Routing inspection path setting and routing inspection image recognition method and device based on dam area environmental hidden danger and storage medium

Abstract

The invention discloses an inspection path setting and inspection image recognition method and device based on dam area environmental hidden dangers and a storage medium, and relates to the technical field of path planning, and the method comprises the steps: determining an initial position point of the hidden dangers on a dam area three-dimensional environmental model, analyzing the diffusion effect of each hidden danger category, expanding the initial position point of the hidden dangers, and obtaining an inspection path setting result; the acquisition reliability of subsequent unmanned aerial vehicle inspection images is facilitated, and relatively accurate position information is provided for subsequent path planning and image recognition. A plurality of selectable paths are evaluated based on flight information, and the most suitable target path is screened out by considering the topographic characteristics of the hidden danger area and the flight condition of the unmanned aerial vehicle. According to the method, different hidden danger areas are subjected to hidden danger feature association, so that the reasonability and adaptability of routing inspection path planning are improved, routing inspection image recognition is more accurate, the requirement for hidden danger routing inspection in a complex dam area environment is met, and timely and safe dam area environmental hidden danger monitoring is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and in particular to a method, device and storage medium for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas. Background Technology

[0002] With the expansion of water conservancy projects and their increasing operational lifespan, the need for monitoring environmental hazards in dam areas (such as cracks, seepage, and surface spalling) is becoming increasingly urgent. Traditional manual inspections suffer from low efficiency, incomplete coverage, and strong subjectivity, making them unsuitable for the dynamic monitoring needs of complex terrain and high-risk areas. In recent years, drone inspections and intelligent image recognition technologies have provided new pathways for dam area hazard investigation. Drones possess flexible deployment and high-resolution imaging capabilities, enabling rapid coverage of high-risk areas such as dam bodies and slopes; while deep learning-based image recognition algorithms (such as YOLO and Mask R-CNN) can automate the detection and feature extraction of hazards such as cracks and seepage. However, existing technologies often focus on identifying single hazards, lacking analysis of spatial correlations, leading to insufficient inference of hazard causes. Furthermore, inspection route planning does not fully consider the hazard risk level and terrain complexity, easily resulting in resource waste or missed detections. Therefore, it is necessary to design dynamic inspection routes based on the spatial distribution characteristics and risk levels of hazards in dam areas, and integrate multimodal image recognition technologies to achieve accurate hazard location, correlation analysis, and risk warning, providing intelligent support for the safe operation and maintenance of dam areas.

[0003] In existing technologies, the planning of drone inspection paths does not take into account the geographical characteristics of the potential hazard area or the drone's flight conditions, resulting in unreasonable inspection path planning and poor adaptability, which leads to low accuracy in inspection image recognition.

[0004] Therefore, how to improve the rationality and adaptability of inspection path planning is a technical problem that needs to be solved. Summary of the Invention

[0005] The purpose of this invention is to address the problems of unreasonable inspection path planning and low adaptability in existing technologies, and to propose a method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, comprising:

[0006] Obtain geographic information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area;

[0007] The initial location points of potential hazards were determined on the three-dimensional environmental model of the dam area. The diffusion effect of each hazard category was analyzed, and the initial location points of the potential hazards were expanded to obtain the location range of the potential hazards.

[0008] Obtain basic information about the inspection drone and match its flight information within the area of ​​potential hazards.

[0009] Based on the location and area of ​​all potential hazards, multiple optional paths for inspection drones are planned. The multiple optional paths are evaluated based on flight information, and the target path is selected. The inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard, thereby collecting inspection images of the potential hazard area.

[0010] Extract hazard features from inspection images of hazard areas, associate hazard features between different hazard areas, and output hazard results.

[0011] In some embodiments of this application, the set of all potential hazard categories involved in the dam area is statistically analyzed, including:

[0012] Accidents that have occurred in the dam area in the past are recorded as potential hazards, and these hazards are classified. The attributes and locations of the accidents are marked on the three-dimensional environmental model of the dam area.

[0013] Accidents that have never occurred in the history of the dam area are recorded as potential hazards that have not yet occurred. These potential hazards are then classified, and the union of the categories of both actual and potential hazards is taken as the set of all hazard categories.

[0014] In some embodiments of this application, the initial location points of potential hazards are determined on a three-dimensional environmental model of the dam area, the diffusion effect of each hazard category is analyzed, and the initial location points of the potential hazards are expanded, including...

[0015] Accident risk analysis is performed on a three-dimensional environmental model of the dam area to obtain a risk probability heat map of the dam area. The risk probability of the existing hidden dangers in the risk probability heat map of the dam area is adjusted according to the attributes and locations of the existing hidden dangers.

[0016] The initial location points and categories of potential hazards are determined based on the risk probability heat map of the dam area;

[0017] Based on the three-dimensional environmental model of the dam area, a multiphysics simulation model of the dam area is established to simulate the diffusion of each type of hidden danger and the diffusion direction of the initial location point of each hidden danger.

[0018] Based on the spread of each type of hazard, determine the spread scale range and condition standards corresponding to each type of hazard. Calculate the condition fit of the initial location point of each hazard based on the condition standards of each type of hazard. Select a spread scale within the spread scale range based on the condition fit of the initial location point of the hazard. Expand the region of the initial location point of the hazard based on the spread direction and spread scale of the initial location point of each hazard.

[0019] In some embodiments of this application, the flight information of the inspection drone within the location area of ​​the potential hazard is matched, including,

[0020] The basic information of the inspection drone includes the drone's structural information and image acquisition equipment information;

[0021] The flight information of the inspection drone within the location area of ​​each potential hazard is determined based on the influence relationship between the drone's structural information, image acquisition equipment information, and the location area of ​​the potential hazard.

[0022] In some embodiments of this application, multiple selectable paths for the inspection drone are planned based on the location area of ​​all potential hazards, including...

[0023] The starting and ending points of the inspection drone are determined, and the location area of ​​each potential hazard is used as a waypoint. Based on the starting point, multiple waypoints and the ending point, the path planning algorithm is used to plan the path of the inspection drone and generate multiple optional paths.

[0024] In some embodiments of this application, multiple alternative paths are evaluated based on flight information to select a target path, including...

[0025] The risk situation and terrain complexity of the location area of ​​the hidden danger are statistically analyzed to obtain risk indicators and terrain complexity indicators. Based on the risk indicators and terrain complexity indicators, a composite indicator of the location area of ​​a single hidden danger is determined, thereby determining the inspection level of each optional path.

[0026] The flight information is marked at the corresponding waypoints on each optional path, and the evaluation information on each optional path is analyzed. The evaluation information includes inspection task information, inspection flight loss information, and inspection wasted effort information.

[0027] Different inspection levels correspond to different combinations of the influence weights of inspection task information and inspection flight loss information. Combined with the inspection waste information, an evaluation index is obtained for each optional path. The target path is then selected based on the evaluation index.

[0028] In some embodiments of this application, different hazard areas are associated with hazard characteristics, and hazard results are output, including:

[0029] A comprehensive analysis of the hazard characteristics under each hazard area yields the hazard risk of each individual hazard area;

[0030] Spatial correlation analysis was conducted on the hazard characteristics and risks of different hazard areas to obtain the hazard risk correlation of multiple hazard areas.

[0031] Establish a hazard correlation map of the dam area to output hazard results.

[0032] Correspondingly, this application also provides a device for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, including,

[0033] The first module is used to acquire geographical information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area.

[0034] The second module is used to determine the initial location points of potential hazards on the three-dimensional environmental model of the dam area, analyze the diffusion effect of each hazard category, expand the initial location points of potential hazards, and obtain the location range of potential hazards.

[0035] The third module is used to obtain the basic information of the inspection drone and match the flight information of the inspection drone in the area of ​​the potential hazard.

[0036] The fourth module is used to plan multiple optional paths for the inspection drone based on the location and area of ​​all potential hazards. It evaluates the multiple optional paths based on flight information, selects the target path, and the inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard to collect inspection images of the potential hazard area.

[0037] The fifth module is used to extract hazard features from inspection images of hazard areas, associate hazard features between different hazard areas, and output hazard results.

[0038] This application also provides a computer-readable storage medium that stores one or more programs, which, when executed by an electronic device including multiple applications, cause the electronic device to perform the methods described above.

[0039] Compared with the prior art, the beneficial effects of this invention are as follows:

[0040] 1. Initial location points of potential hazards are determined on the 3D environmental model of the dam area. The diffusion effect of each hazard category is analyzed, and the initial location points are expanded, taking into account the diffusion of hazards. This expansion facilitates the reliability of subsequent UAV inspection image acquisition and provides more accurate location information for subsequent path planning and image recognition. The flight information of the inspection UAV within the hazard location area is matched, and reasonable flight information is configured according to the different characteristics or terrain conditions of each hazard area to ensure the stability and image quality of the UAV inspection images.

[0041] 2. Based on flight information, multiple optional paths are evaluated to select the target path. The terrain characteristics of the potential hazard area and the flight status of the UAV are considered to select the most suitable target path. Furthermore, the characteristics of different potential hazard areas are correlated, improving the rationality and adaptability of inspection path planning, making inspection image recognition more accurate, meeting the needs of potential hazard inspection in the complex environment of the dam area, and ensuring the timeliness and safety of environmental hazard monitoring in the dam area. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating a method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, as proposed in this invention.

[0043] Figure 2 This is a schematic diagram of the structure of an inspection path setting and inspection image recognition device based on potential environmental hazards in dam areas, as proposed in this invention. Detailed Implementation

[0044] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0045] Reference Figure 1 A method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, comprising the following steps:

[0046] Step S101: Obtain geographical information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area.

[0047] In some embodiments of this application, the set of all potential hazard categories involved in the dam area is statistically analyzed, including:

[0048] Accidents that have occurred in the dam area in the past are recorded as potential hazards, and these hazards are classified. The attributes and locations of the accidents are marked on the three-dimensional environmental model of the dam area.

[0049] Accidents that have never occurred in the history of the dam area are recorded as potential hazards that have not yet occurred. These potential hazards are then classified, and the union of the categories of both actual and potential hazards is taken as the set of all hazard categories.

[0050] In this embodiment, LiDAR scanning is used: an airborne LiDAR (such as Riegl VUX-1) is used to acquire high-precision point cloud data with a resolution of 0.1m, which is used to extract topographic elevation (DEM) and dam structure (such as micro-deformation of concrete surface).

[0051] GIS data integration: Import images and point clouds into ContextCapture or ArcGIS Pro to generate a 3D model of the dam area, including the dam body, spillway, bank slopes and surrounding environment (such as rivers and vegetation).

[0052] 3D model example:

[0053] A 3D model of a reservoir dam area shows that the dam body is 500m long and 80m high, including the location of historical cracks (coordinates: X=250m, Y=30m, Z=75m) and the layout of drainage pipes.

[0054] Hazard classification and labeling

[0055] Data source:

[0056] Historical accident reports (such as the crack in the middle of the dam in 2018 and the seepage on the right bank slope in 2020).

[0057] Hazard attributes: type (cracks, leaks), size (length, width), degree of material deterioration (e.g., depth of concrete carbonation).

[0058] Classification and labeling:

[0059] Cracks: Classified by their cause into shrinkage cracks and settlement cracks, which are marked on the 3D model (e.g., the red polygon represents a crack from 2018, with attributes: length 3m and width 0.05m).

[0060] Leakage: Classified by location into dam body leakage and foundation leakage, marked as blue areas (e.g., leakage point in 2020, flow rate 0.5L / s).

[0061] Example:

[0062] The following are the identified potential hazards: {cracks (shrinkage), cracks (settlement), seepage (dam body), seepage (foundation)}.

[0063] Prediction and classification of no hidden dangers

[0064] Prediction methods:

[0065] Expert experience: Refer to the "Safety Appraisal Code for Hydraulic Structures" to supplement potential hazards (such as concrete carbonization and metal structure corrosion).

[0066] Numerical simulation: Using COMSOL to simulate extreme working conditions (such as a sudden drop in water level) to predict potential hazards (such as slope landslides).

[0067] Category example:

[0068] No hazards have occurred: {concrete carbonization, metal corrosion, landslide}.

[0069] Union of hazard categories

[0070] Final collection:

[0071] Existing hazards ∪ Non-existent hazards = {cracks (shrinkage), cracks (settlement), seepage (dam body), seepage (foundation), concrete carbonization, metal corrosion, landslide}.

[0072] Application scenarios:

[0073] In the 3D model, cracks are marked in red, leaks in blue, carbonization / corrosion in yellow, and landslide risk areas are shaded in orange, providing a target list for subsequent inspections.

[0074] Step S102: Determine the initial location of the hidden danger on the three-dimensional environmental model of the dam area, analyze the diffusion effect of each type of hidden danger, expand the initial location of the hidden danger, and obtain the location range of the hidden danger.

[0075] In this embodiment, the accident risk situation on the 3D environmental model of the dam area is considered to determine the possible hidden dangers and their initial locations. To ensure the reliability of UAV inspection image acquisition, the area of ​​this initial location is expanded according to the previous spread of the hidden dangers, thereby improving the stability of UAV inspection image acquisition.

[0076] In some embodiments of this application, the initial location points of potential hazards are determined on a three-dimensional environmental model of the dam area, the diffusion effect of each hazard category is analyzed, and the initial location points of the potential hazards are expanded, including...

[0077] Accident risk analysis is performed on a three-dimensional environmental model of the dam area to obtain a risk probability heat map of the dam area. The risk probability of the existing hidden dangers in the risk probability heat map of the dam area is adjusted according to the attributes and locations of the existing hidden dangers.

[0078] The initial location points and categories of potential hazards are determined based on the risk probability heat map of the dam area;

[0079] Based on the three-dimensional environmental model of the dam area, a multiphysics simulation model of the dam area is established to simulate the diffusion of each type of hidden danger and the diffusion direction of the initial location point of each hidden danger.

[0080] Based on the spread of each type of hazard, determine the spread scale range and condition standards corresponding to each type of hazard. Calculate the condition fit of the initial location point of each hazard based on the condition standards of each type of hazard. Select a spread scale within the spread scale range based on the condition fit of the initial location point of the hazard. Expand the region of the initial location point of the hazard based on the spread direction and spread scale of the initial location point of each hazard.

[0081] In this embodiment, the input data includes: historical hazard records (type, location, and time of occurrence), environmental data (rainfall and water level changes), and engineering data (dam materials and structure).

[0082] Method: Use machine learning (such as random forest) to generate an initial risk probability heatmap, where pixel values ​​represent risk probabilities (0-1).

[0083] Risk adjustments have been made:

[0084] For existing hidden dangers (such as cracks in 2018), the risk value of the corresponding location is increased based on its attributes (such as crack length of 3m and leakage flow rate of 0.5L / s).

[0085] Determination of the initial location and category of potential hazards

[0086] step:

[0087] High-risk area screening:

[0088] Set a risk threshold (e.g., >0.7) to filter out the initial location points of potential hazards.

[0089] Hazard category determination:

[0090] The hazard category is determined by combining the annotation information (such as cracks and leaks) in the 3D model and the risk type (such as seepage and stress concentration).

[0091] Example:

[0092] The risk value of the middle part of the dam body (X=250m, Y=30m, Z=75m) is 0.7, which is judged as a potential crack.

[0093] The right bank slope (X=400m, Y=50m, Z=60m) has a risk value of 0.75 and is identified as a potential landslide hazard.

[0094] Multiphysics Simulation and Diffusion Analysis

[0095] step:

[0096] Model building:

[0097] Based on the three-dimensional model, a coupled model of seepage field (FLAC3D) and stress field (ABAQUS) is established.

[0098] Diffusion simulation:

[0099] Crack propagation: extending along the direction of maximum principal stress, simulating the propagation length under different seepage rates.

[0100] Leakage diffusion: Simulate the distance the wetting front travels along the seepage path.

[0101] Landslide propagation: Simulating the extent of the sliding surface based on slope stability analysis (such as the Bishop method).

[0102] Example:

[0103] Cracks: At a seepage rate of 0.15 m / s, they spread 3 m along the stress direction;

[0104] Landslide: When the rainfall intensity is 50 mm / h, the sliding surface extends to the toe of the slope.

[0105] Diffusion Scale Range and Condition Criteria

[0106] step:

[0107] Definition of diffusion scale range:

[0108] Cracks: 1-3m along the stress direction, 0.5-1m perpendicular to the stress direction;

[0109] Landslide: The sliding surface depth is 0.5-2m, and the horizontal extension is 5-15m.

[0110] Condition standard formulation (the conditions or characteristics of each type of hazard simulation process):

[0111] Crack propagation conditions: When the seepage rate is >0.1 m / s, the upper limit of the diffusion scale is +50%;

[0112] Landslide diffusion conditions: When pore water pressure > 10 kPa, the sliding surface depth increases by 30%.

[0113] Example:

[0114] If the seepage rate is 0.15 m / s, the crack propagation scale is 3 m (upper limit) × 1.2 m (vertical).

[0115] If the pore water pressure is 12 kPa, the depth of the landslide sliding surface is 2 m × 1.3 = 2.6 m.

[0116] The condition fit of the initial location point of each hazard is calculated based on the condition criteria for each hazard category. The initial location point of each hazard is then matched against the condition criteria (similarity calculation) to determine a condition fit. A reasonable diffusion scale is then selected for expansion. The diffusion scale is chosen based on the fit, and the hazard region (rectangular / elliptical) is generated by combining the diffusion direction.

[0117] It is understandable that the areas with potential hazards mentioned above are areas in the dam area environment where no accidents have occurred yet, but where accidents are very likely to occur.

[0118] Step S103: Obtain the basic information of the inspection drone and match the flight information of the inspection drone in the area of ​​the potential hazard.

[0119] In some embodiments of this application, the flight information of the inspection drone within the location area of ​​the potential hazard is matched, including,

[0120] The basic information of the inspection drone includes the drone's structural information and image acquisition equipment information;

[0121] The flight information of the inspection drone within the location area of ​​each potential hazard is determined based on the influence relationship between the drone's structural information, image acquisition equipment information, and the location area of ​​the potential hazard.

[0122] In this embodiment, the basic information of the UAV is acquired.

[0123] UAV structural information:

[0124] The dimensions of the drone are: length × width × height (e.g., DJIM300RTK: 810 × 670 × 430 mm), used to determine whether the drone can enter narrow areas (such as dam crack monitoring channels).

[0125] Maximum payload: 6kg, ensuring it can carry high-definition cameras (such as Zenmuse H20T, weighing 2.3kg) and lidar (such as L1, weighing 1.2kg).

[0126] Flight duration: e.g., 55 minutes, with flight time planned based on the area of ​​the potential hazard zone.

[0127] Image acquisition device information:

[0128] Camera resolution: such as 20MP (Zenmuse H20T), used to assess whether the image clarity meets the requirements for hazard identification (e.g., crack width ≥ 0.05mm).

[0129] LiDAR point cloud density: e.g., 240,000 points / m 2 (L1) is used to generate high-precision three-dimensional models (such as dam surface deformation accuracy ±2mm).

[0130] Sensor field of view: such as a thermal imaging camera with a FOV of 69°×42°, ensuring coverage of potential hazard areas.

[0131] Matching relationship between the location and area of ​​potential hazards and drone information

[0132] Matching logic:

[0133] Region size constraints:

[0134] If the potential hazard area is a rectangular crack measuring 3m in length and 1.2m in width, the drone must meet the following requirements:

[0135] The fuselage width is less than 1.2m (to avoid collisions);

[0136] The camera's field of view covers a length of 3m (if the flight altitude is 10m, the field of view must be ≥70°).

[0137] Load capacity constraints:

[0138] If both a camera and a lidar are required, the total payload must be ≤6kg (M300RTK meets this requirement).

[0139] Range and flight path:

[0140] The perimeter of the crack area is 8.4m. If data needs to be collected at 0.5m intervals, the required flight path length is 16.8m (round trip). Considering the camera frame rate (e.g., 10fps) and the drone speed (e.g., 2m / s), the single flight time is calculated to be ≤8.4s, and the total mission time must be ≤ the endurance time (55 minutes).

[0141] Camera perspective constraints:

[0142] Calculate the UAV attitude angle based on the camera field of view (FOV) and the size of the potential hazard area (e.g., the area is 3m long × 1.2m wide).

[0143] Posture adjustment:

[0144] If the area is not rectangular (such as a landslide fan-shaped area), use multi-angle circling flight (pitch angle ±30°).

[0145] Flight altitude determined

[0146] method:

[0147] Resolution requirements:

[0148] The minimum crack width detected by visible light is 0.05m. The flight altitude is calculated based on the camera resolution (20MP, pixel size 2.4μm).

[0149] Flight speed determined

[0150] method:

[0151] Image overlap ratio requirements:

[0152] Ensure that the overlap rate between adjacent photos is ≥70%, and calculate the flight speed.

[0153] The required flight information for inspection drones varies depending on the location and area of ​​different potential hazards. This flight information includes attitude, altitude, speed, etc., and it will affect the quality of the inspection images.

[0154] Step S104: Based on the location and area of ​​all potential hazards, multiple optional paths for the inspection drone are planned. The multiple optional paths are evaluated based on flight information, and the target path is selected. The inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard, thereby collecting inspection images of the potential hazard area.

[0155] In this embodiment, the path planning is carried out using the potential hazard area as the waypoint, resulting in multiple optional paths. The image acquisition and flight conditions of the drone during the inspection are taken into account to comprehensively evaluate each path and select the most reasonable path.

[0156] In some embodiments of this application, multiple selectable paths for the inspection drone are planned based on the location area of ​​all potential hazards, including...

[0157] The starting and ending points of the inspection drone are determined, and the location area of ​​each potential hazard is used as a waypoint. Based on the starting point, multiple waypoints and the ending point, the path planning algorithm is used to plan the path of the inspection drone and generate multiple optional paths.

[0158] In this embodiment, the data for the hazardous area is as follows:

[0159] Area with potential cracks: Center point (250, 30, 20), range 3m × 1.2m.

[0160] Landslide hazard area: center point (400, 50, 15), radius 15m.

[0161] Drone start and end point settings:

[0162] Starting point: UAV hangar in the northeast corner of the dam area (coordinates: 500, 100, 25).

[0163] Destination: Same as the starting point (returning home to recharge).

[0164] algorithm:

[0165] A* algorithm: used for global path planning, taking into account terrain elevation and obstacles.

[0166] RRT* (Rapid Exploration Random Tree): Used for dynamic obstacle avoidance in complex environments.

[0167] In some embodiments of this application, multiple alternative paths are evaluated based on flight information to select a target path, including...

[0168] The risk situation and terrain complexity of the location area of ​​the hidden danger are statistically analyzed to obtain risk indicators and terrain complexity indicators. Based on the risk indicators and terrain complexity indicators, a composite indicator of the location area of ​​a single hidden danger is determined, thereby determining the inspection level of each optional path.

[0169] The flight information is marked at the corresponding waypoints on each optional path, and the evaluation information on each optional path is analyzed. The evaluation information includes inspection task information, inspection flight loss information, and inspection wasted effort information.

[0170] Different inspection levels correspond to different combinations of the influence weights of inspection task information and inspection flight loss information. Combined with the inspection waste information, an evaluation index is obtained for each optional path. The target path is then selected based on the evaluation index.

[0171] In this embodiment, the formula for calculating the inspection level of the selectable path is as follows:

[0172]

[0173] in, Let n be the inspection level of the i1th optional path, and n be the number of potential hazard locations along the i1th optional path. The risk indicator for the location area of ​​the i2th hazard on the i1th optional path (determined by analyzing the probability of an accident occurring and the severity of the hazard in that area). The terrain complexity index (determined by terrain undulation, obstacle density, flight difficulty, etc.) is the area of ​​the location of the i2th hazard on the i1th optional path. for The maximum value in, These are the first and second constants for the i1th optional path, respectively. This represents the correction of the maximum value to the average value. The first constant is to balance the size of the correction function, and the second constant is to balance the size of the inspection level.

[0174] To reasonably assign influence weights to inspection mission information and inspection flight loss information, a composite index is defined for measurement, including risk index and terrain complexity index. Different inspection levels correspond to different combinations of influence weights for inspection mission information and inspection flight loss information.

[0175] The formula for calculating the evaluation index of the optional path is as follows:

[0176]

[0177] in, The evaluation metric for the j1th optional path is... Let m1, m2, and m3 be the respective influence weights of the inspection task information and the inspection flight loss information for the j1th optional path. m1, m2, and m3 are the number of parameters for the inspection task information (direct parameters related to the task, such as the location and coverage of potential hazards, and image quality), the inspection flight loss information (flight distance, flight energy consumption, and other flight losses), and the inspection waste information (redundant operation parameters such as redundant coverage counts, invalid flight distances, and attitude adjustment counts). These information are then standardized. These are the combined weights of the information from the j2nd inspection task, the flight loss information from the j3rd inspection, and the wasted energy information from the j4th inspection, respectively. These represent the parameter values ​​for the j2nd inspection task information, the j3rd inspection flight loss information, and the j4th inspection wasted energy information under the j1st optional path. Let j1 be the third constant of the optional path. This represents the correction of the sum of inspection task information and inspection flight loss information by the inspection waste information. The third constant is to balance the magnitude of the correction function. The inspection waste information can indirectly adjust the sum of inspection task information and inspection flight loss information.

[0178] Step S105: Extract the hazard features from the inspection image of the hazard area, associate the hazard features of different hazard areas, and output the hazard results.

[0179] In some embodiments of this application, different hazard areas are associated with hazard characteristics, and hazard results are output, including:

[0180] A comprehensive analysis of the hazard characteristics under each hazard area yields the hazard risk of each individual hazard area;

[0181] Spatial correlation analysis was conducted on the hazard characteristics and risks of different hazard areas to obtain the hazard risk correlation of multiple hazard areas.

[0182] Establish a hazard correlation map of the dam area to output hazard results.

[0183] In this embodiment, the feature extraction of hidden dangers

[0184] step:

[0185] Denoising: Use Gaussian filtering or median filtering to eliminate image noise.

[0186] Enhance contrast: Improve the contrast between the hazard area and the background through histogram equalization or the CLAHE algorithm.

[0187] Image segmentation: Semantic segmentation models (such as U-Net) are used to separate hazardous areas (such as cracks and landslides) from normal areas.

[0188] Feature extraction

[0189] Geometric features:

[0190] Crack: length, width, direction, number of branches.

[0191] Landslide: sliding surface area, displacement vector, and trailing edge crack density.

[0192] Texture features:

[0193] The roughness and contrast of the crack region were extracted using the gray-level co-occurrence matrix (GLCM).

[0194] Spectral characteristics:

[0195] In multispectral images, areas with abnormal vegetation cover (NDVI) and moisture content (NDWI) are extracted.

[0196] Topological relationship analysis:

[0197] Use GIS tools (such as ArcGIS) to determine whether the potential hazard area is adjacent to or contains it.

[0198] Example: Crack A is 50m away from landslide B, and is determined to be "potentially related".

[0199] The distance decay model shows that the risk of a hazard decreases as the distance increases.

[0200] Construction of a correlation map of hidden dangers in the dam area

[0201] Definition of graph nodes and edges

[0202] Node: A single potential hazard area (such as crack A, landslide B).

[0203] side:

[0204] Spatial association: directly adjacent or distance <100m.

[0205] Risk correlation: Correlation risk value > 0.5.

[0206] Visualization of graphs

[0207] tool:

[0208] Use Gephi or Cytoscape to generate network graphs, where node size represents risk level and edge thickness represents correlation strength.

[0209] Example:

[0210] The high-risk landslide B is connected to the medium-risk crack A, forming a "high-medium" correlation cluster.

[0211] Output of hazard results

[0212] Report content:

[0213] Single-area risk table: Lists the risk level and characteristics of all potential hazard areas.

[0214] Association map: Mark key association paths (e.g., landslide B → crack A → seepage point C).

[0215] Recommendations for handling:

[0216] Highly correlated areas: Prioritize reinforcement of landslide B to prevent triggering crack propagation.

[0217] Correspondingly, this application also provides a device for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, such as... Figure 2 As shown, including,

[0218] The first module is used to acquire geographical information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area.

[0219] The second module is used to determine the initial location points of potential hazards on the three-dimensional environmental model of the dam area, analyze the diffusion effect of each hazard category, expand the initial location points of potential hazards, and obtain the location range of potential hazards.

[0220] The third module is used to obtain the basic information of the inspection drone and match the flight information of the inspection drone in the area of ​​the potential hazard.

[0221] The fourth module is used to plan multiple optional paths for the inspection drone based on the location and area of ​​all potential hazards. It evaluates the multiple optional paths based on flight information, selects the target path, and the inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard to collect inspection images of the potential hazard area.

[0222] The fifth module is used to extract hazard features from inspection images of hazard areas, associate hazard features between different hazard areas, and output hazard results.

[0223] This application also provides a computer-readable storage medium that stores one or more programs, which, when executed by an electronic device including multiple applications, cause the electronic device to perform the methods described above.

[0224] Compared with the prior art, the beneficial effects of this invention are as follows:

[0225] 1. Initial location points of potential hazards are determined on the 3D environmental model of the dam area. The diffusion effect of each hazard category is analyzed, and the initial location points are expanded, taking into account the diffusion of hazards. This expansion facilitates the reliability of subsequent UAV inspection image acquisition and provides more accurate location information for subsequent path planning and image recognition. The flight information of the inspection UAV within the hazard location area is matched, and reasonable flight information is configured according to the different characteristics or terrain conditions of each hazard area to ensure the stability and image quality of the UAV inspection images.

[0226] 2. Based on flight information, multiple optional paths are evaluated to select the target path. The terrain characteristics of the potential hazard area and the flight status of the UAV are considered to select the most suitable target path. Furthermore, the characteristics of different potential hazard areas are correlated, improving the rationality and adaptability of inspection path planning, making inspection image recognition more accurate, meeting the needs of potential hazard inspection in the complex environment of the dam area, and ensuring the timeliness and safety of environmental hazard monitoring in the dam area.

[0227] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0228] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing the present invention.

[0229] Those skilled in the art will understand that the modules in the apparatus of the implementation scenario can be distributed within the apparatus of the implementation scenario as described, or they can be located in one or more apparatuses different from this implementation scenario, with corresponding changes. The modules of the above-described implementation scenario can be combined into one module, or they can be further divided into multiple sub-modules.

[0230] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas, characterized in that, include, Obtain geographic information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area; The initial location points of potential hazards were determined on the three-dimensional environmental model of the dam area. The diffusion effect of each hazard category was analyzed, and the initial location points of the potential hazards were expanded to obtain the location range of the potential hazards. Obtain basic information about the inspection drone and match its flight information within the area of ​​potential hazards. Based on the location and area of ​​all potential hazards, multiple optional paths for inspection drones are planned. The multiple optional paths are evaluated based on flight information, and the target path is selected. The inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard, thereby collecting inspection images of the potential hazard area. Extract hazard features from inspection images of hazard areas, associate hazard features between different hazard areas, and output hazard results.

2. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 1, characterized in that, The statistical set includes all categories of potential hazards in the dam area, including: Accidents that have occurred in the dam area in the past are recorded as potential hazards, and these hazards are classified. The attributes and locations of the accidents are marked on the three-dimensional environmental model of the dam area. Accidents that have never occurred in the history of the dam area are recorded as potential hazards that have not yet occurred. These potential hazards are then classified, and the union of the categories of both actual and potential hazards is taken as the set of all hazard categories.

3. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 2, characterized in that, The initial location points of potential hazards were determined on the three-dimensional environmental model of the dam area. The diffusion effect of each hazard category was analyzed, and the initial location points of the hazards were expanded to include... Accident risk analysis is performed on a three-dimensional environmental model of the dam area to obtain a risk probability heat map of the dam area. The risk probability of the existing hidden dangers in the risk probability heat map of the dam area is adjusted according to the attributes and locations of the existing hidden dangers. The initial location points and categories of potential hazards are determined based on the risk probability heat map of the dam area; Based on the three-dimensional environmental model of the dam area, a multiphysics simulation model of the dam area is established to simulate the diffusion of each type of hidden danger and the diffusion direction of the initial location point of each hidden danger. Based on the spread of each type of hazard, determine the spread scale range and condition standards corresponding to each type of hazard. Calculate the condition fit of the initial location point of each hazard based on the condition standards of each type of hazard. Select a spread scale within the spread scale range based on the condition fit of the initial location point of the hazard. Expand the region of the initial location point of the hazard based on the spread direction and spread scale of the initial location point of each hazard.

4. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 1, characterized in that, Match the flight information of the inspection drone within the area where the potential hazard is located. include, The basic information of the inspection drone includes the drone's structural information and image acquisition equipment information; The flight information of the inspection drone within the location area of ​​each potential hazard is determined based on the influence relationship between the drone's structural information, image acquisition equipment information, and the location area of ​​the potential hazard.

5. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 1, characterized in that, Based on the location and area of ​​all potential hazards, multiple optional routes for the inspection drones are planned. include, The starting and ending points of the inspection drone are determined, and the location area of ​​each potential hazard is used as a waypoint. Based on the starting point, multiple waypoints and the ending point, the path planning algorithm is used to plan the path of the inspection drone and generate multiple optional paths.

6. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 5, characterized in that, Multiple alternative routes are evaluated based on flight information to select the target route, including... The risk situation and terrain complexity of the location area of ​​the hidden danger are statistically analyzed to obtain risk indicators and terrain complexity indicators. Based on the risk indicators and terrain complexity indicators, a composite indicator of the location area of ​​a single hidden danger is determined, thereby determining the inspection level of each optional path. The flight information is marked at the corresponding waypoints on each optional path, and the evaluation information on each optional path is analyzed. The evaluation information includes inspection task information, inspection flight loss information, and inspection wasted effort information. Different inspection levels correspond to different combinations of the influence weights of inspection task information and inspection flight loss information. Combined with the inspection waste information, an evaluation index is obtained for each optional path. The target path is then selected based on the evaluation index.

7. The method for setting inspection paths and recognizing inspection images based on potential environmental hazards in dam areas according to claim 1, characterized in that, The system correlates the characteristics of different potential hazard areas and outputs hazard results, including: A comprehensive analysis of the hazard characteristics under each hazard area yields the hazard risk of each individual hazard area; Spatial correlation analysis was conducted on the hazard characteristics and risks of different hazard areas to obtain the hazard risk correlation of multiple hazard areas. Establish a hazard correlation map of the dam area to output hazard results.

8. A device for setting inspection routes and recognizing inspection images based on potential environmental hazards in dam areas, characterized in that, include, The first module is used to acquire geographical information of the dam area, construct a three-dimensional environmental model of the dam area, and statistically analyze the set of all potential hazard categories involved in the dam area. The second module is used to determine the initial location points of potential hazards on the three-dimensional environmental model of the dam area, analyze the diffusion effect of each hazard category, expand the initial location points of potential hazards, and obtain the location range of potential hazards. The third module is used to obtain the basic information of the inspection drone and match the flight information of the inspection drone in the area of ​​the potential hazard. The fourth module is used to plan multiple optional paths for the inspection drone based on the location and area of ​​all potential hazards. It evaluates the multiple optional paths based on flight information, selects the target path, and the inspection drone flies according to the target path and the flight information of the location and area of ​​the potential hazard to collect inspection images of the potential hazard area. The fifth module is used to extract hazard features from inspection images of hazard areas, associate hazard features between different hazard areas, and output hazard results.

9. A computer-readable storage medium storing one or more programs that, when executed by an electronic device including a plurality of applications, cause the electronic device to perform the method of any one of claims 1-7.

Images