Substation site selection data processing and feedback method, system and device based on unmanned aerial vehicle-AI cooperation, and medium
The substation site selection method, which combines drone photography and AI evaluation models, generates avoidance suggestions, overcoming the limitations of traditional manual site selection methods and achieving efficient and accurate substation site selection decisions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- GUANGXI POWER GRID CORP
- Filing Date
- 2025-11-14
- Publication Date
- 2026-04-28
AI Technical Summary
Traditional substation site selection methods that rely on manual feedback cannot meet the requirements for efficient and accurate decision-making, especially in complex geographical environments and large-scale data scenarios, where the limitations of feedback methods are particularly prominent.
The method employs a drone-AI collaborative approach, using drones to capture images of the site selection area, performing preprocessing and feature extraction, then using a pre-trained site selection environment assessment model to evaluate the results, generating avoidance suggestions, and displaying the assessment results through a 3D geographic information system platform.
It enables the acquisition of abundant site selection information in a short period of time, improves the scientificity and feasibility of site selection, reduces the risk of errors caused by human factors, optimizes site selection schemes, and improves decision-making efficiency and economy.
Smart Images
Figure CN121936745A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent site selection technology for power engineering, specifically to a method, system, equipment, and medium for substation site selection data processing and feedback based on UAV-AI collaboration. Background Technology
[0002] With the development of the power industry, the requirements for the scientific and rational selection of substation sites are increasing. In the traditional substation site selection process, data feedback is mostly based on experience summaries or simple reports after manual on-site surveys, which suffers from problems such as untimely feedback, lack of intuitiveness, and limited information. At the same time, with the increasing demand for substation construction and the increasing difficulty of site selection, relying solely on manual feedback methods can no longer meet the requirements for efficient and accurate decision-making, especially in complex geographical environments and large-scale data scenarios, where the limitations of traditional feedback methods become more prominent. Summary of the Invention
[0003] In view of the above-mentioned problems, the present invention provides a method, system, equipment and medium for substation site selection data processing and feedback based on UAV-AI collaboration.
[0004] Therefore, the technical problem solved by this invention is: how to address the fact that relying solely on manual feedback is no longer sufficient to meet the requirements of efficient and accurate decision-making, especially in complex geographical environments and large-scale data scenarios where the limitations of traditional feedback methods are even more pronounced.
[0005] To address the aforementioned technical problems, this invention provides the following technical solution: a substation site selection data processing and feedback method based on UAV-AI collaboration, comprising: controlling a UAV to photograph the substation site selection area according to a preset flight route to obtain images of the site selection area; preprocessing and extracting features from the site selection area images to obtain key environmental element information; inputting the key environmental element information into a pre-trained site selection environment assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity, and comprehensive site selection feasibility score; matching avoidance suggestions based on the assessment results and a preset engineering constraint rule base; and displaying the avoidance suggestions.
[0006] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the preprocessing and feature extraction of the site selection area image to obtain key environmental element information includes: performing radiometric correction, geometric correction and image enhancement on the acquired original site selection area image to obtain a standardized image; and using image segmentation technology based on the standardized image to identify and extract key environmental element information within the site selection area.
[0007] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the pre-trained site selection environment assessment model includes: collecting historical image data and corresponding environmental parameters of the substation site selection area; performing radiometric and geometric correction on the historical image data, normalizing the environmental parameters, and constructing a mapping relationship dataset between key features and environmental parameters; inputting the mapping relationship dataset into a machine learning algorithm for training to obtain the site selection environment assessment model; testing the output of the site selection environment assessment model using a validation dataset; and completing training when the prediction accuracy of the site selection environment assessment model for terrain suitability, infrastructure proximity assessment, and comprehensive site selection feasibility score all reach a preset threshold.
[0008] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the site selection environment assessment model includes a terrain feature extraction module, an infrastructure proximity module, and a multi-task fusion module. The terrain feature extraction module uses a convolutional neural network to analyze the terrain and landform raster data in the key environmental element information and outputs a probability distribution of terrain suitability level. The infrastructure proximity module uses a graph neural network to analyze the spatial topological relationship of infrastructure in the key environmental element information and outputs a weighted proximity score. The multi-task fusion module integrates the terrain suitability level probability distribution and the infrastructure proximity score, and generates a comprehensive site selection feasibility score through a fully connected layer.
[0009] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the key environmental element information includes topographic features, location of existing buildings and infrastructure; the environmental parameters include topographic elevation data, infrastructure distribution data, topographic suitability level labels, infrastructure proximity labels, and comprehensive feasibility score labels.
[0010] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the step of matching avoidance suggestions based on the evaluation results and a preset engineering constraint rule library includes: matching anomalies in the evaluation results based on the preset engineering constraint rule library; generating corresponding avoidance suggestions based on the matched anomalies; if two or more avoidance suggestions are generated, then the spatial parameters in the key environmental element information of each avoidance suggestion are extracted to calculate the implementation cost, and the avoidance suggestions are arranged in ascending order of implementation cost; the anomalies include terrain suitability level lower than a preset level, infrastructure proximity score lower than a first preset threshold, and comprehensive site selection feasibility score lower than a second preset threshold.
[0011] This preferred solution triggers an avoidance suggestion generation mechanism by matching anomalies in the evaluation results, and introduces spatial parameters to evaluate and rank implementation costs, so that the avoidance suggestions are not only targeted, but also take into account the economic feasibility of implementation.
[0012] As a preferred embodiment of the UAV-AI collaborative substation site selection data processing and feedback method described in this invention, the step of displaying avoidance suggestions includes spatially overlaying the avoidance suggestions with the corresponding site selection area image and evaluation results to generate a visual decision report; and displaying the visual decision report through a three-dimensional geographic information system platform.
[0013] This preferred solution generates a visualized decision report by spatially overlaying avoidance suggestions with site selection area images and assessment results, and displays it on a 3D geographic information system platform. This makes complex assessment information presented in an intuitive graphical way, improving the understandability of the site selection solution and the efficiency of decision-making. It is suitable for substation site selection processes that require multi-party collaboration and approval.
[0014] This invention provides a substation site selection data processing and feedback system based on UAV-AI collaboration.
[0015] To address the aforementioned technical problems, this invention provides the following technical solution: a substation site selection data processing and feedback system based on UAV-AI collaboration, comprising: an image acquisition module, a feature extraction module, an environmental assessment module, a constraint rule matching module, and a display module; the image acquisition module controls a UAV to capture images of the substation site selection area according to a preset flight route, obtaining images of the site selection area; the feature extraction module preprocesses and extracts features from the site selection area images to obtain key environmental element information; the environmental assessment module inputs the key environmental element information into a pre-trained site selection environmental assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity, and comprehensive site selection feasibility score; the constraint rule matching module matches avoidance suggestions based on the assessment results and a preset engineering constraint rule library; and the display module displays the avoidance suggestions.
[0016] The present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the substation site selection data processing and feedback method based on UAV-AI collaboration.
[0017] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the aforementioned substation site selection data processing and feedback method based on UAV-AI collaboration.
[0018] The beneficial effects of this invention are as follows: This invention acquires images of the site selection area using drones, achieving high efficiency in data collection. Then, it uses AI technology for data processing and analysis, automatically completing tasks such as image preprocessing, feature extraction, and evaluation, shortening the time from data collection to result output. Compared with traditional manual methods, it can provide site selection suggestions to decision-makers in a shorter time, accelerating the progress of substation construction projects.
[0019] By utilizing a pre-trained site selection environment assessment model, based on historical data and advanced machine learning algorithms, a comprehensive and accurate assessment of the terrain suitability, infrastructure proximity, and overall site selection feasibility of the selected area can be conducted. This enables a more objective and accurate judgment of the merits of a site selection, reducing the risk of site selection errors caused by human factors or insufficient data.
[0020] Based on the assessment results and the pre-set engineering constraint rule library, it can automatically match targeted avoidance suggestions and also sort multiple avoidance suggestions by cost, providing decision-makers with a more scientific and comprehensive reference, helping them to reasonably avoid potential risks, optimize substation site selection schemes, improve the feasibility and economy of site selection, and reduce changes and losses in the subsequent construction process.
[0021] By spatially overlaying avoidance recommendations with site selection area images and assessment results, a visualized decision report is generated and displayed through a 3D geographic information system platform. This allows decision-makers to more intuitively understand the environmental conditions of the site selection area, assessment results, and the spatial distribution of avoidance recommendations, facilitating rapid decision-making and improving the scientific and rational nature of the decision. Attached Figure Description
[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 The above is a flowchart of a method for processing and feeding back substation site selection data based on UAV-AI collaboration, provided as an embodiment of the present invention. Detailed Implementation
[0024] To make the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0025] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a method for substation site selection data processing and feedback based on UAV-AI collaboration, including: S1. Control the drone to take pictures of the substation site selection area according to the preset flight route to obtain images of the site selection area.
[0026] S2. Preprocess and extract features from the images of the selected area to obtain information on key environmental elements.
[0027] S3. Input key environmental element information into the pre-trained site selection environment assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity and comprehensive site selection feasibility score.
[0028] S4. Based on the assessment results and the preset engineering constraint rule library, match avoidance suggestions.
[0029] S5. Display the avoidance suggestions.
[0030] It should be noted that steps S1-S5 of this invention enable the drone to quickly photograph the site selection area along a preset flight route, acquiring a large amount of image data in a short time. Compared with traditional manual surveys, this significantly reduces data collection time and improves the overall efficiency of substation site selection. The evaluation results include multiple dimensions such as terrain suitability level, infrastructure proximity, and comprehensive site selection feasibility score, comprehensively reflecting the advantages and disadvantages of the site selection area. This provides decision-makers with rich information to help them make more scientific and reasonable site selection decisions. Based on the evaluation results and a preset engineering constraint rule base, avoidance suggestions are matched, providing specific solutions to potential problems in the site selection, further improving the scientific nature and feasibility of the site selection decision. The avoidance suggestions are displayed to users, allowing them to promptly understand the evaluation results and improvement suggestions for the site selection area, facilitating timely adjustments to the site selection plan or the implementation of corresponding measures based on feedback.
[0031] Example 2, an embodiment of the present invention, provides a method for substation site selection data processing and feedback based on UAV-AI collaboration, based on the previous embodiment, including: Furthermore, in step S1, the drone is controlled to take pictures of the substation site selection area according to the preset flight route to obtain images of the site selection area.
[0032] In this embodiment of the application, the key environmental element information in step S2 can be a set of environmental elements extracted from the preprocessed image of the site selection area, including the terrain and landform categories (such as flat land and steep slopes), building categories (such as substations and residential buildings), and infrastructure categories (such as roads and power transmission lines) obtained by U-Net segmentation, combined with the results of YOLOv7 detection of targets such as high-voltage towers and drainage ditches. After coordinate transformation, the elements form a vector layer and establish spatial topological relationships between the elements, thereby constituting structured key environmental element information data.
[0033] In an alternative implementation, key environmental element information can also be topographic element information. The slope, undulation shape and surface elevation difference of the site selection area are obtained through joint analysis of image segmentation and DEM data, which is used to support the assessment of the terrain suitability level.
[0034] In another alternative implementation, key environmental element information can also be infrastructure spatial distribution information. Location data such as roads, power transmission lines, and high-voltage towers are obtained through image segmentation and target detection, and the connectivity and proximity of infrastructure are represented by combining spatial topological relationships to support the calculation and analysis of infrastructure proximity.
[0035] This invention extracts key environmental element information from site selection area images, enabling the expression of terrain features, building distribution, and infrastructure structure in complex scenes in the form of structured data, thus allowing the evaluation model to obtain accurate and rich environmental input information.
[0036] Furthermore, in step S2, the selected area image is preprocessed and features are extracted to obtain key environmental element information, including the following steps B1-B2: B1. Perform radiometric correction, geometric correction, and image enhancement on the original images of the selected area to obtain standardized images.
[0037] B2. Based on standardized images, use image segmentation technology to identify and extract key environmental element information within the selected site area.
[0038] In this embodiment, the standardized image in step B1 is the result of radiometric correction, geometric correction, and image enhancement of the original image of the selected area. Specifically, radiometric correction uses histogram equalization or dark channel prior algorithms to eliminate the effects of uneven illumination, shadows, and atmospheric scattering; geometric correction performs orthorectification based on digital elevation models and ground control points to correct distortion errors caused by terrain undulations and UAV attitude changes; image enhancement can use nonlocal mean denoising or super-resolution reconstruction techniques to suppress noise and improve image edge sharpness. The image obtained after the above processing has uniform illumination conditions, spatial reference coordinates, and high-resolution quality, which facilitates subsequent image segmentation and feature extraction operations.
[0039] In one alternative implementation, the standardized imagery can also incorporate the fused imagery obtained after processing with multi-temporal image registration technology. For example, in images acquired by UAVs during multiple flights, the spatial offset caused by temporal differences can be eliminated through a registration algorithm, achieving standardized output under a unified reference across time phases.
[0040] In another alternative implementation, standardized imagery can also be enhanced by fusing multispectral data to improve the image processing results. This enhances the ability to identify the boundaries of green areas and water bodies through NDVI (Normalized Difference Vegetation Index), thereby improving the accuracy of identifying ecologically sensitive areas in the site selection area.
[0041] This invention uses standardized imagery as the basis for site selection analysis, which can significantly improve the accuracy and consistency of subsequent image segmentation and environmental element extraction.
[0042] Specifically, in step B1, histogram equalization or dark channel prior algorithm is used to eliminate the effects of uneven illumination, shadows and atmospheric scattering, and ensure consistent image brightness; based on digital elevation model (DEM) and ground control points (GCP), orthorectification is used to eliminate distortion caused by UAV attitude changes and terrain undulations; adaptive filtering (non-local mean denoising) or super-resolution reconstruction technology is used to suppress noise and improve edge sharpness to obtain standardized image.
[0043] Specifically, in step B2, based on the preprocessed standardized image, key environmental elements are extracted through the following steps: A U-shaped network (U-Net) is used to perform pixel-level classification of terrain (such as flat land and steep slopes), buildings (substations and residences), and infrastructure (roads and power transmission lines) in the imagery; the precise coordinates of key facilities (such as high-voltage towers and drainage ditches) are located using YOLOv7; the extracted feature location information is converted into vector layers, and spatial topological relationships (such as the connectivity between buildings and roads) are established; the generated vector layers and spatial topological relationships are stored in a structured manner to construct key environmental feature information data.
[0044] Key environmental information includes topographic features and the location of existing buildings and infrastructure.
[0045] Furthermore, in step S3, key environmental element information is input into the pre-trained site selection environment assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity, and comprehensive site selection feasibility score, including the following steps C1-C4: C1. Collect historical image data and corresponding environmental parameters of the substation site selection area.
[0046] C2. Perform radiometric and geometric corrections on historical image data, normalize environmental parameters, and construct a dataset of mapping relationships between key features and environmental parameters.
[0047] C3. Input the mapping relationship dataset into the machine learning algorithm for training to obtain the site selection environment evaluation model.
[0048] C4. Test the output of the site selection environment assessment model using the validation dataset. Training is complete when the prediction accuracy of the site selection environment assessment model for terrain suitability, infrastructure proximity assessment and comprehensive site selection feasibility score all reach the preset threshold.
[0049] In this embodiment, the environmental parameters in step C1 can be used to train the site selection environmental assessment model using various data labels closely related to regional site selection. These include: topographic elevation data, infrastructure distribution data, topographic suitability level labels, infrastructure proximity labels, and comprehensive site selection feasibility score labels. These parameters quantitatively represent the site selection area from the perspectives of spatial structure, facility accessibility, and comprehensive evaluation, and are mapped to corresponding historical image data to form a dataset used as model training input, enabling the model to learn the correlation between site selection conditions and environmental characteristics.
[0050] In one alternative implementation, environmental parameters can also incorporate an extended parameter set following meteorological, land use, or policy control data, such as adding average annual precipitation, soil type, and planned use classification, to enhance the site selection model's ability to judge climate adaptability, geological suitability, and policy compliance.
[0051] In another alternative implementation, environmental parameters can also be obtained through real-time parameter data dynamically collected by on-site sensors, such as short-period changing parameters like temperature, humidity, air quality, and traffic flow. These parameters can be fused with static parameters using time series feature extraction methods to support the adaptability and generalization ability of the site selection model under dynamic environmental changes.
[0052] By using structured environmental parameters as training inputs for a site selection assessment model, this invention can effectively establish a quantitative mapping relationship between key environmental factors and site selection feasibility, thereby improving the model's prediction accuracy and generalization ability.
[0053] Specifically, environmental parameters include topographic elevation data, infrastructure distribution data, topographic suitability level labels, infrastructure proximity labels, and comprehensive feasibility score labels.
[0054] The site selection environment assessment model includes a terrain feature extraction module, an infrastructure proximity module, and a multi-task fusion module. The terrain feature extraction module uses a convolutional neural network to analyze the terrain and landform raster data in the key environmental element information and outputs a probability distribution of terrain suitability level. The infrastructure proximity module uses a graph neural network to analyze the spatial topological relationship of infrastructure in the key environmental element information and outputs a weighted proximity score. The multi-task fusion module integrates the terrain suitability level probability distribution and the infrastructure proximity score and generates a comprehensive site selection feasibility score through a fully connected layer.
[0055] In this embodiment of the application, the engineering constraint rule base in step S4 can be used as a set of rules to determine anomalies in the evaluation results. The rule base includes three types of rules: those for terrain suitability level lower than a preset level, infrastructure proximity score lower than a first preset threshold, and comprehensive site selection feasibility score lower than a second preset threshold. When any indicator triggers the corresponding rule, a corresponding avoidance suggestion is generated.
[0056] In one alternative implementation, the engineering constraint rule base can also be based on a rule set constructed from a more refined hierarchical system, dividing terrain suitability into 5 levels, each level corresponding to a different level of engineering recommendation intensity, thereby enhancing the sensitivity and adaptability of the rules.
[0057] In another alternative implementation, the engineering constraint rule base can also introduce an adaptive rule set with a multi-source data dynamic update mechanism, such as combining meteorological data, land use planning or geological disaster early warning information to dynamically adjust the proximity threshold or scoring criteria to adapt to environmental changes or policy changes at different stages of the site selection area.
[0058] This invention uses an engineering constraint rule base to judge and provide feedback on the site selection evaluation results, so that the output of avoidance suggestions has clear triggering logic and operability.
[0059] Furthermore, in step S4, based on the evaluation results and the preset engineering constraint rule base, avoidance suggestions are matched, including the following steps D1-D3: D1. Match the outliers in the evaluation results according to the preset engineering constraint rule library.
[0060] D2. Based on the matched anomalies, generate corresponding avoidance suggestions.
[0061] D3. If two or more avoidance suggestions are generated, the spatial parameters of the key environmental elements are extracted from each avoidance suggestion to calculate the implementation cost, and the avoidance suggestions are arranged in ascending order of implementation cost.
[0062] Specifically, the abnormal items include terrain suitability level lower than the preset level, infrastructure proximity score lower than the first preset threshold, and comprehensive site selection feasibility score lower than the second preset threshold.
[0063] When the terrain suitability level is lower than the preset level, a site leveling plan is recommended; when the infrastructure proximity score is less than the first preset threshold, it is recommended to add a connecting road or power grid extension path; when the comprehensive site selection feasibility score is less than the second preset threshold, output the coordinates of the alternative site selection area and feasibility comparison data. For example, based on a pre-defined library of engineering constraint rules, the following rules can be defined: Rule 1: If the terrain suitability level of a certain area is lower than level 4 (for example, a slope greater than 15° is considered unsuitable), then the terrain unsuitability anomaly is triggered, and a suggestion to level the site is generated to avoid this.
[0064] Rule 2: If the infrastructure proximity score of a candidate point is lower than 60 (the lower the score, the worse the accessibility), the infrastructure accessibility difference constant is triggered, and avoidance suggestions for planning new connecting roads or power grid extension lines are generated.
[0065] Rule 3: If the overall score of a candidate site is below 70 (the lower the score, the worse the feasibility), a suggestion to activate the candidate site selection scheme will be generated.
[0066] If two or more avoidance suggestions are generated, the spatial parameters of key environmental elements are extracted from each avoidance suggestion to calculate the implementation cost, and the avoidance suggestions are arranged in ascending order of implementation cost.
[0067] Furthermore, in step S5, the avoidance suggestions will be presented, including the following steps E1-E2: E1. Spatially overlay the avoidance recommendations with the corresponding site selection area images and assessment results to generate a visual decision report.
[0068] E2. Visualize decision reports through a 3D geographic information system platform.
[0069] Specifically, the terrain suitability level is overlaid on the site selection area image in the form of a heat map; Infrastructure proximity scores use buffer radii to indicate the locations of critical facilities. The comprehensive site selection feasibility score and avoidance suggestions are displayed in a floating pop-up window.
[0070] For example, on a 3D terrain model, suitability zones of different levels are displayed in semi-transparent overlays of different colors (e.g., from green to red). Areas with anomalies are highlighted, and clicking on them triggers a pop-up window displaying detailed information about the area, its proximity score, overall score, and corresponding avoidance suggestions. Users can rotate, zoom, and pan the 3D scene at any angle using their devices (such as computers or tablets).
[0071] Example 3 is an embodiment of the present invention, which provides a method for substation site selection data processing and feedback based on UAV-AI collaboration. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.
[0072] A municipal power company plans to build a new 110kV substation in a mountainous area on the outskirts of the city to meet the growing electricity demand in the region. The area has complex terrain, including hills, woodlands, and a small number of existing residences and infrastructure. To complete the site selection scientifically and efficiently, the project team adopted the method of this invention.
[0073] The drone, equipped with a high-definition visible light camera, automatically cruises and takes pictures along a pre-set grid flight path that covers the entire area.
[0074] High-resolution raw image data of the selected site area was obtained, totaling approximately 2,000 high-resolution photos.
[0075] Radiometric correction (to eliminate uneven lighting) and geometric correction (to eliminate distortion based on DEM and GCPs) were performed on the original images, and image enhancement techniques were used to improve contrast, resulting in a set of seamlessly stitched standardized orthophoto maps.
[0076] The U-Net semantic segmentation model was used to perform pixel-level classification on standardized images, identifying terrain features (such as flat land, steep slopes, and woodlands), existing buildings (residential buildings and factories), and infrastructure (roads and existing power transmission lines). Simultaneously, the YOLOv7 model was used to accurately locate the coordinates of key facilities (such as high-voltage towers).
[0077] A key environmental feature information layer containing vector boundaries and attribute information was generated. For example, area A (candidate area 1) was identified as a gentle slope woodland, while area B (candidate area 2) was adjacent to a main road, but contained a residential building and a steep slope.
[0078] The extracted key environmental element information (converted into raster data and topology diagrams) is input into the pre-trained site selection environment assessment model.
[0079] The terrain feature extraction module (Convolutional Neural Network, CNN) analyzed the elevation and slope data of areas A and B.
[0080] The infrastructure proximity module (Graph Neural Network, GNN) calculates the spatial topological relationships and distances from areas A and B to major roads and power grid access points.
[0081] The assessment results for Area A were obtained: Terrain suitability level: Level 4 (Preset threshold: ≤ Level 3 is suitable. Level 4 indicates a large slope, Level 4 > Level 3 → unqualified / abnormal, requiring a certain degree of site leveling).
[0082] Infrastructure proximity score: 55 points (preset threshold: ≥60 points is acceptable. 55 points indicates that the distance from existing main roads and power grids is relatively far).
[0083] Overall site selection feasibility score: 65 points (preset threshold: ≥70 points is recommended).
[0084] Assessment results for Area B: Terrain suitability level: Level 2 (Suitable).
[0085] Infrastructure proximity score: 75 (good).
[0086] Overall site selection feasibility score: 80 points (recommended).
[0087] The evaluation results for area A are automatically matched with the preset engineering constraint rule base to identify anomalies: When the terrain suitability level (level 3) is greater than the preset level (level 4), an anomaly is obtained: 1.
[0088] An anomaly 2 is obtained when the infrastructure proximity score (55 points) is less than the first preset threshold (60 points).
[0089] An anomaly 3 is obtained when the overall site selection feasibility score (80 points) is less than the second preset threshold (70 points).
[0090] Based on the rule base, the following avoidance suggestions were generated for area A, and the implementation cost was calculated: For abnormal terrain: a "site leveling plan" is generated, with an estimated earthwork volume of 5,000 cubic meters and an estimated implementation cost of 300,000 yuan.
[0091] In response to infrastructure anomalies: A suggestion was generated to "add a new 1.5-kilometer-long connecting road and power grid extension path", with an estimated implementation cost of 1.2 million yuan.
[0092] The system automatically sorts the two recommendations in ascending order of implementation cost, prioritizing the presentation of the lower-cost "site leveling solution" to decision-makers.
[0093] The assessment results of areas A and B (in the form of heat maps and buffer radii) and the generated avoidance suggestions are overlaid with the site selection area image.
[0094] Generate visualized decision reports on a 3D geographic information system platform.
[0095] Decision-makers can clearly see that area A is marked in orange (unsuitable terrain) by opening the 3D scene on a computer terminal.
[0096] When the mouse clicks on area A, a floating pop-up window immediately appears, displaying its terrain level (level 4), proximity score, overall score, and a list of avoidance suggestions sorted by cost.
[0097] Meanwhile, area B is displayed as green (suitable) in the 3D model, and all scores are clearly displayed.
[0098] Example 4 is an embodiment of the present invention. This embodiment provides a substation site selection data processing and feedback system based on UAV-AI collaboration, including an image acquisition module, a feature extraction module, an environmental assessment module, a constraint rule matching module, and a display module.
[0099] The image acquisition module is used to control the drone to take pictures of the substation site selection area according to the preset flight route, and obtain images of the site selection area.
[0100] The feature extraction module is used to preprocess and extract features from the images of the selected area to obtain key environmental element information.
[0101] The environmental assessment module is used to input key environmental element information into a pre-trained site selection environmental assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity and comprehensive site selection feasibility score.
[0102] The constraint rule matching module is used to match avoidance suggestions based on the evaluation results and the preset engineering constraint rule library.
[0103] The display module is used to showcase avoidance suggestions.
[0104] This embodiment also provides an electronic device applicable to the substation site selection data processing and feedback method based on UAV-AI collaboration, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the substation site selection data processing and feedback method based on UAV-AI collaboration proposed in the above embodiment.
[0105] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the substation site selection data processing and feedback method based on UAV-AI collaboration as proposed in the above embodiments.
[0106] The storage medium proposed in this embodiment and the method for processing and feedback substation site selection data based on UAV-AI collaboration proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.
[0107] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.
[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
Claims
1. A method for substation site selection data processing and feedback based on UAV-AI collaboration, characterized by: include, The drone is controlled to take pictures of the substation site selection area according to the preset flight route, and the site selection area is obtained. Preprocessing and feature extraction of images of the site selection area yields information on key environmental elements. Key environmental element information is input into a pre-trained site selection environment assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity and comprehensive site selection feasibility score. Based on the assessment results and the pre-set engineering constraint rule library, match avoidance suggestions; Suggestions for avoiding these issues will be presented.
2. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 1, characterized in that: The preprocessing and feature extraction of the site selection area image yields key environmental element information, including: Radiometric correction, geometric correction, and image enhancement are performed on the original images of the selected site area to obtain standardized images; Based on standardized imagery, image segmentation technology is used to identify and extract key environmental element information within the selected site area.
3. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 2, characterized in that: The pre-trained site selection environment evaluation model includes, Collect historical image data and corresponding environmental parameters of the substation site selection area; Radiometric and geometric corrections were performed on historical image data, environmental parameters were normalized, and a dataset of mapping relationships between key features and environmental parameters was constructed. The mapping relationship dataset is input into a machine learning algorithm for training to obtain a site selection environment assessment model. The site selection environment assessment model output was tested using a validation dataset. Training was completed when the prediction accuracy of the site selection environment assessment model for terrain suitability, infrastructure proximity assessment, and comprehensive site selection feasibility score all reached the preset threshold.
4. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 3, characterized in that: The site selection environment assessment model includes a terrain feature extraction module, an infrastructure proximity module, and a multi-task fusion module; The terrain feature extraction module uses a convolutional neural network to parse the terrain and landform raster data in key environmental element information and outputs the probability distribution of terrain suitability level. The infrastructure proximity module uses a graph neural network to analyze the spatial topological relationships of infrastructure in key environmental element information and outputs a weighted proximity score. The multi-task fusion module integrates the probability distribution of terrain suitability level and infrastructure proximity score, and generates a comprehensive site selection feasibility score through the fully connected layer.
5. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 4, characterized in that: The key environmental element information includes topographic features, location of existing buildings and infrastructure; The environmental parameters include topographic elevation data, infrastructure distribution data, topographic suitability level labels, infrastructure proximity labels, and comprehensive feasibility score labels.
6. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 5, characterized in that: The process of matching avoidance suggestions based on the evaluation results and a pre-set engineering constraint rule base includes: Based on a pre-defined engineering constraint rule library, match outliers in the evaluation results; Based on the matched anomalies, corresponding avoidance suggestions are generated; If two or more avoidance suggestions are generated, the spatial parameters of key environmental elements are extracted from each avoidance suggestion to calculate the implementation cost, and the avoidance suggestions are arranged in ascending order of implementation cost. Abnormal items include terrain suitability level lower than preset level, infrastructure proximity score lower than the first preset threshold, and comprehensive site selection feasibility score lower than the second preset threshold.
7. The substation site selection data processing and feedback method based on UAV-AI collaboration as described in claim 6, characterized in that: The article will present avoidance suggestions. include, The avoidance recommendations are spatially overlaid with the corresponding site selection area images and assessment results to generate a visual decision report; Visualized decision-making reports are presented through a 3D geographic information system platform.
8. A substation site selection data processing and feedback system based on UAV-AI collaboration, employing the substation site selection data processing and feedback method based on UAV-AI collaboration as described in any one of claims 1 to 7, characterized in that, include: The system includes an image acquisition module, a feature extraction module, an environmental assessment module, a constraint rule matching module, and a display module. The image acquisition module is used to control the drone to take pictures of the substation site selection area according to the preset flight route, and obtain images of the site selection area. The feature extraction module is used to preprocess and extract features from the site selection area image to obtain key environmental element information; The environmental assessment module is used to input key environmental element information into a pre-trained site selection environmental assessment model to obtain assessment results, including terrain suitability level, infrastructure proximity and comprehensive site selection feasibility score. The constraint rule matching module is used to match avoidance suggestions based on the evaluation results and the preset engineering constraint rule library; The display module is used to show avoidance suggestions.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the substation site selection data processing and feedback method based on UAV-AI collaboration as described in any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the substation site selection data processing and feedback method based on UAV-AI collaboration as described in any one of claims 1 to 7.