projects, provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of an intelligent monitoring system for the post-construction management of an ecological restoration project, provided as an embodiment of this application. Detailed Implementation

[0032] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0033] This application provides an intelligent monitoring method for the post-construction management of ecological restoration projects, such as... Figure 1 As shown, the intelligent monitoring method for the later-stage management of ecological restoration projects specifically includes steps S101-S105: S101. Based on the spatial information coding of the appendages in the ecological restoration project, periodically collect and process the trace change characteristics of each appendage based on the geometric center point of the current ecological restoration project, and generate a database of appendage trace information related to the ecological restoration project.

[0034] Specifically, a real-time dynamic differential positioning device is first used to calibrate all visible engineering attachments in the restoration area corresponding to the current ecological restoration project, thereby obtaining attachment category information. Among them, visible engineering attachments include at least: anchor bolt ends, protective net nodes, drainage ditch feature points, and lattice beam nodes.

[0035] In one embodiment, an operator carries a handheld real-time dynamic differential positioning device into the repair area. The positioning accuracy of this device reaches the centimeter level (≤2cm). Then, all visible engineering attachments are individually calibrated, including anchor bolt ends, protective net nodes, drainage ditch feature points, lattice beam nodes, etc., thereby obtaining attachment category information through individual calibration.

[0036] Furthermore, each type of attachment in the location information of the marked attachments is marked with its own unique marking information to obtain attachment marking information. The specific execution process is as follows: 1) Perform color recognition processing on the anchor bolt end based on image registration points to obtain anchor bolt marking information; 2) Perform hardware-attached QR code information processing on the protective net nodes to obtain net node QR code positioning information; 3) Perform cross-sectional positioning benchmark calculation on the feature points of the drainage ditch to obtain drainage ditch cross-sectional positioning information. Among them, the attachment marking information includes: anchor bolt marking information, net node QR code positioning information, and drainage ditch cross-sectional positioning information.

[0037] In one embodiment, for the anchor bolt end, a 2cm diameter circular marker can be sprayed onto the center point of the anchor bolt end using red spray paint as a control point for subsequent image registration. For the protective net nodes, a uniquely numbered waterproof aluminum tag, 3cm x 5cm in size, can be hung at each node, with a QR code and numerical code engraved on it. For the drainage ditch, a 2mm deep cross mark is chiseled into the inner side of the ditch wall every 10m as a cross-sectional positioning reference.

[0038] Furthermore, based on the appendage category information and appendage labeling information, the three-dimensional spatial coordinates of each appendage need to be spatially encoded to obtain an appendage spatial index table. Then, based on the three-dimensional spatial coordinates, the spatial distance between each type of appendage is calculated to obtain a spatial adjacency table. The appendage spatial index table includes a unique code, type, spatial coordinates, labeling method, and initial data collection time for each appendage.

[0039] In one embodiment, the three-dimensional spatial coordinates (including X, Y, and Z coordinates) of each appendage collected by the real-time differential positioning device are entered into a database to form an appendage spatial index table. This index table records the unique code, type, spatial coordinates, marking method, and initial acquisition time of each appendage. Furthermore, the system can automatically generate two-dimensional planar distribution maps and three-dimensional spatial distribution maps of the appendages based on the spatial coordinates. On the distribution map, the system automatically calculates and marks the spatial distances between appendages, especially the distances between anchor bolts and adjacent lattice beam nodes, and between anchor bolts and adjacent drainage ditch sections, forming a spatial adjacency table that records information about other appendages within a 3-meter radius of each appendage.

[0040] It should be noted that the data collection equipment must first be configured. For the overall slope in the current ecological restoration project, a multi-rotor drone equipped with a high-resolution mechanical shutter camera can be used for aerial photography. The drone's forward overlap rate is set to 80%, the lateral overlap rate to 70%, and the flight altitude is set to 30m based on the slope distance to ensure a ground resolution of 0.5cm / pixel. For attachments obscured by vegetation or inaccessible to the drone (such as anchor bolts on steep rock faces), a handheld laser scanner is used to perform multi-station scanning from a safe location, with a scanning resolution set to achieve a 6mm dot pitch at a distance of 10m.

[0041] Furthermore, drones are used to automatically photograph the current ecological restoration project at each waypoint along a preset flight path, resulting in a sequence of images covering the entire restoration area. This sequence includes the location, attitude, and timestamp data for each photograph. Then, based on the location data and image feature points in the sequence, point cloud computing is performed on the location and attitude of the appendages to obtain a reconstructed sparse point cloud. Finally, based on the reconstructed sparse point cloud and the corresponding high-quality dense point cloud, a digital surface model and orthophoto of the appendages are generated.

[0042] In one embodiment, the drone first automatically flies along a preset route, triggering the camera to take pictures at each waypoint to acquire a sequence of images covering the entire restoration area. The system automatically records the position data, attitude data, and timestamp data of each photo. For each appendage, it is ensured that it is covered by at least three photos from different angles to support subsequent 3D reconstruction. Next, 3D point cloud reconstruction is performed, which involves importing the acquired sequence of images into 3D reconstruction software and performing the following processing flow: First, the position and attitude of the appendage are calculated based on the position data and image feature points to generate a sparse point cloud; then, a dense point cloud is generated, with the quality parameter set to high level; finally, a digital surface model and orthophoto are generated, with the pixel size of the orthophoto set to 0.5cm. For handheld laser scanning data, multi-station stitching is performed using the accompanying software, with the stitching error controlled within 3mm.

[0043] Furthermore, based on the geometric center point of each appendage, the orthophoto image is cropped into local image blocks. Then, the pixel area of ​​the rust region in the local image block of the anchor bolt end is calculated, and the rust region is subjected to ellipse fitting to obtain the main direction of the rust patch based on the angle of the major axis of the ellipse. The end tilt angle of the local image block of the anchor bolt end is also identified.

[0044] As a feasible implementation method, in the ellipse fitting process for the rusted area, it is first necessary to segment the binary image of the rusted area from the local image block of the anchor bolt end based on RGB color threshold or grayscale threshold; then, the Canny edge detection algorithm can be used to extract the edge pixel set of the rusted area; then, based on the above-mentioned least squares algebraic distance model, the generalized eigenvalue problem is solved to obtain the coefficients of the ellipse equation; and the center, major semi-axis, minor semi-axis and major axis direction angle of the ellipse are calculated according to the coefficients; finally, the main direction of the rusted patch is output with due north as the reference.

[0045] In one embodiment, the set of edge pixels of the rusted area extracted after image segmentation is: , where N is the total number of edge pixels. Ellipse fitting is performed using the least squares method based on algebraic distance. The general equation of the ellipse is: A, B, C, D, E, and F are coefficients to be determined, and they must satisfy the elliptic constraint condition B. 2 4AC < 0. To avoid zero solutions and to treat solutions that are integer multiples of the same ellipse as representations of the same ellipse, the parameters are constrained by the condition: A + C = 1; or the parameters are constrained using a constraint matrix, i.e.: ;in H is the constraint matrix. The objective function is then defined as the sum of squared algebraic distances from all edge pixels to the fitted ellipse: and order The optimization problem can then be expressed as: ;in, Then, based on the principle of extrema, we can take the partial derivatives of each coefficient of the objective function and set them to zero, transforming it into a generalized eigenvalue problem for solution. Where S=U T U is the divergence matrix. Next, we solve this generalized eigenvalue problem, taking the generalized eigenvector corresponding to the smallest positive generalized eigenvalue, which yields the optimal elliptic coefficients: .

[0046] In one embodiment, the optimal ellipse equation coefficients obtained from the above fitting are used to calculate the geometric parameters of the ellipse according to the following formula: (1) Ellipse center coordinates (x0, y0): and (2) The semi-major axis of the ellipse and short half shaft : and And agreed ≥ That is, the semi-major axis corresponds to a larger value. (3) Angle of the direction of the major axis of the ellipse (i.e., the main direction of the rust patches): Where arctan2 is the arctangent function in the fourth quadrant. Finally, taking true north as... Measured clockwise, the principal direction angle of the rust patch is... : ;when When it is negative, take Ensure the angle is Within the range.

[0047] In one embodiment, local image blocks can be cropped from the orthophoto, centered on each appendage. The cropping size for the anchor bolt end is 20×20 mm, and the cropping size for the protective net nodes and drainage ditch walls is 50×50 mm. For each local image block of the anchor bolt end, the rusted area needs to be identified first using image processing technology. An RGB color threshold is set, and pixels that meet the threshold condition are identified as rusted areas. The pixel area of ​​the rusted area is calculated and converted to the actual area according to the ground resolution. An ellipse fitting is performed on the rusted area, and the direction angle of the major axis of the ellipse is calculated. With true north as 0°, the main direction of the rusted patch is obtained by measuring clockwise. The normal vector of the anchor bolt end plane is extracted from the 3D point cloud, and the angle between this normal vector and the horizontal plane is calculated to obtain the end tilt angle.

[0048] Furthermore, it is necessary to calculate the offset vector between the node center and the reference point for the local image block of the protective net node to obtain the net node offset information, and obtain the side length and angle of the four mesh holes around the node to obtain the mesh deformation rate.

[0049] In one embodiment, for a protective mesh node, the center of the aluminum plate QR code can be used as a reference point. The offset vector of the node center relative to the reference point is calculated, and the offset direction and amount are recorded. The side lengths and angles of the four mesh openings around the node are measured, and the mesh deformation rate is calculated.

[0050] Furthermore, it is necessary to calculate the elevation change curve of the ditch wall profile from the local image block of the drainage ditch to obtain the maximum scour depth. Based on the original design cross-section, the deposition thickness at each point on the bottom of the drainage ditch within the local image block should also be determined.

[0051] In one embodiment, for the drainage ditch wall and bottom, the elevation change curve of the ditch wall profile is extracted from the 3D point cloud, and then the maximum scour depth is calculated. By comparing the current point cloud with the original design cross-section, the deposition thickness at each point on the ditch bottom is calculated.

[0052] Furthermore, based on the data acquisition cycle, the main direction of the rust patch, the tip angle, the mesh node offset information, the mesh deformation rate, the maximum scour depth, and the deposition thickness are periodically integrated to obtain the evolution trace characteristics.

[0053] Furthermore, a database is constructed by combining the spatial index table of appendages, the evolutionary trace characteristics of each appendage, and the original image blocks to generate an appendage trace information database for ecological restoration projects. That is, the extracted trace feature values, along with the original image blocks and 3D point cloud slices, are stored in the database to form a baseline record of the appendage's trace. Each baseline record includes the appendage code, acquisition time, acquisition equipment, environmental conditions (which may include temperature, humidity, and cumulative rainfall over the previous three days), and trace feature vectors (which may include specific values ​​such as corrosion area, main corrosion direction, end tilt angle, node offset, mesh deformation rate, scour depth, and deposition thickness).

[0054] S102. Using the appendage trace information database, detect the changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour to obtain significant change feature points.

[0055] Specifically, firstly, the aerial data currently collected by the appendages is compared with the original reference aerial data using the appendage trace information database. If the comparison result shows that there are changes, the homography transformation matrix of each appendage's local image block and the reference image block is calculated under the relevant invariant feature transformation points. Then, the current appendage image is projected onto the coordinate system of the appendage reference image using the homography transformation matrix to generate a difference image.

[0056] As a feasible implementation method, the data collection task can be automatically triggered at a preset cycle (e.g., quarterly) using an appendage trace information database. A data collection drone or ground equipment is then used to repeatedly collect data along the same flight path, at the same altitude, and with the same resolution as the initial collection, ensuring data consistency. After collection is complete, the system automatically registers the current data with the reference database data. If the registration result shows changes, for each appendage's local image block, the scale-invariant feature transformation points between the current image and the reference image need to be extracted first. Then, a random sampling consensus algorithm is used to filter matching point pairs, and a homography transformation matrix is ​​calculated. The current image is then projected onto the coordinate system of the reference image using this transformation matrix, achieving precise alignment of the two images. After alignment, the pixel-by-pixel grayscale difference between the two images is calculated, generating a difference image.

[0057] In one embodiment, the homography transformation matrix can be calculated using the Direct Linear Transform (DLT) algorithm. First, feature point pairs (N≥4) are obtained by extracting features from the current image and the reference image using Scale Invariant Feature Transform (SIFT) and then filtering them through random sampling consistency. The homogeneous coordinates of the feature points in the reference image are then denoted as (x...). i y i ,1), the homogeneous coordinates of the corresponding matching feature points in the current image are (x i ',y i ', 1). The homography transformation matrix H is a 3×3 matrix, and its form is: The matrix has 9 elements, but since the overall scale does not affect the transformation result, it is usually set to 0. =1 is used for normalization, therefore there are actually 8 unknown parameters to be solved. Then, according to the homography transformation relationship, the reference image feature points (x i y i ) and current image feature points (x i ',y i The following mapping relationship needs to be satisfied: x i '=( ·x i + ·y i + ) / ( ·x i + ·y i + ) and y i '=( ·x i + ·y i + ) / ( ·x i + ·y i + Then, by transforming and rearranging the two equations above, we can obtain information about the unknown parameters. to The linear equations are obtained. Each pair of matched feature points provides two linear equations, therefore at least four non-collinear pairs of matched points are needed to solve for all eight unknown parameters.

[0058] In one embodiment, based on the above mapping relationship and matching feature point pairs, and to improve computational accuracy and stability, more than four sets of matching feature point pairs can be used (e.g., N≥10). In this case, the number of equations must be greater than the number of unknowns to construct an overdetermined system of equations. Then, the least squares method is used to solve this overdetermined system of equations, i.e., minimizing the sum of squared reprojection errors of all matching points. In the specific solution, singular value decomposition (SVD) can be performed on the coefficient matrix formed by each set of matching point pairs. The right singular vector corresponding to the smallest singular value is taken as the solution to the system of equations, thus obtaining the eight unknown parameters of the homography transformation matrix H.

[0059] In one embodiment, to further improve the accuracy and robustness of the transformation matrix, the aforementioned DLT algorithm can be embedded within the iterative framework of the random sampling consensus algorithm. In each iteration, four sets of matching point pairs are randomly selected to calculate a candidate homography matrix. This matrix is ​​then used to project and verify all matching point pairs, and the number of interior points that satisfy the projection error threshold is counted. After multiple iterations, the candidate matrix with the largest number of interior points is selected, and the least squares solution is re-executed using all interior points to obtain the final homography transformation matrix.

[0060] Furthermore, based on the state of the anchor bolt end, RGB color thresholding is performed on the difference image to re-identify the areas of newly added rust and fading rust, thus obtaining the change in rust diameter. The difference between the angle between the current rust patch's principal direction and the reference principal direction is then calculated to obtain the change angle of the rust principal direction.

[0061] In one embodiment, for changes in anchor bolt end corrosion, RGB color thresholding is reapplied to the difference image to identify newly added and faded corrosion areas. The newly added and faded corrosion areas are calculated, and the difference between the two is the net change. Simultaneously, the difference between the angle between the current corrosion patch's principal direction and the reference principal direction is calculated to obtain the angle of change in the principal corrosion direction.

[0062] Furthermore, based on the anchor bolt end inclination angle, the change in inclination angle between the current end inclination angle and the reference end inclination angle is calculated from the current point cloud data in the difference image. Then, based on the normal vector projection, the tilting direction is calculated from the change in inclination angle, yielding the tilting direction. The tilting direction includes both in-slope and out-of-slope directions.

[0063] In one embodiment, for changes in the anchor bolt end inclination angle, the normal vector of the anchor bolt end plane can be re-extracted from the current point cloud, the current inclination angle can be calculated, and the difference between the current inclination angle and the reference inclination angle is the change in inclination angle. Then, based on the projection direction of the normal vector, it can be determined whether the change in inclination angle points inward or outward, thus determining the tilting direction.

[0064] Furthermore, based on the displacement state of the protective net nodes, the QR code information in the difference image is identified and processed to obtain the current pixel coordinates. The current pixel coordinates are then compared with the reference pixel coordinates to obtain the transformed net node displacement vector.

[0065] In one embodiment, for the displacement of the protective net node, it is necessary to identify the aluminum plate QR code in the current image, read its pixel coordinates, compare the coordinates with the coordinates in the reference image, calculate the pixel displacement, convert the pixel displacement into the actual displacement based on the ground resolution, and record the direction and magnitude of the displacement vector, which is the converted net node displacement vector.

[0066] Furthermore, it is also necessary to identify the elevation changes between the current drainage ditch cross-section positioning information and the reference drainage ditch cross-section positioning information in the difference image, identify the scour deepening area and the deposition thickening area, and calculate the maximum scour depth increment and deposition thickness increment.

[0067] In one embodiment, to assess changes in drainage ditch scour, it is necessary to compare the current drainage ditch cross-section positioning information in the current point cloud with the reference drainage ditch cross-section positioning information in the reference point cloud, and then calculate the elevation change of each profile. This process identifies areas of increased scour (elevation decrease) and areas of increased deposition (elevation increase), calculates the maximum scour depth increment and deposition thickness increment, and thus obtains the amount of change under drainage ditch scour.

[0068] Furthermore, the data of the rust diameter change, the rust main direction change angle, the tilt angle change, the tilt direction change, the network node displacement vector, and the maximum scour depth increment and deposition thickness increment are normalized, and the appendages that exceed the preset quantization threshold range are identified as significant change feature points.

[0069] In one implementation, the changes in each accessory also need to be normalized by type, and then uniformly mapped to a change index range of 0-100. Normalization can use a linear mapping method, using the minimum and maximum values ​​in the historical data of that type as boundaries to map the current change to that range. When the change index exceeds a preset threshold (e.g., 30), the accessory is finally marked as a significant change feature point.

[0070] S103. Based on the spatial location of the appendages corresponding to the significant change feature points, perform matching processing on the change vector field and the proximity relationship graph under the association rules, and output the association analysis results.

[0071] Specifically, the attachments corresponding to the points of significant change are first subjected to coordinate projection processing on a two-dimensional planar diagram, and the direction of change for each change is marked to obtain a visualized change vector field. That is, the attachments marked as points of significant change are projected onto a two-dimensional planar diagram according to their spatial coordinates, and the direction of change is indicated by an arrow at the position of each point (such as the direction of change of anchor bolt inclination angle, the direction of displacement of protective net, and the direction of corrosion expansion). The length of the arrow is proportional to the change index, forming a visualized change vector field diagram.

[0072] Furthermore, it is necessary to traverse all significant change feature points and, based on the derived spatial adjacency table, identify the spatial neighbor points between the corresponding attachments of the significant change feature points, and then construct a proximity graph. In the proximity graph, nodes represent significant change feature points, and edges are lines connecting spatial neighbors.

[0073] In one embodiment, it is also necessary to traverse all significant change feature points, identify the spatial neighbors (attachments ≤ 3m away) of each change point according to the spatial adjacency table established above, and establish a graph structure, where nodes are change points and edges are lines connecting spatial neighbors.

[0074] Furthermore, based on the change vector field and the proximity graph, each connected component in the graph structure is identified, and the node types and change characteristics contained in each connected component are compared one by one with a predefined association rule base. The association rule base includes several pieces of information on the appearance of appendages and their corresponding risk issues.

[0075] In one embodiment, predefined association rules are applied for pattern matching. Specifically, based on the change vector field and the proximity graph, node combinations in the graph structure are matched against a predefined association rule base. The association rule base is predefined in "IF-THEN" format and includes the following association rules: Rule 1: When three changes occur simultaneously—anchor bolt end inclination angle pointing outwards from the slope, anchor bolt end rust patches expanding downhill, and lattice beam nodes displacing outwards from the slope—it is determined to be a shallow sliding trend.

[0076] Rule 2: When three changes occur simultaneously—uniform displacement direction of the protective net nodes, no change in the inclination angle of the anchor bolt ends, and increased corrosion of the protective net nodes—the protective net is deemed to be loose.

[0077] Rule 3: When three changes occur simultaneously—deepening scouring of the drainage ditch wall, expansion of rust patches at the downstream anchor bolt ends toward the drainage ditch, and thickening of sediment at the bottom of the ditch—it is determined that the water catchment path has changed.

[0078] Rule 4: When both the moss cover boundary at the anchor bolt end recedes upwards and the area of ​​the rust patch at the anchor bolt end shrinks, it is determined to be regional aridification.

[0079] Rule 5: When the anchor bolt end tilt angle changes inward toward the slope, the lattice beam node has no displacement, and the protective net node has no displacement, it is determined that the anchor bolt is locally loose (not as a whole).

[0080] Furthermore, if the node set of a connected component completely matches all node types in the association rule base and the change characteristics of each node conform to the direction or type specified by the rule (i.e., the rule number mentioned above), then the corresponding association rule is determined as the problem determination result for the connected component. The problem determination result includes at least: shallow slippage trend, loosening of protective netting, change in water catchment path, regional drying, and local loosening of anchor bolts. If the node set of a connected component matches multiple association rules, then the association rule with the highest priority is determined as the problem determination result for the connected component.

[0081] As a feasible implementation method, the rule matching execution method described above can be as follows: For each connected component in the graph structure, extract the node types and change features contained in that component, and compare the node feature vector with the rule preconditions in the rule base one by one. If the set of nodes of the connected component completely matches the triggering condition of a certain rule, that is, it contains all the node types required by the rule and the change features of each node conform to the direction or type specified by the rule, then it is determined to be the conclusion corresponding to that rule, that is, the problem judgment result. If multiple rules are matched at the same time, the judgment conclusion with the highest priority is output according to the preset priority. If there is no matching rule, it is marked as an unidentified change pattern, which requires manual review.

[0082] Furthermore, based on the problem determination results for each connected component, a correlation analysis report is generated, and the final correlation analysis results are output. These results include: the problem region number, a list of related accessory codes, the judgment conclusion, and the confidence level.

[0083] As a feasible implementation method, based on the problem determination result of each connected component, the system outputs the association analysis result of each connected component, including the problem region number, the list of related accessory codes, the determination conclusion, and the confidence level. The confidence level is calculated based on the completeness of the matching rules: if all rule condition nodes exist and the direction of change is completely matched, the confidence level is 100%; if a non-critical node is missing, the confidence level is reduced accordingly.

[0084] S104. Perform structured inversion state determination on the problem-related data in the correlation analysis results to obtain the inversion state determination table of the current ecological restoration project.

[0085] Specifically, the inversion state processing is first performed on each connected component of the problem-related data in the association analysis results to obtain the inversion state judgment table, including: 1) statistically analyzing the distribution of problem appendage types contained in the connected components; 2) statistically summarizing the change indicators of each connected component to generate a quantitative summary result; 3) calculating the spatial influence range of the problem area of ​​each connected component; 4) determining the risk level of the problem judgment result according to the matching corresponding association rule number.

[0086] In one embodiment, the system receives the correlation analysis results output above, which are organized using connected components as basic units. Each connected component contains the following raw information: a unique identifier for the connected component, a list of appendage codes contained in the component, the spatial coordinates of each appendage, quantitative indicators of change for each appendage (change in dip angle, displacement vector, corrosion expansion area, scour depth, etc.), the rule number matched by the component, and the confidence level of the matching rule. The system parses the information of each connected component, extracting key fields for subsequent judgment. The parsing process includes: 1) Statistically analyze the distribution of attachment types contained in the connected component. For example, if the component contains 8 anchor bolt ends, 3 lattice beam nodes, and 1 protective net node, the statistical results are: 8 anchor bolt ends, 3 lattice beam nodes, and 1 protective net node.

[0087] 2) Summarize the change indicators for each type of attachment. For anchor bolt ends, summarize the maximum, minimum, and average values ​​of the change in inclination angle and the distribution of the change direction; for protective net nodes, summarize the maximum, average, and displacement direction consistency; for drainage ditch walls, summarize the maximum, average, and distribution range of scour deepening.

[0088] 3) Calculate the spatial extent of this component. Extract the spatial coordinates of all attachments within this component, calculate the vertex coordinates of the minimum convex hull, and simultaneously calculate the geometric centroid coordinates as the positioning reference for the problem region.

[0089] As a feasible implementation method, it is necessary to confirm the judgment conclusion and determine the risk level according to the matching rule number. For example, if the matching rule is one (shallow sliding trend), the system executes the following confirmation logic: verify whether the direction of change of the anchor bolt end inclination angle all points outward from the slope (the direction of change is within the range of 90°±30°); verify whether the direction of corrosion patch expansion is consistent with the slope direction (i.e., downhill); verify whether the displacement direction of the lattice beam node is consistent with the direction of change of the anchor bolt inclination angle; if all three items pass the verification, the judgment conclusion is confirmed as "shallow sliding trend" and the risk level is determined as "high"; if some verifications fail, the confidence level is reduced proportionally, and an explanation is added to the judgment conclusion, such as "shallow sliding trend" (confidence level reduced, anchor bolt inclination angle direction is inconsistent). For example, if the matching rule is 2 (protective netting relaxation), the system executes the following confirmation logic: verify the consistency of the displacement direction of the protective netting nodes (the included angle of displacement direction of each node ≤ 30°); verify whether the anchor bolt end has no change in tilt angle (the change in tilt angle ≤ 0.5°); verify whether the degree of corrosion of the protective netting nodes exceeds the threshold (the corrosion area increases by ≥ 10 cm² or the corrosion depth increases by ≥ 1 mm); if the verification passes, the conclusion is "protective netting relaxation", and the risk level is determined to be "medium".

[0090] Furthermore, based on the distribution of problem-related incidental types, quantitative summary results, spatial impact range, and risk level, a structured inversion state determination table is generated. This inversion state determination table is a set of analysis results for multiple problems associated with significantly changing feature points.

[0091] As a feasible implementation method, the above analysis results, judgment conclusions, risk levels, and quantitative change indicators are finally summarized to generate a structured inversion state judgment table. This inversion state judgment table adopts a two-dimensional table format, with each row corresponding to a connected component (i.e., a problem region), as shown in Table 1: Table 1 Inversion State Judgment Table

[0092] S105. Based on the spatial impact range of the problem area in the inversion state determination table, perform corresponding matching processing of management and protection measures for the problem area state, and generate a risk response strategy for the current ecological restoration project based on the hierarchical response mechanism.

[0093] Specifically, the first step is to obtain the spatial coordinates of all appendages within the spatial influence range of the problem area. Then, the minimum convex hull of these spatial coordinates is calculated. The vertex coordinates of the minimum convex hull are then determined as the boundary coordinates of the spatial influence range of the problem area. Based on these boundary coordinates, the centroid coordinates of the spatial influence range of the problem area are obtained.

[0094] Furthermore, navigation target points for maintenance personnel must be generated based on the regional boundary coordinates and the regional center of gravity coordinates; then, the problem analysis results in the inversion status judgment table are processed for risk type judgment to obtain risk judgment conclusion data; and based on the risk judgment conclusion data, the preset solution correspondence table is queried to generate specific maintenance instructions.

[0095] In one embodiment, based on the risk assessment conclusion data, the system first queries a preset solution mapping table, and then generates specific maintenance instructions. For example: 1) For high-risk issues identified as having a “shallow sliding tendency”, the management measures include: inserting inclinometer tubes at the centroid coordinates of the problem area, with a depth of 3m and an outer diameter of 70mm; manually reading inclinometer data once a week; stacking soil bags at the toe of the slope for counterweight, with soil bags weighing 50kg / bag, stacked 2m wide and 1.5m high; and installing vibrating wire anchor rod dynamometers at the ends of the anchor rods.

[0096] 2) For medium-risk issues identified as “loose protective netting”, the maintenance measures include: for nodes with a displacement greater than 5mm, use a torque wrench to retighten the anchor bolts, with the tightening torque set at 120N·m; for areas with a mesh deformation rate greater than 15%, cut and replace the damaged mesh panels, with the new mesh panels having 8mm steel wire and a mesh size of 10cm×10cm.

[0097] 3) For medium-risk issues identified as “change in water catchment path”, the maintenance measures include: repairing the scour-deepened section of the drainage ditch with M10 cement mortar, with a thickness of 3cm and a width the same as the ditch width; and adding a sedimentation tank with dimensions of 1m×1m×0.8m 5m upstream of the scour section.

[0098] 4) For low-risk issues identified as “regional aridification”, management measures include: installing drip irrigation tape in the problem area with a dripper spacing of 0.5m and a flow rate of 2L / h; setting the irrigation cycle to once a week, with each irrigation lasting 2 hours; recording the irrigation volume, and assessing the aridification improvement at the next comparison.

[0099] Furthermore, based on the navigation target point, the information on each risk type corresponding to the ancillary structures within the problem area is processed by marking the route nodes under the maintenance route to generate a structured maintenance instruction sheet. The structured maintenance instruction sheet includes: instruction number, generation time, problem area number, area centroid coordinates, area boundary coordinate list, problem judgment conclusion, risk level, maintenance measure list (each measure includes a sequence number, specific action, execution location, execution parameters), and response time limit.

[0100] Furthermore, if the structured management instruction sheet is an emergency management instruction sheet, an emergency push message is generated through the application programming interface of the instant messaging tool. The emergency push message is then repeatedly pushed based on a countdown timer mechanism.

[0101] Furthermore, if the structured maintenance instruction sheet is a high-risk maintenance instruction sheet, the corresponding structured maintenance instruction sheet will be imported into the maintenance work order system, and a work order generation notification will be issued.

[0102] Furthermore, if the structured maintenance instruction form is a low-risk maintenance instruction form, the corresponding structured maintenance instruction form will be packaged into a quarterly maintenance monitoring report, and the quarterly maintenance monitoring report will be identified as a key monitoring target in the next data collection cycle.

[0103] In one embodiment, for high-risk issues, the system can automatically push a message to the project manager and technical director via the application programming interface (API) of an instant messaging tool. The message includes a summary of the instruction and a screenshot of the problem area. Simultaneously, the system displays the high-risk task prominently on the homepage of the project management platform with a 48-hour countdown. If no "processed" confirmation is received within 48 hours, the system will send a reminder every hour. For medium-risk issues, the system imports the instruction into the maintenance work order system and automatically assigns it to the maintenance team responsible for the area. The maintenance team receives the work order notification on their mobile application and can view a navigation map of the problem area and specific operational parameters. The work order is valid for 7 days; if not completed by the deadline, it automatically escalates to high-risk. For low-risk issues, the system records the information in the quarterly monitoring report as a baseline for subsequent comparisons, without triggering an immediate work order. However, in the next data collection cycle, the system will focus on comparing whether the aridification index of the area has improved.

[0104] As a feasible implementation method, after the maintenance work is completed, new evolutionary trace features are extracted again. If the new evolutionary trace features do not reach the baseline evolutionary trace features and the project is in a permanent engineering modification state, then the new evolutionary trace features are used as the current baseline evolutionary trace features. After extracting the new evolutionary trace features again, the changes are calculated and the maintenance effect is evaluated. If the trace features recover to within ±10% of the baseline value after maintenance, the original baseline is maintained, and the current state is stored as the new current baseline evolutionary trace feature. If the trace features do not recover after maintenance but are confirmed to be a permanent engineering modification (such as the addition of concrete cladding), then the current state is set as the new baseline evolutionary trace feature, and the modification content is recorded in the annotation. If the trace features continue to deteriorate after maintenance, a manual on-site inspection process is triggered, and the engineer determines whether more in-depth engineering measures are needed. Finally, the current baseline evolutionary trace features are uploaded to the appendage trace information database and updated to the new appendage trace information database.

[0105] In addition, this application also provides an intelligent monitoring system for the post-construction management of ecological restoration projects, such as... Figure 2 As shown, the intelligent monitoring system for maintenance 200 specifically includes: The intelligent monitoring system for maintenance has instructions 202 that can be executed by at least one processing module 201, so that at least one processing module 201 can perform the following: Based on the spatial information coding of appendages in ecological restoration projects, the trace change characteristics of each appendage are periodically collected and processed based on the geometric center point of the current ecological restoration project, and a database of appendage trace information related to the ecological restoration project is generated. By using the appendage trace information database, the changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour were detected to obtain significant change feature points. Based on the spatial location of the appendages corresponding to the significant change feature points, the change vector field and the proximity relationship graph are matched under the association rules, and the association analysis results are output. The problem-related data in the correlation analysis results are structured and inverted to determine the state, resulting in the inversion state determination table for the current ecological restoration project. Based on the spatial impact range of the problem area in the inversion state determination table, corresponding management and protection measures are matched to the problem area status, and risk response strategies for the current ecological restoration project are generated based on the hierarchical response mechanism.

[0106] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0107] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0108] The above description is merely an embodiment of this application and is not intended to limit this application. For those skilled in the art, various modifications and variations can be made to the embodiments of this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the embodiments of this application should be included within the scope of the claims of this application.

Claims

1. An intelligent monitoring method for the post-construction management of ecological restoration projects, characterized in that, The method includes: Based on the spatial information coding of appendages in ecological restoration projects, the trace change characteristics of each appendage are periodically collected and processed based on the geometric center point of the current ecological restoration project, and a database of appendage trace information related to the ecological restoration project is generated. Using the aforementioned appendage trace information database, trace change detection is performed on the changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour, to obtain significant change feature points. Based on the spatial location of the appendages corresponding to the significantly changed feature points, the change vector field and the proximity relationship graph are matched according to association rules, and the association analysis results are output, specifically including: The appendages corresponding to the significant change feature points are subjected to coordinate projection processing of a two-dimensional planar graph, and the change trend direction of each change is marked to obtain the visualized change vector field. Traverse all significant change feature points, and based on the derived spatial adjacency table, identify the spatial neighbor points between the corresponding attachments of the significant change feature points, and establish the adjacency graph; wherein, the nodes in the adjacency graph are significant change feature points, and the edges are the lines connecting spatial neighbors; Based on the change vector field and the proximity graph, each connected component in the proximity graph structure is identified, and the node types and change features contained in the connected components are compared one by one with the data of the predefined association rule base; wherein, the association rule base includes several appendage appearance information and corresponding risk problem results; If the set of nodes of the connected component completely matches all node types in the association rule base, and the change characteristics of each node conform to the direction or type specified by the rule, then the corresponding association rule is determined as the problem judgment result of the connected component; wherein, the problem judgment result includes at least: shallow sliding trend, loosening of protective net, change of water catchment path, regional drying, and local loosening of anchor bolts; If the set of nodes of the connected component matches multiple association rules, then the association rule with the highest priority is determined as the problem determination result of the connected component. Based on the problem determination result for each connected component, a correlation analysis report is generated and the correlation analysis results are output; wherein, the correlation analysis results include: problem region number, a list of related accessory codes, judgment conclusion, and confidence level; The problem-related data in the correlation analysis results are subjected to structured inversion state determination to obtain the inversion state determination table of the current ecological restoration project; Based on the spatial impact range of the problem area in the inversion state determination table, the corresponding management and protection measures are matched for the problem area status, and a risk response strategy for the current ecological restoration project is generated based on the hierarchical response mechanism.

2. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 1, characterized in that, Before periodically collecting and processing the trace change characteristics of each appendage based on its geometric center point according to the spatial information encoding of appendages in the ecological restoration project, and generating a database of appendage trace information related to the ecological restoration project, the method further includes: Using a real-time dynamic differential positioning device, all visible engineering attachments in the restoration area corresponding to the current ecological restoration project are individually calibrated to obtain attachment category information; wherein, the visible engineering attachments include at least: anchor bolt ends, protective net nodes, drainage ditch feature points, and lattice beam nodes; Each appendage in the appendage location information is marked with its own unique marking information to obtain appendage marking information, including: Color recognition processing based on image registration points is performed on the anchor bolt end to obtain anchor bolt marking information; hardware-attached QR code information is performed on the protective net nodes to obtain net node QR code positioning information; cross-sectional positioning benchmark calculation is performed on the drainage ditch feature points to obtain drainage ditch cross-sectional positioning information. The attached marking information includes: the anchor bolt marking information, the network node QR code positioning information, and the drainage ditch cross-section positioning information; Based on the appendage category information and the appendage marking information, the three-dimensional spatial coordinates of each appendage are spatially encoded to obtain an appendage spatial index table; and based on the three-dimensional spatial coordinates, the spatial distance between each type of appendage is calculated to obtain a spatial adjacency table; wherein, the appendage spatial index table includes the unique code, type, spatial coordinates, marking method and initial acquisition time of each appendage.

3. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 1, characterized in that, Based on the spatial information coding of appendages in ecological restoration projects, the trace change characteristics of each appendage are periodically collected and processed according to the geometric center point of the current ecological restoration project, and a database of appendage trace information of the ecological restoration project is generated, specifically including: Using drones, the current ecological restoration project is automatically photographed at each waypoint along a preset flight path to obtain a sequence of image information covering the entire restoration area; wherein, the sequence of image information includes: position data, attitude data, and timestamp data for each photo; Based on the location data and image feature points in the sequence image information, point cloud computing is performed on the position and orientation of the appendage to obtain a reconstructed sparse point cloud; and based on the reconstructed sparse point cloud and the corresponding high-quality dense point cloud, a digital surface model and orthophoto of the appendage are generated. Based on the geometric center point of each appendage, the orthophoto is cropped into local image blocks; The pixel area of ​​the rust region in the local image block of the anchor bolt end is calculated, and the rust region is subjected to ellipse fitting to obtain the main direction of the rust patch based on the angle of the major axis of the ellipse; and the tip angle of the local image block of the anchor bolt end is identified. The offset vector between the node center and the reference point is calculated for the local image block of the protective net node to obtain the net node offset information, and the side length and angle of the four mesh holes around the node are obtained to obtain the mesh hole deformation rate. The elevation change curve of the ditch wall profile was calculated from the local image block of the drainage ditch to obtain the maximum scour depth; and the deposition thickness at each point on the bottom of the ditch was determined based on the original design section. Based on the data acquisition cycle, the main direction of the rust patch, the tip tilt angle, the mesh node offset information, the mesh deformation rate, the maximum scour depth, and the deposition thickness are periodically integrated into feature points to obtain evolution trace features; A database is constructed by combining the spatial index table of appendages, the evolutionary trace characteristics of each appendage, and the original image blocks to generate an information database of appendage traces related to ecological restoration projects.

4. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 3, characterized in that, Using the aforementioned appendage trace information database, trace change detection is performed on the changes in anchor bolt traces, anchor bolt end inclination angle, protective net node displacement, and drainage ditch scour to obtain significant change feature points, specifically including: The data collected by aerial photography and the original baseline aerial photography data are compared using the appendage trace information database. If the comparison result shows that there are changes, then the local image block of each appendage is compared with the reference image block to calculate the homography transformation matrix under the relevant invariant feature transformation points, and the current appendage image is projected onto the coordinate system of the appendage reference image through the homography transformation matrix to generate a difference image. Based on the state of the anchor bolt end, the difference image is segmented using RGB color thresholding to re-identify the area of ​​newly added rust and the area of ​​faded rust, thus obtaining the change in rust diameter; and the difference between the angle between the current rust patch's main direction and the reference main direction is calculated to obtain the change angle of the rust main direction. Based on the anchor bolt end tilt angle, the change in tilt angle between the current end tilt angle and the reference end tilt angle is calculated from the current point cloud data in the difference image; and based on the normal vector projection, the tilt direction is calculated from the change in tilt angle to obtain the tilt direction quantity; wherein, the tilt direction quantity includes: inside the slope and outside the slope; Based on the displacement state of the protective net nodes, the QR code information in the difference image is identified and processed to obtain the current pixel coordinates; and the current pixel coordinates are compared with the reference pixel coordinates to obtain the converted net node displacement vector. The elevation change between the current drainage ditch cross-section positioning information and the reference drainage ditch cross-section positioning information in the difference image is identified, the scour deepening area and the deposition thickening area are identified, and the maximum scour depth increment and deposition thickness increment are calculated. The data of the corrosion diameter change, corrosion main direction change angle, tilt angle change, tilt direction change, network node displacement vector, maximum scour depth increment and deposition thickness increment are normalized, and appendages exceeding the preset quantization threshold range are identified as significant change feature points.

5. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 1, characterized in that, The problem-related data in the correlation analysis results are subjected to structured inversion state determination to obtain the inversion state determination table for the current ecological restoration project, which specifically includes: Statistically analyze the distribution of problem appendage types contained in the connected components; The change indicators of each connected component are statistically summarized to generate a quantitative summary result. Calculate the spatial influence range of the problem region for each of the connected components; The risk level of the problem judgment result is determined based on the matching association rule number; Based on the distribution of the types of problem-related items, the quantitative summary results, the spatial impact range, and the risk level, a structured inversion state determination table is generated; wherein, the inversion state determination table is a set of multiple problem analysis results associated with the significant change feature points.

6. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 1, characterized in that, Based on the spatial influence range of the problem area in the inversion state determination table, corresponding matching processing of management and protection measures is performed on the problem area state, specifically including: Obtain the spatial coordinates of all attachments within the spatial influence range of the problem area; and calculate the minimum convex hull of the spatial coordinates. The vertex coordinates of the minimum convex hull are determined as the boundary coordinates of the spatial influence range of the problem region; and based on the boundary coordinates, the centroid coordinates of the spatial influence range of the problem region are obtained. Based on the area boundary coordinates and the area centroid coordinates, navigation target points are generated for maintenance personnel; The problem analysis results in the inversion state determination table are processed for risk type determination to obtain risk determination conclusion data; and based on the risk determination conclusion data, a preset solution correspondence table is queried to generate specific management instructions. Based on the navigation target point, the information of each risk type corresponding to the attachments in the problem area is processed by the line node marking under the maintenance route to generate a structured maintenance instruction sheet; wherein, the structured maintenance instruction sheet includes: instruction number, generation time, problem area number, area centroid coordinates, area boundary coordinate list, problem judgment conclusion, risk level, maintenance measures list and response time limit.

7. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 6, characterized in that, Based on a tiered response mechanism, a risk response strategy for the current ecological restoration project is generated, specifically including: If the structured management instruction sheet is an emergency management instruction sheet, an emergency push message is generated through the application programming interface of the instant messaging tool; and the emergency push message is repeatedly pushed according to the time countdown mechanism. If the structured maintenance instruction sheet is a high-risk maintenance instruction sheet, the corresponding structured maintenance instruction sheet will be imported into the maintenance work order system, and a work order generation notification will be issued. If the structured maintenance instruction sheet is a low-risk maintenance instruction sheet, the corresponding structured maintenance instruction sheet will be packaged into a quarterly maintenance monitoring report, and the quarterly maintenance monitoring report will be identified as a key monitoring target in the next data collection cycle.

8. The intelligent monitoring method for post-construction management of ecological restoration projects according to claim 1, characterized in that, After the maintenance work is completed, new evolutionary trace features are extracted again; If the new evolutionary trace feature does not reach the baseline evolutionary trace feature and belongs to the state of permanent engineering modification, then the new evolutionary trace feature shall be used as the current baseline evolutionary trace feature. The current baseline evolution trace features are uploaded to the appendage trace information database and updated to a new appendage trace information database.

9. An intelligent monitoring system for the post-construction management of ecological restoration projects, characterized in that, The intelligent monitoring system for ecological restoration has instructions that can be executed by at least one processing module, so that the at least one processing module can execute an intelligent monitoring method for the later management of an ecological restoration project according to any one of claims 1-8.

Images