Photovoltaic array shadow ecological evaluation method and system based on single board geometry inversion

Abstract

The application discloses a photovoltaic array shadow ecological evaluation method and system based on single-board geometric inversion. The method extracts two-dimensional attributes of single photovoltaic boards based on remote sensing images, combines three-dimensional surface data and a digital terrain model to perform candidate board point set extraction, ground surface constraint filtering, plane fitting purification and array grouping statistical smoothing, inverts board reference elevation, inclination and azimuth, compares the board uphill direction with the rectangular direction to restore the actual size of the inclination direction, and constructs a single-board geometric parameter dataset; according to the dataset, ground surface pixel space shielding is determined, and the accumulated photosynthetic effective radiation reduction is calculated. The system comprises a basic data acquisition and coordinate unification module, a single-board identification and geometric inversion module, a shadow distribution generation module and an ecological evaluation result output module. The application realizes single-board geometric parameter-driven shadow dynamic quantification and ecological evaluation result output.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of new energy and ecological environment, specifically to a method and system for ecological assessment of photovoltaic array shadows based on single-plate geometric inversion. Background Technology

[0002] The dynamic quantification of photovoltaic array shadows and the assessment of ecological effects involve remote sensing identification, three-dimensional geometric reconstruction, and radiation quantification calculation. In existing technologies, one approach focuses on analyzing surface irradiance, shadow occupancy, and spatial distribution under conditions considering weather changes and cloud cover; another approach focuses on estimating the contour, slope, and height of target surface components based on imagery and three-dimensional data to reconstruct a three-dimensional model. These approaches provide a certain technical foundation for radiation analysis or three-dimensional structure estimation. However, in the ecological assessment of photovoltaic array shadows based on remote sensing inversion of single-panel geometric parameters for photovoltaic power plant scenarios, existing technologies still have the following problems: First, existing schemes have not established mechanisms for two-dimensional contour extraction, panel reference elevation inversion, tilt and azimuth inversion, and three-dimensional coordinate construction of the four corner points for individual photovoltaic panels, making it difficult to directly generate a single-panel geometric parameter dataset for determining spatial shading at the surface pixel level. Second, existing schemes have not established a complete processing chain for extracting candidate panel point sets, surface constraint filtering, panel fitting and purification, array grouping statistical smoothing, and panel tilt direction dimension recovery for individual photovoltaic panels in photovoltaic power plants, nor have they further combined solar position parameters, shadow status indicators, and scattered radiation ratio coefficients to calculate the spatial distribution map of cumulative photosynthetically active radiation reduction. Therefore, it is necessary to propose a photovoltaic array shadow ecological assessment method and system based on single-panel geometric inversion to achieve integrated processing of single-panel geometric parameter inversion, time-series shadow distribution generation, and ecological assessment result output. Summary of the Invention

[0003] The purpose of this invention is to solve the technical problems mentioned above and to propose a photovoltaic array shadow ecology assessment method based on single-panel geometric inversion, including the following steps: S1. Acquire high-resolution remote sensing images, 3D surface data, and digital terrain models of the target area, and unify the high-resolution remote sensing images, 3D surface data, and digital terrain models into the same plane coordinate system; based on the high-resolution remote sensing images, extract the vector contour, 2D center point coordinates, 2D projection length, 2D projection width, and long side orientation angle of each photovoltaic panel in the photovoltaic power station area; S2. Based on the vector contour of each individual photovoltaic panel, perform polygon interior point retrieval on the 3D surface data to obtain the original candidate panel point set; extract the ground elevation corresponding to the position of each point in the original candidate panel point set based on the digital terrain model and calculate the ground height, and remove ground points with ground height less than the preset lower threshold and abnormal high points with ground height greater than the preset upper threshold; perform plane fitting purification processing on the filtered point set, remove the external points that do not meet the same plane model and retain the interior point set as the panel fitting point set; invert the panel reference elevation, initial tilt angle and initial spatial azimuth angle based on the panel fitting point set, and obtain the final tilt angle and final azimuth angle by combining the array group of each individual photovoltaic panel; determine the projection of the uphill direction of the panel on the horizontal plane according to the final azimuth angle, compare the uphill direction of the panel with the direction of the rectangle determined by the direction angle of the long side of the rectangle, determine the dimensions corresponding to the tilt direction of the panel in the 2D projection length and 2D projection width, and restore the dimensions corresponding to the tilt direction of the panel to the actual physical dimensions along the tilt direction of the panel according to the final tilt angle, generate the 3D coordinates of the four corner points of each individual photovoltaic panel, and form a single panel geometric parameter dataset; S3. Based on the solar position parameters, digital terrain model and single-board geometric parameter dataset at each moment within the preset simulation period, perform spatial occlusion determination based on solar direction on the surface pixels of the target area to obtain the time series shadow distribution results. S4. Based on the time-series shadow distribution results and the solar position parameters at each time point, calculate the instantaneous photosynthetically active radiation of the surface pixels under unshaded conditions, and calculate the actual photosynthetically active radiation under shaded conditions by combining the shadow status indicators and the scattered radiation ratio coefficient at each time point. The instantaneous photosynthetically active radiation reduction is obtained by the difference between the instantaneous photosynthetically active radiation under unshaded conditions and the actual photosynthetically active radiation under shaded conditions. The instantaneous photosynthetically active radiation reduction is accumulated or integrated over the simulation period to obtain the spatial distribution map of the cumulative photosynthetically active radiation reduction as the ecological assessment result.

[0004] In the preferred embodiment, step S1 involves extracting the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and the orientation angle of the long side of the rectangle for each individual photovoltaic panel within the photovoltaic power station area, including: A multi-feature set is constructed based on high-resolution remote sensing images, and pixel-level classification is performed on the multi-feature set to obtain preliminary photovoltaic identification results. The preliminary photovoltaic identification results are processed by region merging and shape filtering to obtain the polygon of the photovoltaic power station area. Within the polygonal area of ​​the photovoltaic power station region, instance segmentation is performed on the image to obtain a single photovoltaic panel instance mask; Connectivity separation, polygon fitting, and minimum circumscribed rectangle fitting are performed on the mask of each individual photovoltaic panel instance to obtain the vector profile, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and long side orientation angle of the rectangle for each individual photovoltaic panel.

[0005] In the preferred embodiment, step S2, which involves performing planar fitting and purification processing to obtain a set of fitting points on the board surface, includes: Using the vector contour of each individual photovoltaic panel as the two-dimensional query boundary, polygon interior point retrieval is performed on the three-dimensional surface data to extract three-dimensional points whose planar coordinates fall inside the vector contour, forming the original candidate panel surface point set; Extract the surface elevation corresponding to each point in the original candidate panel point set from the digital terrain model, calculate the height above the ground, and remove ground points with a height above the ground that is less than a preset lower threshold and abnormal high points with a height above the ground that is greater than a preset upper threshold. The point set after surface constraint filtering is further fitted using a random sampling consensus algorithm. External points that do not satisfy the same plane model are removed, and the internal point set is retained as the plate fitting point set.

[0006] In the preferred embodiment, step S2 involves retrieving the reference elevation of the photovoltaic panel, the initial tilt angle, and the initial spatial azimuth angle, and combining this with the array grouping of each individual photovoltaic panel to obtain the final tilt angle and the final azimuth angle, including: Perform the best-fit plane calculation on the set of fitted points on the plate surface to obtain the plate surface normal vector and the centroid of the fitted plane. Substitute the coordinates of the two-dimensional center point into the fitted plane equation to solve for the reference elevation of the plate surface. The initial tilt angle is determined by the angle between the plate surface normal vector and the vertical direction, and the initial spatial azimuth angle is determined by the projection direction of the plate surface normal vector onto the horizontal plane. Based on the spatial proximity of each individual photovoltaic panel, the similarity of its two-dimensional projection size, and the similarity of the orientation angle of its long side, array grouping is performed on each individual photovoltaic panel; Outlier removal and statistical smoothing are performed on the initial tilt angle and initial spatial azimuth angle according to the array group to obtain the final tilt angle and final azimuth angle.

[0007] In the preferred embodiment, step S2 involves generating the three-dimensional coordinates of the four corner points of each individual photovoltaic panel, including: The projection of the upslope direction of the plate onto the horizontal plane is determined based on the final azimuth angle. The upslope direction of the plate is then compared with the direction of the long side of the rectangle determined by the direction angle of the long side of the rectangle and its orthogonal direction. The dimensions corresponding to the tilt direction of the plate and the dimensions corresponding to the transverse direction of the plate are then determined in the two-dimensional projection length and two-dimensional projection width. The dimensions corresponding to the tilt direction of the plate are restored to the actual physical dimensions along the tilt direction of the plate, and the center point of the plate is constructed based on the plate reference elevation and the coordinates of the two-dimensional center point. Based on the center point of the panel, the actual physical dimensions along the tilt direction of the panel, and the transverse dimensions of the panel, construct the three-dimensional coordinates of the four corner points of each individual photovoltaic panel; The average elevation of the two corner points on the lower side is determined as the bottom elevation, and the average elevation of the two corner points on the upper side is determined as the top elevation.

[0008] In the preferred embodiment, step S3 involves performing a spatial occlusion determination based on the solar direction for the surface pixels of the target area, including: Construct the solar direction vector based on the solar altitude angle and solar azimuth angle at each moment; The boundary points of the three-dimensional rectangular surface of each individual photovoltaic panel are projected onto the surface reference surface corresponding to the digital terrain model in the opposite direction of solar incidence, forming the candidate shadow envelope range of each individual photovoltaic panel at the corresponding time, and a spatial index structure is established based on the candidate shadow envelope range of each individual photovoltaic panel. Based on the planar position of the surface pixels, the candidate shading plate set is queried in the spatial index structure. The intersection of the line of sight and the three-dimensional rectangular plate surface is determined only for the individual photovoltaic panels in the candidate shading plate set. The shadow state raster at each time moment is generated and the time series shadow distribution result is formed.

[0009] This invention also provides a photovoltaic array shading ecological assessment system based on single-plate geometry inversion, comprising: The basic data acquisition and coordinate unification module is used to acquire high-resolution remote sensing images, 3D surface data and digital terrain models of the target area, and unify the high-resolution remote sensing images, 3D surface data and digital terrain models into the same plane coordinate system; The module for identifying individual photovoltaic panels and extracting two-dimensional attributes is used to extract the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and rectangular long side orientation angle of each individual photovoltaic panel in the photovoltaic power station area based on high-resolution remote sensing images. The candidate panel point set extraction and surface constraint filtering module is used to extract candidate panel point sets based on the vector contour, three-dimensional surface data and digital terrain model of each individual photovoltaic panel. It calculates the ground height based on the digital terrain model and removes ground points and abnormal high points. It performs plane fitting purification processing on the filtered point set, and retains the inner point set as the panel fitting point set after removing outliers. The single-panel geometric parameter inversion and 3D panel construction module is used to invert the panel reference elevation, initial tilt angle and initial spatial azimuth angle based on the panel fitting point set. Combined with the array grouping of each individual photovoltaic panel, the final tilt angle and final azimuth angle are obtained. Based on the final azimuth angle, the projection of the uphill direction on the horizontal plane of the panel is determined. The uphill direction on the panel is compared with the rectangular direction to determine the dimension corresponding to the tilt direction of the panel in the 2D projection dimension. Based on the final tilt angle, the actual physical dimension along the tilt direction of the panel is restored, and the 3D coordinates of the four corner points of each individual photovoltaic panel are generated to form a single-panel geometric parameter dataset. The solar position parameter calculation module is used to calculate the solar position parameters at each time point within a preset simulation period. The time-series shadow distribution generation module is used to perform spatial occlusion determination based on the sun direction for surface pixels in the target area based on solar position parameters, digital terrain model and single-board geometric parameter dataset, and obtain time-series shadow distribution results. The module for calculating and evaluating the cumulative photosynthetically active radiation reduction (CARERR) outputs a spatial distribution map of CARERR as an ecological assessment result, based on time-series shadow distribution results and solar position parameters.

[0010] In the preferred embodiment, the single photovoltaic panel identification and two-dimensional attribute extraction module includes a photovoltaic power station area extraction unit, an instance segmentation unit, and a two-dimensional attribute extraction unit; The photovoltaic power station area extraction unit constructs a multi-feature set based on high-resolution remote sensing images, performs pixel-level classification, region merging, and shape filtering on the multi-feature set, and outputs a polygon of the photovoltaic power station area. The instance segmentation unit performs instance segmentation on the image within the polygonal area of ​​the photovoltaic power station region and outputs a single photovoltaic panel instance mask. The two-dimensional attribute extraction unit performs connected component separation, polygon fitting, and minimum bounding rectangle fitting on the mask of a single photovoltaic panel instance, and outputs the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and long side orientation angle of the rectangle for each single photovoltaic panel.

[0011] In the preferred embodiment, the single-board geometric parameter inversion and 3D board surface construction module includes an array grouping unit, an attitude inversion unit, an attitude smoothing unit, and a 3D board surface construction unit; The array grouping unit performs array grouping on each individual photovoltaic panel based on the spatial proximity relationship, similarity of two-dimensional projection size, and similarity of the orientation angle of the long side of the rectangle, and outputs the array grouping result; The attitude inversion unit performs best-fit plane calculation based on the plate surface fitting point set to obtain the plate surface normal vector and the centroid of the fitting plane, and calculates the plate surface reference elevation, initial tilt angle and initial spatial azimuth angle based on the fitting plane equation; The attitude smoothing unit performs outlier removal and statistical smoothing on the initial tilt angle and initial spatial azimuth angle by array grouping, and outputs the final tilt angle and final azimuth angle; The three-dimensional panel construction unit determines the tilt direction of the panel based on the final tilt angle, final azimuth angle, direction angle of the long side of the rectangle, coordinates of the two-dimensional center point, and two-dimensional projection dimensions, restores the actual physical dimensions along the tilt direction of the panel, and generates the three-dimensional coordinates of the four corner points of each individual photovoltaic panel.

[0012] In the preferred embodiment, the time-series shadow distribution generation module includes a solar direction construction and candidate shading range generation unit and a shadow state determination unit, and the cumulative photosynthetically effective radiation reduction calculation and evaluation result output module includes a radiation calculation unit and a cumulative evaluation unit. The solar direction construction and candidate shading range generation unit constructs a solar direction vector based on the solar elevation angle and solar azimuth angle at each moment, and projects the boundary points of the three-dimensional rectangular panel of each individual photovoltaic panel onto the surface reference surface corresponding to the digital terrain model along the opposite direction of solar incidence, forming the surface candidate shadow envelope range and establishing a spatial index structure. The shadow state determination unit queries the candidate shading plate set in the spatial index structure based on the planar position of the surface pixels, and performs the intersection determination of the line of sight and the three-dimensional rectangular plate surface only for the individual photovoltaic panels in the candidate shading plate set, and outputs the time series shadow distribution results; The radiation calculation unit calculates the instantaneous photosynthetically active radiation under unshaded conditions based on the time series shadow distribution results and solar position parameters, and calculates the actual photosynthetically active radiation under shaded conditions by combining the shadow status indicator and the scattered radiation ratio coefficient, and then calculates the instantaneous photosynthetically active radiation reduction. The cumulative evaluation unit sums or integrates the instantaneous reduction in photosynthetically active radiation at each moment during the simulation period and outputs a spatial distribution map of the cumulative reduction in photosynthetically active radiation.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention unifies the coordinate processing of high-resolution remote sensing images, 3D surface data, and digital terrain models. It then combines this with 2D contour extraction of individual photovoltaic panels, surface constraint filtering of candidate panel point sets, panel fitting, and array grouping statistical smoothing to obtain the reference elevation, final tilt angle, final azimuth angle, and 3D coordinates of the four corner points for each individual photovoltaic panel, forming a single-panel geometric parameter dataset. Based on this, a solar direction vector is constructed according to the solar position parameters at each time point. The boundary points of the 3D rectangular panel surface of the individual photovoltaic panel are projected onto the surface reference plane along the opposite direction of solar incidence to form candidate shadow envelope ranges. Based on the candidate shadow envelope range, a spatial index structure is established to limit the spatial occlusion determination of surface pixels to the candidate occlusion plate set, thereby obtaining the time series shadow distribution results. The present invention further calculates the instantaneous photosynthetically active radiation under unshaded conditions, the actual photosynthetically active radiation under shaded conditions, and the instantaneous photosynthetically active radiation reduction based on the time series shadow distribution results, solar position parameters, and scattered radiation ratio coefficient. The instantaneous photosynthetically active radiation reduction is accumulated or integrated within the simulated period, and the spatial distribution map of the cumulative photosynthetically active radiation reduction is output as the ecological assessment result. Attached Figure Description

[0014] Figure 1 The flowchart shows the photovoltaic array shadow ecological assessment method based on single-board geometric inversion. Figure 2 This is a schematic diagram of projective geometry; Figure 3 This is a schematic diagram of the time series shaded distribution results; Figure 4 This is a structural diagram of a photovoltaic array shadow ecological assessment system based on single-plate geometric inversion. Detailed Implementation

[0015] Example 1 like Figure 1 As shown, this embodiment takes a large-scale ground-mounted fixed-tilt photovoltaic power station as the object, acquires high-resolution remote sensing images, three-dimensional surface data, and a digital terrain model (DTM) of the target area, extracts the two-dimensional contour of individual photovoltaic panels based on the high-resolution remote sensing images, inverts the spatial attitude and reference elevation of individual photovoltaic panels based on the three-dimensional surface data, generates the time-series shadow distribution of the photovoltaic array on the ground surface based on the digital terrain model, and quantifies the reduction of photosynthetically active radiation to obtain the shadow ecological assessment results, including the following steps: S101, Basic Data Acquisition and Coordinate Unification High-resolution remote sensing imagery, 3D surface data, and a digital terrain model of the target area are acquired. The high-resolution remote sensing imagery is preferably orthorectified and georegistered UAV imagery or high spatial resolution satellite imagery, used for individual photovoltaic panel identification and 2D contour extraction. The 3D surface data is preferably airborne LiDAR point cloud, UAV real-scene 3D point cloud, or a digital surface model generated from the point cloud, used to invert the tilt angle, azimuth angle, and reference elevation of individual photovoltaic panels. The digital terrain model is used to characterize surface undulations and for surface constraint filtering of the candidate point set for individual panels and subsequent determination of surface shading status.

[0016] Coordinate system checks are performed on high-resolution remote sensing images, 3D surface data, and digital terrain models. When the coordinate systems of the three are inconsistent, they are uniformly reprojected to the same plane coordinate system so that the single-board outline, board surface points, and ground elevation satisfy a consistent overlay relationship.

[0017] S102, Photovoltaic Power Station Area Extraction A multi-feature set is constructed based on high-resolution remote sensing imagery. This set includes at least red, green, blue, and near-infrared bands, normalized vegetation index (NDI), and texture features. A classification model is trained using labeled samples, and pixel-level classification is performed on the multi-feature set to obtain preliminary photovoltaic (PV) identification results. Region merging is then performed on these preliminary PV identification results, and shape selection is conducted using aspect ratio and area-to-perimeter ratio to obtain the PV power plant region polygon. This polygon is used to define the subsequent identification range for individual PV panels.

[0018] S103, Individual Photovoltaic Panel Identification and Two-Dimensional Attribute Extraction Instance segmentation is performed on the image within the photovoltaic power station area to obtain individual photovoltaic panel instance masks. Connectivity separation, polygon fitting, and minimum circumscribed rotation rectangle fitting are then performed on the individual photovoltaic panel instance masks to obtain the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and long side orientation angle of each individual photovoltaic panel.

[0019] For any single photovoltaic panel, the coordinates of the center point of its minimum circumscribed rectangle are denoted as . The length of the longer side is denoted as The length of the shorter side is denoted as The clockwise angle between the longer side and the true north direction of the plane coordinate system is denoted as... . and The two orthogonal dimensions used to characterize the rectangle do not presuppose that either dimension necessarily corresponds to the tilt direction of the plate surface.

[0020] Establish attribute records for each individual photovoltaic panel. The attribute records should include at least the panel number, array number, coordinates of the two-dimensional center point, two-dimensional projection length, two-dimensional projection width, direction angle of the long side of the rectangle, final azimuth angle, final tilt angle, reference elevation of the panel surface, bottom elevation, top elevation, and three-dimensional coordinate fields of the four corner points.

[0021] S104, Array Grouping Based on the spatial proximity of individual photovoltaic panels, the similarity of their two-dimensional projected dimensions, and the similarity of the orientation angles of their long sides, clustering is performed on each individual photovoltaic panel to form array groups. The planar distance between the center points of any two individual photovoltaic panels is calculated, and the differences in their two-dimensional projected length and width are compared, along with the differences in the orientation angles of their long sides. When both distances and differences fall within a preset threshold range, the corresponding individual photovoltaic panels are grouped into the same array. Array grouping is used for subsequent outlier removal and statistical smoothing of tilt and azimuth angles.

[0022] S105, Candidate Plate Surface Point Set Extraction and Surface Constraint Filtering Using the vector contour of each individual photovoltaic panel as the two-dimensional query boundary, polygon interior point retrieval is performed on the three-dimensional surface data of the target area to extract three-dimensional points whose planar coordinates fall inside the vector contour, forming the original candidate panel point set for the corresponding individual photovoltaic panel. A spatial index structure is established on the three-dimensional surface data to perform polygon interior point retrieval. The spatial index structure is preferably a KD-tree index, a two-dimensional raster index, or a local 2.5D index structure.

[0023] Perform surface constraint filtering on the original candidate plate surface point set. For any point in the original candidate plate surface point set... Extracting surface elevation at the same location from the digital terrain model Calculate the height above the ground As shown in equation (1): (1) Ground points with a height below a preset lower threshold and abnormally high points with a height above a preset upper threshold are removed. The point set after surface constraint filtering is further fitted using a random sampling consensus algorithm, removing external points that do not conform to the same planar model and retaining the internal point set as the panel fitting point set for that individual photovoltaic panel. Addressing the technical issue of interference from ground points, support points, and local noise points in the candidate panel point set, this step employs a combination of digital terrain model constraints and the random sampling consensus algorithm to achieve panel fitting point set purification.

[0024] S106, Plate surface normal vector, plate surface reference elevation, and initial spatial attitude inversion For each individual photovoltaic panel, the best-fit plane is calculated using the fitted point set on the panel surface to obtain the corresponding panel surface normal vector. Let the fitted point set on the panel surface of a certain individual photovoltaic panel be... Its center of mass is Construct the covariance matrix As shown in equation (2): (2) For covariance matrix Perform eigenvalue decomposition and extract the eigenvector corresponding to the smallest eigenvalue. As the initial plate surface normal vector. When At that time, the normal vector of the plate surface is flipped to .when When the value is less than the preset minimum value, the corresponding photovoltaic panel will be marked as an abnormal attitude panel, and the array dominant attitude parameter will be used to backfill the value in the subsequent array statistical smoothing stage.

[0025] Assume the fitted plane passes through the centroid. Then the plane equation of the plate is shown in equation (3): (3) The coordinates of the two-dimensional center point obtained in the two-dimensional recognition stage Substituting into equation (3), the reference elevation of the photovoltaic panel surface is calculated. As shown in equation (4): (4) Slab reference elevation Corresponding to the coordinates of the two-dimensional center point Elevation position on the fitted plane.

[0026] Unit vector in the vertical direction For reference, the initial tilt angle is determined based on the angle between the normal vector and the vertical direction. As shown in equation (5): (5) The projection direction of the normal vector of the plate onto the horizontal plane is used as the spatial reference direction of the light-receiving surface to obtain the initial spatial azimuth angle. .when hour, As shown in equation (6): (6) when When, add it Normalization to Within the range. When When the dominant azimuth angle of other individual photovoltaic panels in the same array is used as the initial spatial azimuth angle of that individual photovoltaic panel; when all individual photovoltaic panels in the corresponding array meet the near-horizontal condition, the orientation angle of the long side of the rectangle is... and its orthogonal direction As a candidate direction, the direction of due south was selected. The candidate direction with the smaller absolute value of the included angle is used as the temporary initial spatial azimuth.

[0027] S107, Final tilt angle, final azimuth angle and single-plate 3D panel construction Initial tilt angle of each individual photovoltaic panel grouped by array and initial spatial azimuth Outlier removal and statistical smoothing are performed to obtain the final tilt angle of each individual photovoltaic panel. and final azimuth For boards with abnormal attitudes, array-dominated attitude parameters are used for backfilling.

[0028] According to the final azimuth Determine the unit projection vector of the normal vector of the illuminated surface onto the horizontal plane. As shown in equation (7): (7) Take its opposite direction as the projection of the upslope direction of the board onto the horizontal plane. As shown in equation (8): (8) Based on the direction angle of the long side of the rectangle Construct the horizontal unit vector of the longer side of the rectangle As shown in equation (9): (9) Construction and The vertical unit vector of the shorter side of the rectangle. As shown in equation (10): (10) Calculate separately and .when When determining the tilt direction of the rectangular long side corresponding to the plate surface, the tilt direction is a horizontal unit vector. As shown in equation (11): (11) Horizontal unit vector As shown in equation (12): (12) Dimensions in the tilt direction The horizontal dimension is taken as .

[0029] when When determining the tilt direction of the rectangular short side corresponding to the plate surface, the tilt direction is a horizontal unit vector. As shown in equation (13): (13) Horizontal unit vector As shown in equation (14): (14) Dimensions in the tilt direction The horizontal dimension is taken as To address the technical issue that the long side of the rectangle does not correspond to a fixed tilt direction of the board surface, this step adopts a determination method based on comparing the projection of the rectangle with the direction of the slope, thus realizing the identification of the tilt direction under different installation orientations.

[0030] After determining the tilt direction, construct a three-dimensional unit vector along the plate surface from the lower edge to the higher edge. As shown in equation (15): (15) Construct the horizontal three-dimensional unit vector of the plate surface As shown in equation (16): (16) The tilt direction dimension is restored to the actual physical dimension along the tilt direction of the plate surface. As shown in equation (17): (17) in, The corresponding horizontal direction is parallel to the horizontal plane, and its actual physical size is consistent with its two-dimensional projection size. When At that time, the corresponding photovoltaic panel is marked as a panel with abnormal size recovery, and the three-dimensional panel construction is performed after backfilling with the array dominant tilt angle parameter, or the photovoltaic panel is removed from the panel-level geometric reconstruction at the current moment.

[0031] Let the center point of the board be The three-dimensional coordinates of the four corner points are shown in equations (18) to (21) respectively: (18) (19) (20) (twenty one) in, and Located on the lower side, and Located on the high side. Bottom elevation As shown in equation (22): (twenty two) Top elevation As shown in equation (23): (twenty three) The final azimuth angle, final tilt angle, reference elevation of the panel surface, bottom elevation, top elevation, and 3D coordinates of the four corner points are backfilled into the attribute record of the corresponding individual photovoltaic panel to form a single-panel geometric parameter dataset. This single-panel geometric parameter dataset serves as the geometric input for subsequent determination of the surface shading state.

[0032] S108, Calculation of Solar Position Parameters Set a preset simulation period and time step. For each moment within the preset simulation period, calculate the solar altitude angle based on the latitude, longitude, date, and time of the target area's center point. and solar azimuth This forms a time series of solar position parameters.

[0033] Rectangle long side orientation angle Final azimuth and solar azimuth A unified direction reference under the same plane coordinate system is adopted; when the solar azimuth is calculated from the true north direction, the solar azimuth is converted to the direction reference corresponding to the plane coordinate system before participating in the subsequent shadow calculation.

[0034] S109, Generation of Time Series Shadow Distribution For any moment within the preset simulation period Read the solar position parameters at that moment and the three-dimensional coordinates of the four corner points of each individual photovoltaic panel, and then calculate based on the solar altitude angle. and solar azimuth Construct the solar direction vector Using the three-dimensional rectangular surface determined by the three-dimensional coordinates of the four corner points of each individual photovoltaic panel as the shading body, and the surface pixels in the digital terrain model as the objects to be judged, spatial shading judgment based on the solar direction is performed on each surface pixel in the target area.

[0035] Based on the solar direction vector at the current moment Candidate shading ranges are generated for each individual photovoltaic panel. The boundary points of the 3D rectangular surface of each individual photovoltaic panel are projected onto the corresponding surface reference plane of the digital terrain model along the opposite direction of solar incidence, forming the shading range of that individual photovoltaic panel at time [time value missing]. The candidate shadow envelope extent on the ground, as shown in the diagram. Figure 2 As shown, a spatial index structure is established based on the candidate shadow envelope range of each individual photovoltaic panel on the ground surface. The spatial index structure is preferably a two-dimensional raster index, an R-tree index, or a regular block index.

[0036] For any surface pixel Based on the planar location of the surface pixel, the candidate masking set is queried in the spatial index structure; when the candidate masking set is empty, the surface pixel is determined to be an empty pixel. At any moment In a non-shaded state; when the candidate occlusion set is not empty, surface pixels are constructed along the opposite direction of solar incidence. The line of sight is determined, and the intersection between the line of sight and the three-dimensional rectangular panel is performed only for individual photovoltaic panels in the candidate shading panel set. If an intersection exists, the surface pixel is determined. At any moment If the target area is in shadow, it is considered to be in non-shadow state; otherwise, it is considered to be in non-shadow state. After determining all surface pixels in the target area, the time is generated. Shadow state grid The above process is repeated for all times within the preset simulation period to obtain the time series shadow distribution results, as illustrated below. Figure 3 As shown.

[0037] S110, Output of calculation and evaluation results for cumulative photosynthetically active radiation reduction. Calculate the normal direct irradiance at each time point based on the solar position parameters. Based on the normal direct irradiance and solar altitude angle, the instantaneous effective photosynthetic radiation of the surface pixel at the corresponding time is calculated under unobstructed conditions. As shown in equation (24): (twenty four) in, This is the radiation conversion factor.

[0038] Based on the time-series shading distribution results, the actual photosynthetically active radiation of surface pixels at corresponding times under shading conditions is calculated. Set time surface pixel The shadow state is When a pixel is in shadow, When a pixel is in a non-shadow state, When considering scattered radiation, As shown in equation (25): (25) in, This is the proportionality coefficient for scattered radiation.

[0039] Based on the difference between the instantaneous photosynthetically active radiation of a surface pixel under unobstructed conditions and the actual photosynthetically active radiation of a surface pixel under shaded conditions, the instantaneous photosynthetically active radiation reduction of the surface pixel at the corresponding time is calculated. As shown in equation (26): (26) The cumulative photosynthetically active radiation reduction is obtained by summing or integrating the instantaneous reduction in photosynthetically active radiation at each moment within the simulation period. As shown in equation (27): (27) in, For time step, This represents the total number of time steps within the simulated time period.

[0040] Output a spatial distribution map of the cumulative photosynthetically active radiation reduction of surface pixels in the target area, and use this spatial distribution map as the result of the photovoltaic array shadow ecological assessment.

[0041] To address the technical challenges of dynamic quantification of photovoltaic array shadows and assessment of ecological effects, this embodiment employs a combined processing approach of high-resolution remote sensing imagery, three-dimensional surface data, and digital terrain models. This approach enables the identification of two-dimensional contours of individual photovoltaic panels, inversion of panel geometric parameters, generation of time-series shadow distribution, and quantification of cumulative photosynthetically active radiation reduction, thereby obtaining ecological assessment results of photovoltaic array shadows based on panel geometric inversion.

[0042] Example 2 like Figure 4 As shown, this embodiment provides a photovoltaic array shadow ecological assessment system based on single-panel geometric inversion. The system includes a basic data acquisition and coordinate unification module, a photovoltaic power station area extraction module, a single photovoltaic panel identification and two-dimensional attribute extraction module, an array grouping module, a candidate panel point set extraction and surface constraint filtering module, a panel normal vector, panel reference elevation and initial spatial attitude inversion module, a final tilt angle, final azimuth angle and single-panel three-dimensional panel construction module, a solar position parameter calculation module, a time series shadow distribution generation module, and a cumulative photosynthetically active radiation reduction calculation and assessment result output module.

[0043] The basic data acquisition and coordinate unification module includes: The basic data receiving and coordinate checking unit is used to receive high-resolution remote sensing images, three-dimensional surface data and digital terrain models of the target area, and to perform coordinate system checks on the high-resolution remote sensing images, three-dimensional surface data and digital terrain models, and output the coordinate system check results to the coordinate unification unit.

[0044] The coordinate unification unit is used to receive coordinate system check results, high-resolution remote sensing images, three-dimensional surface data and digital terrain models, and to perform unified reprojection processing on high-resolution remote sensing images, three-dimensional surface data and digital terrain models when the coordinate systems are inconsistent. It outputs high-resolution remote sensing images, three-dimensional surface data and digital terrain models in a unified plane coordinate system to the photovoltaic power station area extraction module, the candidate panel point set extraction and surface constraint filtering module, the solar position parameter calculation module and the time series shadow distribution generation module.

[0045] The photovoltaic power station area extraction module includes: The multi-feature element construction unit is used to receive high-resolution remote sensing images in a unified plane coordinate system and construct a multi-feature element set based on the high-resolution remote sensing images. The multi-feature element set includes at least red, green, blue, and near-infrared bands, normalized vegetation index, and texture features. The multi-feature element set is then output to the classification and region selection unit.

[0046] The classification and region filtering unit receives multi-feature set and labeled samples, trains a classification model using the labeled samples, performs pixel-level classification on the multi-feature set to obtain preliminary photovoltaic identification results, and then performs region merging and shape filtering on the preliminary photovoltaic identification results to output the photovoltaic power station region polygon to the single photovoltaic panel identification and two-dimensional attribute extraction module.

[0047] The module for identifying individual photovoltaic panels and extracting two-dimensional attributes includes: The instance segmentation unit is used to receive the polygon of the photovoltaic power station area and the high-resolution remote sensing image under a unified planar coordinate system, and to perform instance segmentation on the image within the range defined by the polygon of the photovoltaic power station area, and output the individual photovoltaic panel instance mask to the two-dimensional attribute extraction unit.

[0048] The two-dimensional attribute extraction unit receives the mask of a single photovoltaic panel instance and sequentially performs connected component separation, polygon fitting, and minimum bounding rectangle fitting, outputting the vector contour and two-dimensional center point coordinates of each single photovoltaic panel. 2D projection length 2D projection width , Direction angle of the long side of the rectangle It also includes modules for recording attributes to array grouping, extracting candidate plate point sets and surface constraint filtering, and constructing the final tilt angle, final azimuth angle, and single-plate 3D plate surface.

[0049] The array grouping module includes: The array clustering unit receives the two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and rectangular long side orientation angle of each individual photovoltaic panel. It performs clustering based on spatial proximity, similarity of two-dimensional projection size, and similarity of rectangular long side orientation angle, and outputs the array grouping results to the panel surface normal vector, panel surface reference elevation, initial spatial attitude inversion module, final tilt angle, final azimuth angle, and single panel three-dimensional panel surface construction module.

[0050] The candidate plate point set extraction and surface constraint filtering module includes: The candidate point set extraction unit is used to receive the vector contours of each individual photovoltaic panel and the three-dimensional surface data in a unified planar coordinate system. It uses the vector contours of each individual photovoltaic panel as the two-dimensional query boundary to perform polygon interior point retrieval on the three-dimensional surface data and outputs the original candidate panel point set to the ground surface constraint filtering unit.

[0051] The surface constraint filtering unit receives the original candidate plate surface point set and the digital terrain model, extracts the surface elevation corresponding to the position of each point in the original candidate plate surface point set from the digital terrain model, calculates the ground height, removes ground points with ground height less than the preset lower threshold and abnormal high points with ground height greater than the preset upper threshold, then performs a random sampling consensus algorithm to fit the filtered point set, retains the interior point set as the plate surface fitting point set, and outputs the plate surface fitting point set to the plate surface normal vector, plate surface reference elevation and initial spatial attitude inversion module.

[0052] The module for inverting the plate surface normal vector, plate surface reference elevation, and initial spatial attitude includes: The plate surface normal vector calculation unit receives the plate surface fitting point set, performs best-fit plane calculation on the plate surface fitting point set, constructs the covariance matrix and performs eigenvalue decomposition, and outputs the initial plate surface normal vector. Fitted plane centroid And mark the abnormal attitude of the single board to the reference elevation of the board surface and the initial attitude calculation unit.

[0053] The plate surface reference elevation and initial attitude calculation unit is used to receive the initial plate surface normal vector. Fitted plane centroid Two-dimensional center point coordinates Array grouping results and the orientation angle of the long side of the rectangle And calculate the reference elevation of the plate surface based on the fitted plane equation. Then, based on the initial plate surface normal vector unit vector in the vertical direction Calculate the initial tilt angle from the included angle. Based on the initial plate surface normal vector Calculate the initial spatial azimuth angle by projecting it onto the horizontal plane. Output panel reference elevation Initial tilt angle and initial spatial azimuth The final tilt angle, final azimuth angle, and single-board 3D panel construction module are then used.

[0054] Slab reference elevation As shown in equation (1): (1) In equation (1), , and Represents the initial plate surface normal vector The three components, Represents the coordinates of the two-dimensional center point. This represents the coordinates of the centroid of the fitted plane.

[0055] The final tilt angle, final azimuth angle, and single-board 3D panel construction module include: The attitude smoothing unit is used to receive array grouping results and initial tilt angle. Initial spatial azimuth angle And mark the single board with abnormal attitude, and group the initial tilt angles by array. and initial spatial azimuth Outlier removal and statistical smoothing are performed, array-dominant attitude parameters are backfilled for boards with abnormal attitudes, and the final tilt angle is output. and final azimuth To construct 3D panel units.

[0056] 3D panel building blocks are used to receive the final tilt angle. Final azimuth , Rectangle long side direction angle 2D projection length 2D projection width Two-dimensional center point coordinates Slab surface reference elevation And array grouping results, and based on the final azimuth angle Determine the projection of the upslope direction onto the horizontal plane, then compare the rectangular direction with the upslope direction to determine the tilt direction of the slab. Based on the determination result, reconstruct the actual physical dimensions along the tilt direction of the slab and construct the center point of the slab. 3D coordinates of the four corner points, bottom elevation and top elevation The final azimuth, final tilt, reference elevation of the board surface, bottom elevation, top elevation, and three-dimensional coordinates of the four corner points are then backfilled into the attribute record, and the single-board geometric parameter dataset is output to the time series shadow distribution generation module.

[0057] The solar position parameter calculation module includes: The solar position parameter calculation unit receives the preset simulation period, time step, latitude and longitude of the target area's center point, date, and time, and calculates the solar altitude angle at each time point within the preset simulation period. and solar azimuth Then adjust the solar azimuth angle. The orientation reference is unified to a unified plane coordinate system, and the time series solar position parameters are output to the time series shadow distribution generation module and the cumulative photosynthetically active radiation reduction calculation and evaluation result output module.

[0058] The time series shadow distribution generation module includes: The solar orientation construction and candidate occlusion range generation unit receives time-series solar position parameters, single-board geometric parameter datasets, and a digital terrain model, and generates data based on the solar altitude angle at each time point during a preset simulation period. and solar azimuth Construct the solar direction vector Then, the boundary points of the three-dimensional rectangular surface of each individual photovoltaic panel are projected onto the corresponding surface reference surface of the digital terrain model along the opposite direction of solar incidence, forming the candidate shadow envelope range of each individual photovoltaic panel at each time. Based on the candidate shadow envelope range, a spatial index structure is established, and the solar direction vector is output. And spatial index structure to shadow state determination unit.

[0059] The shadow state determination unit is used to receive the solar direction vector. The system uses a spatial index structure, a single-panel geometric parameter dataset, and surface pixels from a digital terrain model. Based on the planar position of these surface pixels, it queries the spatial index structure for a set of candidate shading panels. If the candidate shading panel set is empty, it outputs a non-shaded state. If the candidate shading panel set is not empty, it constructs a line of sight for the surface pixels along the opposite direction of solar incidence. It only performs line-of-sight intersection checks on individual photovoltaic panels within the candidate shading panel set, outputting a raster of shadowed states at each time step. The module outputs the results of time-series shadow distribution calculation and evaluation of cumulative photosynthetically active radiation reduction.

[0060] The output module for calculating and evaluating the cumulative photosynthetically active radiation reduction includes: The radiation calculation unit is used to receive time-series solar position parameters and shadow state raster. The normal direct irradiance was calculated based on the solar position parameters at each time. Then based on the normal direct radiation irradiance and solar altitude angle Calculate the instantaneous photosynthetic effective radiation of a surface pixel under unobstructed conditions. and combined with the shadow state grid Calculate the actual photosynthetically active radiation of surface pixels under shaded conditions. and instantaneous photosynthetic effective radiation reduction Output instantaneous photosynthetic effective radiation reduction To the cumulative evaluation unit.

[0061] Instantaneous photosynthetic effective radiation of surface pixels under unobstructed conditions As shown in equation (2): (2) Actual photosynthetically active radiation of a surface pixel under shaded conditions As shown in equation (3): (3) In equations (2) and (3), This represents the radiation conversion factor. This represents the proportionality coefficient of scattered radiation.

[0062] The cumulative assessment unit is used to receive instantaneous photosynthetic effective radiation reduction. and time step And the instantaneous effective reduction of photosynthetic radiation at each moment during the simulation period. Perform summation or integration to obtain the cumulative photosynthetically active radiation reduction. Then, the spatial distribution map of the cumulative photosynthetically active radiation reduction of surface pixels in the target area is output as the result of the photovoltaic array shadow ecological assessment.

[0063] Cumulative photosynthetically active radiation reduction As shown in equation (4): (4) In equation (4), This represents the total number of time steps within the simulation period.

Claims

1. A photovoltaic array shading ecological assessment method based on single-plate geometric inversion, characterized in that, Includes the following steps: S1. Acquire high-resolution remote sensing images, 3D surface data, and digital terrain models of the target area, and unify the high-resolution remote sensing images, 3D surface data, and digital terrain models into the same plane coordinate system; based on the high-resolution remote sensing images, extract the vector contour, 2D center point coordinates, 2D projection length, 2D projection width, and long side orientation angle of each photovoltaic panel in the photovoltaic power station area; S2. Based on the vector contour of each individual photovoltaic panel, perform polygon interior point retrieval on the 3D surface data to obtain the original candidate panel point set; extract the ground elevation corresponding to the position of each point in the original candidate panel point set based on the digital terrain model and calculate the ground height, and remove ground points with ground height less than the preset lower threshold and abnormal high points with ground height greater than the preset upper threshold; perform plane fitting purification processing on the filtered point set, remove the external points that do not meet the same plane model and retain the interior point set as the panel fitting point set; invert the panel reference elevation, initial tilt angle and initial spatial azimuth angle based on the panel fitting point set, and combine the array grouping of each individual photovoltaic panel to obtain the final tilt angle and final azimuth angle; The projection of the uphill direction on the horizontal plane is determined based on the final azimuth angle. The uphill direction on the panel is compared with the direction of the rectangle determined by the direction angle of the long side of the rectangle. The dimensions corresponding to the tilt direction of the panel are determined in the two-dimensional projection length and two-dimensional projection width. Based on the final tilt angle, the dimensions corresponding to the tilt direction of the panel are restored to the actual physical dimensions along the tilt direction of the panel. The three-dimensional coordinates of the four corner points of each individual photovoltaic panel are generated to form a single panel geometric parameter dataset. S3. Based on the solar position parameters, digital terrain model and single-board geometric parameter dataset at each moment within the preset simulation period, perform spatial occlusion determination based on solar direction on the surface pixels of the target area to obtain the time series shadow distribution results. S4. Based on the time-series shadow distribution results and the solar position parameters at each time point, calculate the instantaneous photosynthetically active radiation of the surface pixels under unshaded conditions, and calculate the actual photosynthetically active radiation under shaded conditions by combining the shadow status indicators and the scattered radiation ratio coefficient at each time point. The instantaneous photosynthetically active radiation reduction is obtained by the difference between the instantaneous photosynthetically active radiation under unshaded conditions and the actual photosynthetically active radiation under shaded conditions. The instantaneous photosynthetically active radiation reduction is accumulated or integrated over the simulation period to obtain the spatial distribution map of the cumulative photosynthetically active radiation reduction as the ecological assessment result.

2. The photovoltaic array shadow ecology assessment method based on single-plate geometric inversion according to claim 1, characterized in that, In step S1, the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and rectangular long side orientation angle of each individual photovoltaic panel within the photovoltaic power station area are extracted, including: A multi-feature set is constructed based on high-resolution remote sensing images, and pixel-level classification is performed on the multi-feature set to obtain preliminary photovoltaic identification results. The preliminary photovoltaic identification results are processed by region merging and shape filtering to obtain the polygon of the photovoltaic power station area. Within the polygonal area of ​​the photovoltaic power station region, instance segmentation is performed on the image to obtain a single photovoltaic panel instance mask; Connectivity separation, polygon fitting, and minimum circumscribed rectangle fitting are performed on the mask of each individual photovoltaic panel instance to obtain the vector profile, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and long side orientation angle of the rectangle for each individual photovoltaic panel.

3. The photovoltaic array shadow ecology assessment method based on single-plate geometric inversion according to claim 1, characterized in that, In step S2, a plane fitting and purification process is performed to obtain a set of fitting points on the board surface, including: Using the vector contour of each individual photovoltaic panel as the two-dimensional query boundary, polygon interior point retrieval is performed on the three-dimensional surface data to extract three-dimensional points whose planar coordinates fall inside the vector contour, forming the original candidate panel surface point set; Extract the surface elevation corresponding to each point in the original candidate panel point set from the digital terrain model, calculate the height above the ground, and remove ground points with a height above the ground that is less than a preset lower threshold and abnormal high points with a height above the ground that is greater than a preset upper threshold. The point set after surface constraint filtering is further fitted using a random sampling consensus algorithm. External points that do not satisfy the same plane model are removed, and the internal point set is retained as the plate fitting point set.

4. The photovoltaic array shadow ecology assessment method based on single-plate geometric inversion according to claim 1, characterized in that, In step S2, the reference elevation, initial tilt angle, and initial spatial azimuth of the photovoltaic panel are inverted, and the final tilt angle and final azimuth angle are obtained by combining the array grouping of each individual photovoltaic panel, including: Perform the best-fit plane calculation on the set of fitted points on the plate surface to obtain the plate surface normal vector and the centroid of the fitted plane. Substitute the coordinates of the two-dimensional center point into the fitted plane equation to solve for the reference elevation of the plate surface. The initial tilt angle is determined by the angle between the plate surface normal vector and the vertical direction, and the initial spatial azimuth angle is determined by the projection direction of the plate surface normal vector onto the horizontal plane. Based on the spatial proximity of each individual photovoltaic panel, the similarity of its two-dimensional projection size, and the similarity of the orientation angle of its long side, array grouping is performed on each individual photovoltaic panel; Outlier removal and statistical smoothing are performed on the initial tilt angle and initial spatial azimuth angle according to the array group to obtain the final tilt angle and final azimuth angle.

5. The photovoltaic array shadow ecology assessment method based on single-plate geometric inversion according to claim 4, characterized in that, In step S2, the three-dimensional coordinates of the four corner points of each individual photovoltaic panel are generated, including: The projection of the upslope direction of the plate onto the horizontal plane is determined based on the final azimuth angle. The upslope direction of the plate is then compared with the direction of the long side of the rectangle determined by the direction angle of the long side of the rectangle and its orthogonal direction. The dimensions corresponding to the tilt direction of the plate and the dimensions corresponding to the transverse direction of the plate are then determined in the two-dimensional projection length and two-dimensional projection width. The dimensions corresponding to the tilt direction of the plate are restored to the actual physical dimensions along the tilt direction of the plate, and the center point of the plate is constructed based on the plate reference elevation and the coordinates of the two-dimensional center point. Based on the center point of the panel, the actual physical dimensions along the tilt direction of the panel, and the transverse dimensions of the panel, construct the three-dimensional coordinates of the four corner points of each individual photovoltaic panel; The average elevation of the two corner points on the lower side is determined as the bottom elevation, and the average elevation of the two corner points on the upper side is determined as the top elevation.

6. The photovoltaic array shadow ecology assessment method based on single-plate geometric inversion according to claim 1, characterized in that, In step S3, spatial occlusion determination based on solar direction is performed on the surface pixels of the target area, including: Construct the solar direction vector based on the solar altitude angle and solar azimuth angle at each moment; The boundary points of the three-dimensional rectangular surface of each individual photovoltaic panel are projected onto the surface reference surface corresponding to the digital terrain model in the opposite direction of solar incidence, forming the candidate shadow envelope range of each individual photovoltaic panel at the corresponding time, and a spatial index structure is established based on the candidate shadow envelope range of each individual photovoltaic panel. Based on the planar position of the surface pixels, the candidate shading plate set is queried in the spatial index structure. The intersection of the line of sight and the three-dimensional rectangular plate surface is determined only for the individual photovoltaic panels in the candidate shading plate set. The shadow state raster at each time moment is generated and the time series shadow distribution result is formed.

7. A photovoltaic array shading ecological assessment system based on single-plate geometric inversion, characterized in that, include: The basic data acquisition and coordinate unification module is used to acquire high-resolution remote sensing images, 3D surface data and digital terrain models of the target area, and unify the high-resolution remote sensing images, 3D surface data and digital terrain models into the same plane coordinate system; The module for identifying individual photovoltaic panels and extracting two-dimensional attributes is used to extract the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and rectangular long side orientation angle of each individual photovoltaic panel in the photovoltaic power station area based on high-resolution remote sensing images. The candidate panel point set extraction and surface constraint filtering module is used to extract candidate panel point sets based on the vector contour, three-dimensional surface data and digital terrain model of each individual photovoltaic panel. It calculates the ground height based on the digital terrain model and removes ground points and abnormal high points. It performs plane fitting purification processing on the filtered point set, and retains the inner point set as the panel fitting point set after removing outliers. The single-panel geometric parameter inversion and 3D panel construction module is used to invert the panel reference elevation, initial tilt angle and initial spatial azimuth angle based on the panel fitting point set. Combined with the array grouping of each individual photovoltaic panel, the final tilt angle and final azimuth angle are obtained. Based on the final azimuth angle, the projection of the uphill direction on the horizontal plane of the panel is determined. The uphill direction on the panel is compared with the rectangular direction to determine the dimension corresponding to the tilt direction of the panel in the 2D projection dimension. Based on the final tilt angle, the actual physical dimension along the tilt direction of the panel is restored, and the 3D coordinates of the four corner points of each individual photovoltaic panel are generated to form a single-panel geometric parameter dataset. The solar position parameter calculation module is used to calculate the solar position parameters at each time point within a preset simulation period. The time-series shadow distribution generation module is used to perform spatial occlusion determination based on the sun direction for surface pixels in the target area based on solar position parameters, digital terrain model and single-board geometric parameter dataset, and obtain time-series shadow distribution results. The module for calculating and evaluating the cumulative photosynthetically active radiation reduction (CARERR) outputs a spatial distribution map of CARERR as an ecological assessment result, based on time-series shadow distribution results and solar position parameters.

8. The photovoltaic array shadow ecological assessment system based on single-plate geometric inversion according to claim 7, characterized in that, The module for identifying individual photovoltaic panels and extracting two-dimensional attributes includes a photovoltaic power station area extraction unit, an instance segmentation unit, and a two-dimensional attribute extraction unit. The photovoltaic power station area extraction unit constructs a multi-feature set based on high-resolution remote sensing images, performs pixel-level classification, region merging, and shape filtering on the multi-feature set, and outputs a polygon of the photovoltaic power station area. The instance segmentation unit performs instance segmentation on the image within the polygonal area of ​​the photovoltaic power station region and outputs a single photovoltaic panel instance mask. The two-dimensional attribute extraction unit performs connected component separation, polygon fitting, and minimum bounding rectangle fitting on the mask of a single photovoltaic panel instance, and outputs the vector contour, two-dimensional center point coordinates, two-dimensional projection length, two-dimensional projection width, and long side orientation angle of the rectangle for each single photovoltaic panel.

9. The photovoltaic array shadow ecological assessment system based on single-plate geometric inversion according to claim 7, characterized in that, The single-board geometric parameter inversion and 3D board surface construction module includes array grouping units, attitude inversion units, attitude smoothing units, and 3D board surface construction units; The array grouping unit performs array grouping on each individual photovoltaic panel based on the spatial proximity relationship, similarity of two-dimensional projection size, and similarity of the orientation angle of the long side of the rectangle, and outputs the array grouping result; The attitude inversion unit performs best-fit plane calculation based on the plate surface fitting point set to obtain the plate surface normal vector and the centroid of the fitting plane, and calculates the plate surface reference elevation, initial tilt angle and initial spatial azimuth angle based on the fitting plane equation; The attitude smoothing unit performs outlier removal and statistical smoothing on the initial tilt angle and initial spatial azimuth angle by array grouping, and outputs the final tilt angle and final azimuth angle; The three-dimensional panel construction unit determines the tilt direction of the panel based on the final tilt angle, final azimuth angle, direction angle of the long side of the rectangle, coordinates of the two-dimensional center point, and two-dimensional projection dimensions, restores the actual physical dimensions along the tilt direction of the panel, and generates the three-dimensional coordinates of the four corner points of each individual photovoltaic panel.

10. The photovoltaic array shadow ecological assessment system based on single-plate geometric inversion according to claim 7, characterized in that, The time-series shadow distribution generation module includes a solar direction construction and candidate occlusion range generation unit and a shadow state determination unit; the cumulative photosynthetically effective radiation reduction calculation and evaluation result output module includes a radiation calculation unit and a cumulative evaluation unit. The solar direction construction and candidate shading range generation unit constructs a solar direction vector based on the solar elevation angle and solar azimuth angle at each moment, and projects the boundary points of the three-dimensional rectangular panel of each individual photovoltaic panel onto the surface reference surface corresponding to the digital terrain model along the opposite direction of solar incidence, forming the surface candidate shadow envelope range and establishing a spatial index structure. The shadow state determination unit queries the candidate shading plate set in the spatial index structure based on the planar position of the surface pixels, and performs the intersection determination of the line of sight and the three-dimensional rectangular plate surface only for the individual photovoltaic panels in the candidate shading plate set, and outputs the time series shadow distribution results; The radiation calculation unit calculates the instantaneous photosynthetically active radiation under unshaded conditions based on the time series shadow distribution results and solar position parameters, and calculates the actual photosynthetically active radiation under shaded conditions by combining the shadow status indicator and the scattered radiation ratio coefficient, and then calculates the instantaneous photosynthetically active radiation reduction. The cumulative evaluation unit sums or integrates the instantaneous reduction in photosynthetically active radiation at each moment during the simulation period and outputs a spatial distribution map of the cumulative reduction in photosynthetically active radiation.

Images