Photovoltaic station three-dimensional model and GIS deep fusion method and system

Abstract

The invention belongs to the technical field of image processing and data fusion, and particularly provides a photovoltaic station three-dimensional model and GIS deep fusion method and system, and the method comprises the steps: collecting multi-source data of a photovoltaic station, and carrying out the preprocessing; performing feature extraction on the image data to obtain feature points, and then performing feature matching and feature point filtering to obtain a feature point set of the photovoltaic station; performing panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; extracting three-dimensional point cloud data, and generating a photovoltaic station three-dimensional model in combination with the panoramic image; converting the three-dimensional model of the photovoltaic station into a format compatible with a GIS (Geographic Information System), and carrying out bidirectional mapping on GIS data and the three-dimensional model of the photovoltaic station; and performing technical fusion and real-time data integration on the three-dimensional model of the photovoltaic station and a GIS platform to form a multi-dimensional visual model of the photovoltaic station. According to the method and the system, the three-dimensional model of the photovoltaic station and the GIS data can be fused, and the visualization degree and the intuition of the photovoltaic station data are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing and data fusion, and specifically, relates to a method and system for deep fusion of a photovoltaic station three-dimensional model and GIS. Background Art

[0002] As the global demand for renewable energy continues to increase, photovoltaic power stations, as an important green energy generation method, have been widely used and developed rapidly. Photovoltaic power stations are usually distributed over vast areas of land, with a large number of equipment, which poses a huge challenge to their management and maintenance. In order to meet the needs of modern and efficient operation and maintenance, it has become a trend to use advanced 3D visualization technology to digitally and intelligently manage photovoltaic power stations.

[0003] However, the existing technologies usually use simplified 3D modeling methods or low-precision scanning technologies, which cannot accurately capture the detailed features of photovoltaic power station equipment, resulting in insufficient accuracy of the modeling model. In addition, there are obstacles in integrating the high-precision 3D model of a large photovoltaic station with a GIS map, making it difficult to achieve seamless integration and unified display.

[0004] Therefore, finding a method that can not only perform high-precision three-dimensional modeling of photovoltaic station equipment, but also seamlessly integrate the high-precision three-dimensional model with the GIS system is a technical problem that needs to be urgently solved by technical personnel in this field. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a method and system for deep integration of a photovoltaic station three-dimensional model and GIS, which can integrate the photovoltaic station three-dimensional model with GIS data, thereby improving the visualization and intuitiveness of the photovoltaic station data.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a method for deep integration of a photovoltaic station three-dimensional model and GIS, comprising the following steps: S1. Collect multi-source data of the photovoltaic station using data acquisition equipment and pre-process the multi-source data; the multi-source data includes point cloud data and image data; S2, extracting features from the processed image data to obtain feature points, performing feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station; S3, performing panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; S4, extracting three-dimensional point cloud data from the panoramic image, performing fine processing on the three-dimensional point cloud data, and generating a three-dimensional model of the photovoltaic station in combination with the panoramic image; S5. Convert the data format of the PV station three-dimensional model into a format compatible with GIS, adjust the coordinates, and perform bidirectional mapping between the GIS data and the PV station three-dimensional model; S6. Based on bidirectional mapping, the adjusted PV station three-dimensional model is technically integrated with the GIS platform and real-time data is integrated to form a multi-dimensional visualization model of the PV station.

[0007] In a preferred solution, in step S2, feature extraction includes: An adaptive scale parameter is calculated according to the area of ​​the photovoltaic panel of the photovoltaic station and the area of ​​the processed image data, a Gaussian pyramid is constructed using the adaptive scale parameter, and the difference of adjacent Gaussian images is calculated to generate a Gaussian difference pyramid; Detect and locate key points in the Gaussian difference pyramid to determine the location of key points; Construct a circular area around the key point, calculate the gradient magnitude and direction of each pixel in the circular area, and calculate the weight of each direction, multiply the gradient magnitude by the weight of the corresponding direction, and add them to the histogram. Determine the main direction of the key point based on the peak value in the histogram; Taking the key point as the center, select the neighborhood and divide it into sub-regions. Calculate the gradient histogram of the main direction of the key point in each sub-region to generate a feature description vector. The feature description vector is normalized to obtain feature points.

[0008] In a preferred solution, the calculation formula of the adaptive scale parameter is: ; in, represents the basic scale parameter, represents the photovoltaic panel area of ​​the photovoltaic station, Represents the area of ​​the processed image data.

[0009] In a preferred embodiment, step S4 comprises the following steps: S41, generating a depth map according to the panoramic image, and converting the depth map into three-dimensional point cloud data, retaining the color information of each point; S42, denoising the three-dimensional point cloud data, and performing point cloud registration with the processed point cloud data to obtain comprehensive point cloud data; S43, performing feature point matching on the panoramic images of different viewing angles, and generating a sparse point cloud and a dense point cloud, and performing point cloud fusion on the sparse point cloud and the dense point cloud to obtain fused point cloud data; S44, converting the fused point cloud data into a triangular mesh, and smoothing the triangular mesh to obtain a three-dimensional model of the photovoltaic station.

[0010] In a preferred solution, in step S43, the point cloud fusion operation is as follows: Use the coarse registration method to align the coordinates of the sparse point cloud and the dense point cloud. For each point pi in the sparse point cloud P, find the nearest corresponding point qi in the dense point cloud Q. Calculate the center of mass and center of mass coordinates of pi and qi, and determine the covariance matrix of the sparse point cloud and the dense point cloud according to the center of mass coordinates; Decompose the covariance matrix of the sparse point cloud and the dense point cloud, calculate the rotation matrix and the translation vector, obtain the transformation matrix according to the rotation matrix and the translation matrix, and apply the transformation matrix to the sparse point cloud to obtain the transformed sparse point cloud; Determine the overlapping area of ​​the sparse point cloud and the dense point cloud according to the distance between the transformed sparse point cloud and the dense point cloud, assign weights to each point in the overlapping area according to the reliability, merge the overlapping areas according to the weights, and directly merge the sparse point cloud and the dense point cloud in the non-overlapping area; The merged point cloud is deduplicated and filtered, and the fused point cloud is smoothed to obtain fused point cloud data.

[0011] In a preferred solution, the step S44 further includes optimizing the resolution of the triangular mesh, as follows: Reconstruct the equipment elements in the 3D model of the photovoltaic station, and use an adaptive mesh subdivision algorithm to increase the mesh density to an average side length of no more than 2 cm; The non-equipment factors in the 3D model of the photovoltaic station were simplified, and the grid density was adjusted to an average side length of no more than 10 cm using a grid downsampling algorithm.

[0012] In a preferred solution, step S5 specifically includes: Convert the PV station 3D model into geojson format and perform coordinate system conversion to WGS84 coordinates; Establish spatial index for GIS data and PV station 3D model, create mapping table, and establish attribute mapping mechanism for GIS data and PV station 3D model; Listeners are set for GIS data and PV station three-dimensional models respectively. When the listener detects changes in GIS data and / or changes in the data of the PV station three-dimensional model, the listener is triggered to synchronize and update the data.

[0013] In a preferred solution, step S6 includes: Encapsulate the 3D JavaScript library of the adjusted PV station 3D model and set a standardized API interface; Collect the information data of the photovoltaic station, clean, convert the format and structure the information data, associate it with the three-dimensional model of the photovoltaic station through the API interface, and obtain the real-time three-dimensional model of the photovoltaic station; A 3D scene is created in the GIS platform, the real-time PV station 3D model is loaded into the 3D scene, and 3D rendering is performed using WebGL to obtain a visualization model of the PV station.

[0014] In a preferred solution, step S6 further includes: A multi-layer layer structure is set for the visualization model of the photovoltaic station; the layer structure includes a device layer, a monitoring layer and a fault layer; A user-friendly interface is provided in each layer structure, allowing the user to view detailed information of the layer structure.

[0015] The present invention also provides a photovoltaic station three-dimensional model and GIS deep fusion system, which uses the above photovoltaic station three-dimensional model and GIS deep fusion method to form a multi-dimensional visualization model of the photovoltaic station, including: A data processing module configured to collect multi-source data of the photovoltaic station using a data acquisition device and pre-process the multi-source data; A feature point extraction module is configured to perform feature extraction on the processed image data to obtain feature points, and perform feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station; A panoramic stitching module, configured to perform panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; A model generation module is configured to extract three-dimensional point cloud data from the panoramic image, perform fine processing on the three-dimensional point cloud data, and generate a three-dimensional model of the photovoltaic station in combination with the panoramic image; A bidirectional mapping module is configured to convert the data format of the photovoltaic station three-dimensional model into a format compatible with GIS, and adjust the coordinates to perform bidirectional mapping between the GIS data and the photovoltaic station three-dimensional model; The visualization module is configured to perform technical fusion and real-time data integration of the adjusted PV station three-dimensional model with the GIS platform based on bidirectional mapping to form a multi-dimensional visualization model of the PV station.

[0016] The present invention provides a method and system for deep integration of a photovoltaic station three-dimensional model and GIS, which has the following beneficial effects: 1. The present invention improves the visualization and intuitiveness of photovoltaic station data by integrating the three-dimensional model of the photovoltaic station with GIS data, realizes real-time data integration and dynamic update, and greatly enhances the monitoring, management and analysis capabilities of the photovoltaic station. Users can also view and analyze various information of the photovoltaic station from all directions and angles on the GIS platform.

[0017] 2. By dynamically calculating the adaptive scale parameters according to the area of ​​the photovoltaic panel and the area of ​​the image data, the feature extraction process is made more accurate and efficient, and can effectively capture features of different scales, which not only ensures the accurate extraction of key equipment details, but also can effectively process the overall characteristics of large-area photovoltaic panel arrays and improve the accuracy of three-dimensional modeling.

[0018] 3. By converting the 3D model into geojson format and connecting it to the WGS84 coordinate system, seamless integration with the GIS platform is achieved. Spatial indexes are established for GIS data and the 3D model of the photovoltaic station, a mapping table is created, and listeners are set up for both. Any data changes can be captured in real time, ensuring the consistency and real-time nature of the data between the GIS platform and the 3D model, and greatly improving the efficiency and accuracy of data management.

[0019] 4. By fusing sparse point clouds with dense point clouds, not only the accuracy and detail restoration of the 3D model are improved, but also the limitations of a single algorithm in dealing with large-scale complex scenes are effectively solved, so that the PV 3D model contains both overall structural information and retains detailed features. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present invention will be further described below in conjunction with the accompanying drawings and embodiments: Figure 1 This is a flow chart of the method for deep integration of the photovoltaic station three-dimensional model and GIS of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0022] Embodiment 1: like Figure 1 As shown, the present invention provides a method for deep integration of a photovoltaic station three-dimensional model and GIS, comprising the following steps: S1. Use data acquisition equipment to collect multi-source data of the photovoltaic station and pre-process the multi-source data; the multi-source data includes point cloud data and image data.

[0023] The data acquisition device can be a high-precision three-dimensional laser scanner, a high-resolution handheld camera (at least 4k resolution), a drone equipped with a 4k camera and a high-precision airborne laser radar, or it can be other acquisition devices. The present invention does not impose specific restrictions on the data acquisition device.

[0024] In this embodiment, a 3D laser scanner is used to perform a 3D scan of the photovoltaic area and booster station equipment, a drone is used to collect oblique photography data, a handheld camera is used to take detailed pictures of the equipment, and an airborne laser radar is used to obtain point cloud data within the photovoltaic station range.

[0025] In this embodiment, preprocessing is performed on multiple sources, specifically including: De-noising the point cloud data and the image data respectively, and performing image enhancement and geometric correction on the denoised image data to obtain corrected image data; The corrected image data and the denoised point cloud data are format converted into a unified format to obtain processed image data and processed point cloud data.

[0026] S2. Extract features from the processed image data to obtain feature points, and perform feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station.

[0027] Specifically, feature extraction includes: The adaptive scale parameter is calculated according to the area of ​​the photovoltaic panel of the photovoltaic station and the area of ​​the processed image data, and the Gaussian pyramid is constructed using the adaptive scale parameter. The difference of adjacent Gaussian images is calculated to generate a Gaussian difference pyramid. The calculation formula of the adaptive scale parameter is as follows: ; in, represents the basic scale parameter, represents the photovoltaic panel area of ​​the photovoltaic station, Represents the area of ​​the processed image data; Detect and locate key points in the Gaussian difference pyramid to determine the location of key points; Construct a circular area around the key point, calculate the gradient magnitude and direction of each pixel in the circular area, and calculate the weight of each direction, multiply the gradient magnitude by the weight of the corresponding direction, and add them to the histogram. Determine the main direction of the key point based on the peak value in the histogram; With the key point as the center, select a 16x16 neighborhood and divide the neighborhood into 4x4 sub-regions. Calculate the gradient histogram of the main direction of the key point in each sub-region to generate a 128-dimensional feature description vector. The 128-dimensional feature description vector is normalized to obtain feature points.

[0028] By dynamically calculating adaptive scale parameters based on the area of ​​the photovoltaic panel and the area of ​​the image data, the feature extraction process is made more accurate and efficient, and can effectively capture features of different scales, which not only ensures the accurate extraction of key equipment details, but also can effectively process the overall characteristics of large-area photovoltaic panel arrays, improve the accuracy of three-dimensional modeling, and greatly reduce the waste of computing resources.

[0029] In this embodiment, the feature point matching and feature point filtering use the RANSAC algorithm.

[0030] S3. Perform panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station.

[0031] S4. Extract three-dimensional point cloud data from the panoramic image, perform fine processing on the three-dimensional point cloud data, and generate a three-dimensional model of the photovoltaic station in combination with the panoramic image.

[0032] Furthermore, step S4 specifically includes: S41, generating a depth map according to the panoramic image, and converting the depth map into three-dimensional point cloud data, retaining the color information of each point; S42, denoising the three-dimensional point cloud data, and performing point cloud registration with the processed point cloud data to obtain comprehensive point cloud data; S43, performing feature point matching on the panoramic images of different viewing angles, and generating a sparse point cloud and a dense point cloud, and performing point cloud fusion on the sparse point cloud and the dense point cloud to obtain fused point cloud data; S44, converting the fused point cloud into a triangular mesh, and smoothing the triangular mesh to obtain a three-dimensional model of the photovoltaic station.

[0033] By fusing sparse point clouds with dense point clouds, not only the accuracy and detail restoration of the three-dimensional model are improved, but also the limitations of a single algorithm in dealing with large-scale complex scenes are effectively solved, so that the photovoltaic three-dimensional model contains both overall structural information and retains detailed features.

[0034] In an embodiment of the present application, the smoothing process in step S44 specifically includes: Check the topological structure of the triangular mesh, identify abnormal triangles in the triangular mesh, and mark the key processing areas; For each vertex of the triangle mesh, the weighted average position of its neighboring vertices is calculated, and the vertex is moved to a new position according to the vertex weight; The moved triangular mesh is optimized using the Laplace smoothing algorithm and iterated until a preset smoothness is achieved.

[0035] It is understandable that the preset smoothness level is set according to actual usage requirements, and the present invention does not impose any specific limitation on this.

[0036] In this embodiment, the SfM algorithm is used to generate a sparse point cloud, and the MVS algorithm is used to generate a dense point cloud.

[0037] The point cloud fusion specifically includes: Use the coarse registration method to align the coordinates of the sparse point cloud and the dense point cloud. For each point pi in the sparse point cloud P, find the nearest corresponding point qi in the dense point cloud Q. Calculate the center of mass and center of mass coordinates of pi and qi, and determine the covariance matrix of the sparse point cloud and the dense point cloud according to the center of mass coordinates; Decompose the covariance matrix of the sparse point cloud and the dense point cloud, calculate the rotation matrix and the translation vector, and obtain the transformation matrix according to the rotation matrix and the translation matrix, and apply the transformation matrix to the sparse point cloud to obtain the transformed sparse point cloud; Determine the overlapping area of ​​the sparse point cloud and the dense point cloud according to the distance between the transformed sparse point cloud and the dense point cloud, assign weights to each point in the overlapping area according to the reliability, merge the overlapping areas according to the weights, and directly merge the sparse point cloud and the dense point cloud in the non-overlapping area; The merged point cloud is deduplicated and filtered, and the fused point cloud is smoothed to obtain fused point cloud data.

[0038] Among them, the coarse registration method adopts the ICP (Iterative Closest Point) iterative closest point algorithm, as follows: Extract key feature points from sparse point clouds and dense point clouds, use the RANSAC algorithm to match feature points, and calculate the initial transformation matrix based on the matching results; Find the nearest point in the target point cloud from each point in the sparse point cloud, and minimize the distance between corresponding point pairs to solve the rigid body transformation; Apply the calculated transformation to the sparse point cloud and calculate the mean square error after the transformation; Repeat the above steps until the mean square error is less than the preset threshold, then stop the iteration and determine the nearest corresponding point in the dense point cloud for each point in the sparse point cloud.

[0039] Furthermore, local smoothing is performed on the fused point cloud, that is, K nearest neighbor points of each fused point cloud are calculated and local smoothing is performed using bilateral filtering.

[0040] In this embodiment, after obtaining the three-dimensional model of the photovoltaic station, step S44 further includes optimizing the resolution of the triangular mesh, as follows: Reconstruct the equipment elements in the 3D model of the photovoltaic station, and use an adaptive mesh subdivision algorithm to increase the mesh density to an average side length of no more than 2 cm; The non-equipment factors in the 3D model of the photovoltaic station were simplified, and the grid density was adjusted to an average side length of no more than 10 cm using a grid downsampling algorithm.

[0041] Specifically, the adaptive mesh subdivision algorithm includes: The equipment elements in the three-dimensional model of the photovoltaic station form a first triangular mesh; Analyze the curvature value of each triangle in the first triangle mesh and detect edge features; Calculate the geometric complexity of the local area of ​​each triangle; Set the curvature threshold and edge length threshold (2cm). When the local curvature of the local area is greater than the curvature threshold or the edge length exceeds 2cm, insert a new vertex at the midpoint of the triangle mesh and connect the new vertices to form a new triangle. Use interpolation methods to optimize the positions of new vertices.

[0042] Furthermore, the grid downsampling algorithm specifically includes: The non-equipment factors in the 3D model of the photovoltaic station form a second triangular mesh; Calculate the importance of each vertex, merge the edges with a length less than 10cm, update the triangles, and form a new second triangular mesh; Calculate the optimal positions of the new vertices in the new second triangle mesh.

[0043] S5. Convert the data format of the PV station three-dimensional model into a format compatible with GIS, adjust the coordinates, and perform bidirectional mapping between the GIS data and the PV station three-dimensional model.

[0044] Specifically, step S5 specifically includes: Convert the PV station 3D model into geojson format and perform coordinate system conversion to WGS84 coordinates; Establish spatial index for GIS data and PV station 3D model, create mapping table, and establish attribute mapping mechanism for GIS data and PV station 3D model; Listeners are set for GIS data and PV station three-dimensional models respectively. When the listener detects changes in GIS data and / or data changes in the PV station three-dimensional model, the listener is triggered and the data is synchronously updated according to the mapping mechanism.

[0045] The present invention realizes seamless integration with the GIS platform by converting the three-dimensional model into the geojson format and docking it with the WGS84 coordinate system, and establishes spatial indexes for the GIS data and the three-dimensional model of the photovoltaic station, creates a mapping table, and sets listeners for both of them respectively, which can capture any data changes in real time, ensure the consistency and real-time performance of the data between the GIS platform and the three-dimensional model, and greatly improve the efficiency and accuracy of data management.

[0046] Preferably, step S5 further includes: Establish a multi-resolution pyramid model to classify the 3D model of the photovoltaic station; Use dynamic LOD to adaptively adjust the 3D model of the photovoltaic station.

[0047] Specifically, models with different levels of detail (LOD) are generated according to the three-dimensional model of the photovoltaic station, and the models with different LODs are organized into a pyramid structure; According to the preset distance threshold, models of different LOD levels are dynamically selected for rendering to achieve a smooth transition between LOD levels and avoid visual discomfort caused by sudden changes.

[0048] S6. Based on bidirectional mapping, the adjusted PV station three-dimensional model is technically integrated with the GIS platform and real-time data is integrated to form a multi-dimensional visualization model of the PV station.

[0049] Specifically, step S6 includes: Encapsulate the 3D JavaScript library of the adjusted PV station 3D model and set a standardized API interface; Collect information data of the photovoltaic station, clean, convert and structure the information data, and associate it with the three-dimensional model of the photovoltaic station through the API interface to obtain a real-time three-dimensional model of the photovoltaic station; the information data includes basic equipment data, equipment operation data and equipment fault status of the photovoltaic station; A 3D scene is created in the GIS platform, the real-time PV station 3D model is loaded into the 3D scene, and 3D rendering is performed using WebGL to obtain a visualization model of the PV station.

[0050] Furthermore, step S6 also includes: A multi-layer layer structure is set for the visualization model of the photovoltaic station, wherein the layer structure includes a device layer, a monitoring layer, and a fault layer; A user-friendly interface is provided in each layer structure, allowing the user to view detailed information of the layer structure.

[0051] It can be understood that the visualization model of the PV station can realize a quick preview of the PV station equipment, allowing the user to directly view the equipment details, such as equipment ID, model, operating status, current output, etc., by clicking on the equipment in the visualization model.

[0052] By integrating the 3D model of the PV station with GIS data, Ben Amin has improved the visualization and intuitiveness of the PV station data, realized real-time data integration and dynamic updating, and greatly enhanced the monitoring, management and analysis capabilities of the PV station. Users can also view and analyze various information of the PV station from all directions and angles on the GIS platform, including equipment status, operating data and spatial distribution.

[0053] Embodiment 2: The present invention also provides a photovoltaic station three-dimensional model and GIS deep fusion system, which adopts the photovoltaic station three-dimensional model and GIS deep fusion method as described above, including: A data processing module, which is configured to collect multi-source data of the photovoltaic station using a data acquisition device and pre-process the multi-source data; wherein the multi-source data includes point cloud data and image data; A feature point extraction module is configured to perform feature extraction on the processed image data to obtain feature points, and perform feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station; A panoramic stitching module, configured to perform panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; A model generation module is configured to extract three-dimensional point cloud data from the panoramic image, perform fine processing on the three-dimensional point cloud data, and generate a three-dimensional model of the photovoltaic station in combination with the panoramic image; A bidirectional mapping module is configured to convert the data format of the photovoltaic station three-dimensional model into a format compatible with GIS, and adjust the coordinates to perform bidirectional mapping between the GIS data and the photovoltaic station three-dimensional model; The visualization module is configured to perform technical fusion and real-time data integration of the adjusted PV station three-dimensional model with the GIS platform based on bidirectional mapping to form a multi-dimensional visualization model of the PV station.

[0054] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for deep integration of a photovoltaic station three-dimensional model and GIS, characterized in that: The following steps are involved: S1. Collect multi-source data of the photovoltaic station using data acquisition equipment and pre-process the multi-source data; The multi-source data includes point cloud data and image data; S2, extracting features from the processed image data to obtain feature points, performing feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station; S3, performing panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; S4, extracting three-dimensional point cloud data from the panoramic image, performing fine processing on the three-dimensional point cloud data, and generating a three-dimensional model of the photovoltaic station in combination with the panoramic image; S5. Convert the data format of the PV station three-dimensional model into a format compatible with GIS, adjust the coordinates, and perform bidirectional mapping between the GIS data and the PV station three-dimensional model; S6. Based on bidirectional mapping, the adjusted PV station three-dimensional model is technically integrated with the GIS platform and real-time data is integrated to form a multi-dimensional visualization model of the PV station.

2. A photovoltaic station three-dimensional model and GIS deep integration method according to claim 1, characterized in that: In step S2, feature extraction includes: An adaptive scale parameter is calculated according to the area of ​​the photovoltaic panel of the photovoltaic station and the area of ​​the processed image data, a Gaussian pyramid is constructed using the adaptive scale parameter, and the difference of adjacent Gaussian images is calculated to generate a Gaussian difference pyramid; Detect and locate key points in the Gaussian difference pyramid to determine the location of key points; Construct a circular area around the key point, calculate the gradient magnitude and direction of each pixel in the circular area, and calculate the weight of each direction, multiply the gradient magnitude by the weight of the corresponding direction, and add them to the histogram. Determine the main direction of the key point based on the peak value in the histogram; Taking the key point as the center, select the neighborhood and divide it into sub-regions. Calculate the gradient histogram of the main direction of the key point in each sub-region to generate a feature description vector. The feature description vector is normalized to obtain feature points.

3. A photovoltaic station three-dimensional model and GIS deep integration method according to claim 2, characterized in that: The calculation formula of the adaptive scale parameter is: ； in, represents the basic scale parameter, represents the photovoltaic panel area of ​​the photovoltaic station, Represents the area of ​​the processed image data.

4. The method for deep integration of a photovoltaic station three-dimensional model and GIS according to claim 1, characterized in that: The step S4 comprises the following steps: S41, generating a depth map according to the panoramic image, and converting the depth map into three-dimensional point cloud data, retaining the color information of each point; S42, denoising the three-dimensional point cloud data, and performing point cloud registration with the processed point cloud data to obtain comprehensive point cloud data; S43, performing feature point matching on the panoramic images of different viewing angles, and generating a sparse point cloud and a dense point cloud, and performing point cloud fusion on the sparse point cloud and the dense point cloud to obtain fused point cloud data; S44, converting the fused point cloud data into a triangular mesh, and smoothing the triangular mesh to obtain a three-dimensional model of the photovoltaic station.

5. A photovoltaic station three-dimensional model and GIS deep integration method according to claim 4, characterized in that: In step S43, the point cloud fusion operation is as follows: Use the coarse registration method to align the coordinates of the sparse point cloud and the dense point cloud. For each point pi in the sparse point cloud P, find the nearest corresponding point qi in the dense point cloud Q. Calculate the center of mass and center of mass coordinates of pi and qi, and determine the covariance matrix of the sparse point cloud and the dense point cloud according to the center of mass coordinates; Decompose the covariance matrix of the sparse point cloud and the dense point cloud, calculate the rotation matrix and the translation vector, obtain the transformation matrix according to the rotation matrix and the translation matrix, and apply the transformation matrix to the sparse point cloud to obtain the transformed sparse point cloud; Determine the overlapping area of ​​the sparse point cloud and the dense point cloud according to the distance between the transformed sparse point cloud and the dense point cloud, assign weights to each point in the overlapping area according to the reliability, merge the overlapping areas according to the weights, and directly merge the sparse point cloud and the dense point cloud in the non-overlapping area; The merged point cloud is deduplicated and filtered, and the fused point cloud is smoothed to obtain fused point cloud data.

6. A photovoltaic station three-dimensional model and GIS deep integration method according to claim 4, characterized in that: The step S44 also includes optimizing the resolution of the triangular mesh, as follows: Reconstruct the equipment elements in the 3D model of the photovoltaic station, and use an adaptive mesh subdivision algorithm to increase the mesh density to an average side length of no more than 2 cm; The non-equipment factors in the 3D model of the photovoltaic station were simplified, and the grid density was adjusted to an average side length of no more than 10 cm using a grid downsampling algorithm.

7. The method for deep integration of a photovoltaic station three-dimensional model and GIS according to claim 1, characterized in that: The step S5 specifically includes: Convert the PV station 3D model into geojson format and perform coordinate system conversion to WGS84 coordinates; Establish spatial index for GIS data and PV station 3D model, create mapping table, and establish attribute mapping mechanism for GIS data and PV station 3D model; Listeners are set for GIS data and PV station three-dimensional models respectively. When the listener detects changes in GIS data and / or changes in the data of the PV station three-dimensional model, the listener is triggered to synchronize and update the data.

8. The method for deep integration of a photovoltaic station three-dimensional model and GIS according to claim 1, characterized in that: The step S6 comprises: Encapsulate the 3D JavaScript library of the adjusted PV station 3D model and set a standardized API interface; Collect the information data of the photovoltaic station, clean, convert the format and structure the information data, associate it with the three-dimensional model of the photovoltaic station through the API interface, and obtain the real-time three-dimensional model of the photovoltaic station; A 3D scene is created in the GIS platform, the real-time PV station 3D model is loaded into the 3D scene, and 3D rendering is performed using WebGL to obtain a visualization model of the PV station.

9. A photovoltaic station three-dimensional model and GIS deep integration method according to claim 8, characterized in that: The step S6 further comprises: A multi-layer layer structure is set for the visualization model of the photovoltaic station; the layer structure includes a device layer, a monitoring layer and a fault layer; A user-friendly interface is provided in each layer structure, allowing the user to view detailed information of the layer structure.

10. A photovoltaic station three-dimensional model and GIS deep integration system, characterized in that: A multi-dimensional visualization model of a photovoltaic station is formed by using the photovoltaic station three-dimensional model and GIS deep integration method as described in any one of claims 1 to 9, including: A data processing module configured to collect multi-source data of the photovoltaic station using a data acquisition device and pre-process the multi-source data; A feature point extraction module is configured to perform feature extraction on the processed image data to obtain feature points, and perform feature matching and feature point filtering on the feature points to obtain a feature point set of the photovoltaic station; A panoramic stitching module, configured to perform panoramic stitching based on the feature point set to generate a panoramic image of the photovoltaic station; A model generation module is configured to extract three-dimensional point cloud data from the panoramic image, perform fine processing on the three-dimensional point cloud data, and generate a three-dimensional model of the photovoltaic station in combination with the panoramic image; A bidirectional mapping module is configured to convert the data format of the photovoltaic station three-dimensional model into a format compatible with GIS, and adjust the coordinates to perform bidirectional mapping between the GIS data and the photovoltaic station three-dimensional model; The visualization module is configured to perform technical fusion and real-time data integration of the adjusted PV station three-dimensional model with the GIS platform based on bidirectional mapping to form a multi-dimensional visualization model of the PV station.

Images