An automatic exploration method and system based on unmanned aerial vehicle rooftop aerial photography

Abstract

The present application relates to the technical field of drone exploration, and in particular to an automatic exploration method and system based on drone aerial photography of roofs. The present application first uses the three-dimensional visual sensor carried by the drone to obtain high-precision three-dimensional point cloud data and multi-angle two-dimensional images; then builds an accurate three-dimensional model of the roof based on the three-dimensional point cloud data and multi-angle two-dimensional images; then identifies the roof plane area and obstructions through model analysis; then calculates the optimal layout plan in combination with the photovoltaic panel parameters; finally generates a three-dimensional rendering image for the user to confirm and adjust; by considering the plane area and obstruction information, combined with the input photovoltaic panel parameters, a more accurate layout plan is calculated, which improves the accuracy of exploration, has strong environmental adaptability, and has more comprehensive layout optimization.

Description

Technical Field

[0001] The present application relates to the technical field of drone exploration, and in particular to an automatic exploration method and system based on drone aerial photography of roofs. Background Art

[0002] With the rapid development of renewable energy, rooftop photovoltaic power generation systems are becoming increasingly popular. However, the installation efficiency and power generation efficiency of rooftop photovoltaic systems are highly dependent on the structural characteristics and lighting conditions of the roof. Therefore, accurate roof exploration before installation becomes a key step to improve the performance of photovoltaic systems and is of great significance to promoting the widespread application of clean energy.

[0003] At present, rooftop exploration mainly relies on drone aerial photography technology. Drones are equipped with high-definition cameras and lidar to collect images and three-dimensional data of the roof. Computer vision algorithms process these data to build a three-dimensional model of the roof. Based on the three-dimensional model of the roof, the system can automatically plan the system to generate a preliminary photovoltaic panel layout plan to provide a reference for installation.

[0004] However, there are still some problems with the existing roof exploration technology. The accuracy of 3D model construction and roof feature recognition is limited, and the environmental adaptability is poor, which easily leads to inaccurate subsequent layout plans. At the same time, the automatically generated layout plan lacks comprehensive consideration of factors such as shadow effects and seasonal changes. The layout optimization is not comprehensive, making it difficult to achieve the optimal configuration of the photovoltaic system. This situation needs to be further improved. Summary of the invention

[0005] In order to solve the problems of poor environmental adaptability and incomplete layout optimization of existing roof exploration technologies, this application provides an automatic exploration method and system based on drone aerial photography of roofs, which adopts the following technical solutions:

[0006] In a first aspect, the present application provides an automatic exploration method based on drone aerial photography of roofs, comprising the following steps:

[0007] The 3D point cloud data and multi-angle 2D images of the target roof are obtained through the 3D visual sensor carried by the drone;

[0008] Constructing a three-dimensional roof model of the target roof according to the three-dimensional point cloud data and the multi-angle two-dimensional images;

[0009] According to the three-dimensional roof model, identifying and classifying the plane area and obstruction information on the target roof;

[0010] Calculate the photovoltaic panel layout plan according to the plane area and the obstruction information and the input photovoltaic panel parameters;

[0011] According to the photovoltaic panel layout plan, a three-dimensional rendering image is generated and displayed on the user terminal for user confirmation or further adjustment.

[0012] By adopting the above technical solution, the present application first uses the three-dimensional vision sensor carried by the drone to obtain high-precision three-dimensional point cloud data and multi-angle two-dimensional images; then builds an accurate three-dimensional model of the roof based on the three-dimensional point cloud data and multi-angle two-dimensional images; then identifies the roof plane area and obstructions through model analysis; then calculates the optimal layout plan in combination with the photovoltaic panel parameters; finally, generates a three-dimensional rendering image for the user to confirm and adjust; by considering the plane area and obstruction information, combined with the input photovoltaic panel parameters, a more accurate layout plan is calculated, which improves the exploration accuracy, has strong environmental adaptability and more comprehensive layout optimization.

[0013] Optionally, identifying and classifying the plane area and obstruction information on the target roof according to the three-dimensional roof model specifically includes the following steps:

[0014] According to the three-dimensional model of the roof, extract the main plane area to obtain the basic structure of the target roof;

[0015] According to the basic structure, cluster analysis is performed on the remaining non-planar points to obtain potential occluder areas;

[0016] According to the clustered geometric features, the obstruction area is classified by a preset decision tree algorithm to determine the type and position of the obstruction corresponding to the obstruction area.

[0017] By adopting the above technical solution, the present application first extracts the main plane area from the three-dimensional model of the roof to determine the basic structure of the roof; then performs cluster analysis on the remaining non-planar points to identify potential obstruction areas; finally, uses the preset decision tree algorithm to accurately classify and locate obstructions based on the geometric characteristics of the clusters; it can not only accurately identify complex roof structures, but also accurately classify and locate various obstructions, greatly improving the accuracy and comprehensiveness of roof exploration.

[0018] Optionally, according to the plane area and the obstruction information, combined with the input photovoltaic panel parameters, a photovoltaic panel layout solution is calculated, specifically including the following steps:

[0019] Determine a preliminary effective area where photovoltaic panels can be installed based on the plane area information, and subdivide the preliminary effective area based on input photovoltaic panel size parameters to obtain a final effective installation area;

[0020] Calculate the shadow coverage range and shadow coverage duration of each effective installation area according to the obstruction information;

[0021] According to the input photovoltaic panel parameters, determine the power generation model of a single photovoltaic panel;

[0022] A photovoltaic panel layout plan that maximizes annual power generation is calculated based on the effective installation areas, the shadow coverage range of each effective installation area, the shadow coverage duration, and the power generation model.

[0023] By adopting the above technical scheme, the existing photovoltaic system layout scheme has not fully considered factors such as the complexity of the roof structure, the influence of obstructions and the parameters of photovoltaic panels, resulting in low layout efficiency and unsatisfactory power generation; the present application first determines the final effective installation area based on the plane area information and the size of the photovoltaic panel; then calculates the shadow influence of each area based on the obstruction information; then establishes a power generation model for a single photovoltaic panel; finally, comprehensively considering the above factors, calculates the optimal layout scheme; not only takes into account the actual available space on the roof and the influence of obstructions, but also combines the specific parameters of the photovoltaic panels, which can effectively improve the space utilization and the overall power generation efficiency of the system.

[0024] Optionally, according to the geometric features of the clusters, the occluder region is classified by a preset decision tree algorithm to determine the occluder type and position corresponding to the occluder region, which specifically includes the following steps:

[0025] According to the geometric features of the cluster and a preset height threshold, the occluder area is preliminarily divided into high-level occluders and low-level occluders;

[0026] Classifying the high-rise obstructions according to their volumes and shape factors;

[0027] According to the area and shape factor of the low-level obstructions, the low-level obstructions are classified into detailed categories;

[0028] According to the refined classification result, the type of each shielding object is determined, and the position coordinates of each shielding object in the three-dimensional model of the roof are recorded.

[0029] By adopting the above technical scheme, since different types of obstructions have different effects on the photovoltaic system, this application first preliminarily divides the obstructions into high-level and low-level according to the height threshold; then refines the high-level and low-level obstructions according to volume, area and shape factors; finally, determines the specific type of each obstruction and records its position coordinates; it can not only accurately distinguish different types of obstructions, but also accurately locate each obstruction, greatly improving the accuracy and detail of obstruction identification.

[0030] Optionally, according to the obstruction information, calculating the shadow coverage range and shadow coverage duration of each effective installation area specifically includes the following steps:

[0031] Based on the sun trajectory model, calculate the sun's azimuth and altitude at different times throughout the year;

[0032] Based on the three-dimensional position and shape of each obstruction and the sun's trajectory, the shadow distribution at each time point throughout the year is simulated;

[0033] The shadow distribution is mapped to each effective installation area to obtain the shadow coverage range of each effective installation area; and the cumulative duration of each part of each effective installation area being covered by the shadow throughout the year is counted to obtain the shadow coverage duration.

[0034] By adopting the above technical solution, the present application first calculates the changes in the sun's position throughout the year based on the sun trajectory model; then simulates the shadow distribution throughout the year in combination with the three-dimensional information of the obstruction; then maps the shadow distribution to the effective installation area; finally, counts the shadow coverage duration of each area; not only takes into account the changes in the sun's position throughout the year, but also combines the precise three-dimensional information of the obstruction, and can provide detailed shadow coverage data for each installation area throughout the year, thereby improving the annual power generation and overall system efficiency of the photovoltaic system.

[0035] Optionally, according to the effective installation area, the shadow coverage range of each effective installation area, the shadow coverage duration and the power generation model, a photovoltaic panel layout plan for maximizing annual power generation is calculated, which specifically includes the following steps:

[0036] Call the preset objective function of maximizing annual power generation;

[0037] Generate constraint conditions according to the boundary restrictions of each effective installation area and the predetermined minimum spacing requirements between photovoltaic panels; use an improved particle swarm optimization algorithm to generate multiple initial candidate layout solutions that meet the constraint conditions;

[0038] According to the shadow coverage range, shadow coverage duration and power generation model, the objective function value of each candidate layout scheme is calculated; according to the objective function value and the constraint conditions, the position and orientation of the photovoltaic panel are iteratively optimized to obtain the photovoltaic panel layout scheme.

[0039] By adopting the above technical solution, this application first establishes an objective function for maximizing annual power generation; then sets constraints based on the installation area and spacing requirements; then uses an improved particle swarm algorithm to generate initial candidate solutions; then evaluates each solution based on shadow data and power generation model; finally, the best layout solution is obtained through iterative optimization; not only the complex roof environment and multiple constraints are taken into account, but also precise shadow analysis data is used to find the optimal solution in a global scope, which effectively improves the space utilization and overall power generation efficiency of the photovoltaic system, provides users with the most optimized photovoltaic panel layout solution, and improves the economic benefits of the system.

[0040] Optionally, the objective function is:

[0041] F(x)=∑(P i *E i *(1-S i )*H i ), where x represents the location and orientation of the photovoltaic panel, P i is the rated power of the ith photovoltaic panel, E i is the effective sunshine coefficient, S i is the shadow loss factor, S i Determined according to the shadow coverage range and the shadow coverage duration, H i It is the annual effective working hours of photovoltaic panels.

[0042] By adopting the above technical solution, the present application establishes a mathematical model based on key factors such as the location and orientation of photovoltaic panels, rated power, effective sunshine, shadow impact, and annual effective working time, so as to achieve accurate calculation and optimization of annual power generation.

[0043] In a second aspect, the present application provides an automatic exploration system based on drone aerial photography of roofs, comprising:

[0044] A data acquisition module is used to acquire 3D point cloud data and multi-angle 2D images of the target roof through a 3D visual sensor carried by the UAV;

[0045] A model building module, used to build a three-dimensional roof model of the target roof according to the three-dimensional point cloud data and the multi-angle two-dimensional image;

[0046] An identification and classification module, used to identify and classify the plane area and obstruction information on the target roof according to the three-dimensional roof model;

[0047] A layout generation module, used to calculate a photovoltaic panel layout plan based on the plane area and the obstruction information and the input photovoltaic panel parameters;

[0048] The terminal display module is used to generate a three-dimensional rendering image according to the photovoltaic panel layout plan and display it on the user terminal for the user to confirm or make further adjustments.

[0049] In a third aspect, the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned automatic exploration method based on drone aerial photography of roofs when executing the computer program.

[0050] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-mentioned automatic exploration method based on drone aerial photography of roofs.

[0051] In summary, the present application includes at least one of the following beneficial technical effects:

[0052] 1. This application first uses the 3D visual sensor carried by the drone to obtain high-precision 3D point cloud data and multi-angle 2D images; then builds an accurate 3D model of the roof based on the 3D point cloud data and multi-angle 2D images; then identifies the roof plane area and obstructions through model analysis; then calculates the optimal layout plan in combination with the photovoltaic panel parameters; and finally generates a 3D rendering image for the user to confirm and adjust; by considering the plane area and obstruction information, combined with the input photovoltaic panel parameters, a more accurate layout plan is calculated, the exploration accuracy is improved, the environmental adaptability is strong, and the layout optimization is more comprehensive;

[0053] 2. This application first extracts the main plane area from the 3D model to determine the basic structure of the roof; then performs cluster analysis on the remaining non-plane points to identify potential obstruction areas; finally, uses the preset decision tree algorithm to accurately classify and locate obstructions based on the geometric features of the clusters; it can not only accurately identify complex roof structures, but also accurately classify and locate various obstructions, greatly improving the accuracy and comprehensiveness of roof exploration;

[0054] 3. This application first preliminarily divides the obstructions into high-level and low-level objects according to the height threshold; then, the high-level and low-level obstructions are further classified according to the volume, area and shape factors; finally, the specific type of each obstruction is determined and its position coordinates are recorded; it can not only accurately distinguish different types of obstructions, but also accurately locate each obstruction, greatly improving the accuracy and detail of obstruction recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of an automatic exploration method for rooftop aerial photography by a drone according to an embodiment of the present application;

[0056] Figure 2 It is a flowchart of step S300 in an automatic exploration method for rooftop aerial photography by a drone in an embodiment of the present application;

[0057] Figure 3 It is a flowchart of step S330 in an automatic exploration method of a roof by drone aerial photography in an embodiment of the present application;

[0058] Figure 4 This is a flow chart of step S400 in a method for automatically surveying a roof by taking aerial photos of a drone in an embodiment of the present application;

[0059] Figure 5 It is a flowchart of step S420 in an automatic exploration method of a roof by drone aerial photography in an embodiment of the present application;

[0060] Figure 6 It is a flowchart of step S440 in a method for automatically surveying a roof by taking aerial photos of a drone in an embodiment of the present application;

[0061] Figure 7 This is a schematic diagram of a module of an automatic exploration system for drone aerial photography of roofs according to an embodiment of the present application;

[0062] Figure 8 It is a diagram of the internal structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0063] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more listed items.

[0064] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0065] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0066] In the first aspect, the present application provides an automatic exploration method based on drone aerial photography of roofs, referring to Figure 1 , including the following steps:

[0067] S100, obtains the 3D point cloud data and multi-angle 2D images of the target roof through the 3D visual sensor carried by the drone.

[0068] The 3D vision sensor can be a laser radar or structured light sensor, etc., which is used to obtain accurate 3D spatial information of the target roof. The multi-angle 2D image is a photo of the roof taken from different angles by a high-resolution camera carried by a drone, which is used to provide texture and color information of the roof.

[0069] Specifically, the drone can take aerial photos above the target roof according to a preset flight path. The flight altitude is adjusted according to the roof area and the required accuracy, usually between 10-30 meters. During the flight, the 3D vision sensor continuously scans the roof surface, while the camera takes multi-angle images at certain intervals. To ensure the integrity and accuracy of the data, multiple repeated scanning and shooting methods are used, and data fusion and optimization are performed in subsequent processing.

[0070] S200: construct a three-dimensional roof model of the target roof based on the three-dimensional point cloud data and the multi-angle two-dimensional images.

[0071] Among them, the three-dimensional roof model is a digital representation of the geometric shape and surface features of the target roof, including information such as the overall outline, slope, orientation, and surface material of the roof.

[0072] Specifically, the acquired 3D point cloud data is first preprocessed, including denoising, downsampling and other operations. Then, the point cloud data obtained by multiple scans are aligned and fused using a point cloud registration algorithm such as the iterative closest point algorithm. Next, the point cloud is converted into a triangular mesh model using a meshing algorithm. Finally, the texture information in the multi-angle 2D image is mapped to the 3D model surface to obtain a 3D roof model with real texture.

[0073] S300: According to the three-dimensional roof model, identify and classify the plane area and obstruction information on the target roof.

[0074] The plane area refers to the relatively flat area on the roof where photovoltaic panels can be installed, and the obstruction information includes data such as the location, size, and shape of various protruding structures, equipment, or obstacles on the roof.

[0075] Specifically, the region growing algorithm is first used to perform plane detection on the three-dimensional model to identify the main plane areas. For non-planar areas, a clustering method based on curvature and normal vectors is used for preliminary segmentation. Then, geometric features such as area, height, shape factor, etc. are extracted from each segmented region. Finally, a pre-trained machine learning model, such as support vector machine (SVM) or random forest, is used to classify these areas and identify common roof obstructions such as chimneys, skylights, and air conditioning outdoor units. In this process, the visual information provided by multi-angle two-dimensional images is combined to improve the accuracy of classification.

[0076] S400: Calculate a photovoltaic panel layout plan based on the plane area and the obstruction information and the input photovoltaic panel parameters.

[0077] In this embodiment, the photovoltaic panel parameters include technical indicators such as size, efficiency, and weight of the photovoltaic panel, as well as installation requirements such as installation angle and minimum spacing.

[0078] Specifically, a heuristic algorithm is used to calculate the layout plan. First, a possible layout grid is generated based on the shape of the plane area and the size of the photovoltaic panel. Then, the position and height of the obstruction are considered to calculate the shadow distribution at each time of the year. Next, the position and orientation of the photovoltaic panels are optimized with the objective function of maximizing the annual power generation. For example, the crossover rate can be set to 0.8, the mutation rate to 0.1, the population size to 100, and the iterations to run the genetic algorithm 500 times. The final layout plan should include the precise position, orientation angle, and estimated annual power generation of each photovoltaic panel.

[0079] S500 , generating a three-dimensional rendering image according to the photovoltaic panel layout plan and displaying it on a user terminal for user confirmation or further adjustment.

[0080] Among them, the three-dimensional rendering image is the process of visualizing the calculated photovoltaic panel layout scheme, with the aim of allowing users to intuitively understand the layout effect and make interactive adjustments. The user terminal can be a computer, tablet or smart phone and other devices, which need to have sufficient graphics processing capabilities to achieve a smooth three-dimensional interactive experience. In the user interface, interactive functions such as zooming, rotating, and panning are provided to allow users to view the layout effect from different angles. In addition, a dynamic shadow simulation function can be added to allow users to observe the shadow changes at different times of the day. To facilitate user adjustments, the drag-and-drop photovoltaic panel position fine-tuning function can be implemented, and the relevant power generation data can be updated in real time. Finally, the save and export functions are provided to allow users to save or export the confirmed scheme into a construction drawing format.

[0081] In one embodiment, referring to Figure 2 In step S300, according to the three-dimensional roof model, the plane area and the obstruction information on the target roof are identified and classified, which specifically includes the following steps:

[0082] S310: Extract the main plane area according to the three-dimensional roof model to obtain the basic structure of the target roof.

[0083] The main plane area refers to the large continuous flat surface on the roof, which is usually a potential area for installing photovoltaic panels. The basic structure refers to the main shape of the roof, including the slope and platform of the roof.

[0084] Specifically, a random sample consensus algorithm is used to extract the main plane. First, a set of points are randomly selected from the 3D model and a plane equation is fitted. Then, the distance from all points to this plane is calculated, and the points with a distance less than the threshold are marked as inliers. This process is repeated many times, and the plane with the most inliers is selected as the main plane. For complex roof structures, multiple main planes may need to be extracted. For example, the number of RANSAC iterations can be set to 1000 and the distance threshold to 0.1 meters. After extracting the main planes, the region growing algorithm can be used to expand these planes to obtain a more complete plane area. Finally, these main planes are combined to form the basic structural model of the roof.

[0085] S320: Based on the basic structure, cluster analysis is performed on the remaining non-planar points to obtain potential occluder regions.

[0086] In this embodiment, non-planar points refer to point cloud data that does not belong to the main plane area, representing various protrusions or equipment on the roof. The purpose of cluster analysis is to group these non-planar points into potential obstruction areas to prepare for subsequent classification.

[0087] Specifically, a density-based clustering algorithm is used, and the neighborhood radius and the minimum number of points are set. For each point, if there are at least more than the minimum number of points in the neighborhood, it is marked as a core point, and all density-reachable core points are classified into one category. For points that do not belong to any class, they can be marked as noise points and temporarily ignored. For each clustering area, the center point coordinates, area occupied, and height features are calculated to prepare for the next classification.

[0088] S330: Classify the occluder region according to the clustered geometric features by using a preset decision tree algorithm to determine the occluder type and position corresponding to the occluder region.

[0089] The geometric features refer to the quantifiable attributes of each cluster area, such as shape, size, and height. The decision tree algorithm is a rule-based classification method that can gradually determine the type of obstruction based on these features. The types of obstructions may include common roof facilities such as chimneys, skylights, air conditioner outdoor units, and solar water heaters.

[0090] Specifically, a training data set containing various occluder features and corresponding types is pre-established, and the information gain ratio is used as the feature selection criterion to build a decision tree. When applied, the corresponding features are extracted for each clustering area, and then the branches of the decision tree are gradually judged to finally obtain the type of occluder. If a situation is difficult to classify, it is set to "unknown" and confirmed or corrected in subsequent manual review.

[0091] In one embodiment, referring to Figure 3In step S330, according to the geometric features of the cluster, the occluder region is classified by a preset decision tree algorithm to determine the type and position of the occluder corresponding to the occluder region, which specifically includes the following steps:

[0092] S331. Preliminarily divide the occluder area into high-level occluders and low-level occluders according to the clustered geometric features and a preset height threshold.

[0093] The height threshold is a reference value pre-set based on the statistical data of common rooftop obstructions, which is used to preliminarily distinguish between high- and low-level obstructions. High-level obstructions usually include vertically extending structures such as chimneys and antennas, while low-level obstructions may be relatively flat devices such as skylights and air-conditioning outdoor units. For example, the height threshold can be set to 1.5 meters. For each clustering area, if its height exceeds 1.5 meters, it is classified as a high-level obstruction; otherwise, it is classified as a low-level obstruction.

[0094] S332: Classify high-rise obstructions according to their volumes and shape factors.

[0095] In this embodiment, volume refers to the size of the three-dimensional space occupied by the high-rise shielding object, and shape factor is a numerical indicator describing the geometric shape of the shielding object, such as roundness, slenderness, etc. These features can help distinguish different types of high-rise shielding objects, such as distinguishing chimneys from antennas.

[0096] S333. Classify the low-level obstructions in detail according to their areas and shape factors.

[0097] Among them, area refers to the projected area of ​​the low-level obstruction on the horizontal plane, which is an important feature for distinguishing different types of low-level obstructions. Shape factor can help identify the geometric outline of the obstruction, such as rectangularity, aspect ratio, etc., which is of great significance for distinguishing skylights, air conditioner outdoor units, etc.

[0098] S334. Determine the type of each shielding object according to the refined classification result, and record the position coordinates of each shielding object in the three-dimensional roof model.

[0099] In this embodiment, the type of the obstruction refers to a specific category obtained after detailed classification, such as a chimney, an antenna, a skylight, an air conditioner outdoor unit, etc. The position coordinates refer to the precise position of each obstruction in the three-dimensional roof model coordinate system, and are represented by the center point coordinates.

[0100] In one embodiment, referring to Figure 4 In step S400, the photovoltaic panel layout scheme is calculated according to the plane area and the obstruction information and the input photovoltaic panel parameters, which specifically includes the following steps:

[0101] S410: Determine a preliminary effective area where photovoltaic panels can be installed based on the plane area information, and subdivide the preliminary effective area based on the input photovoltaic panel size parameters to obtain a final effective installation area.

[0102] The initial effective area refers to the flat area on the roof that meets the basic installation conditions, taking into account the structural characteristics of the roof, load-bearing capacity and other factors. The PV panel size parameters include the length, width and inclination angle of a single PV panel during installation. The final effective installation area is the actual installation area obtained after considering the specific size of the PV panel based on the initial effective area.

[0103] Specifically, a polygon clipping algorithm is used to determine the preliminary effective area. First, the roof plane area is represented as one or more polygons. Then, the safety distance of the roof edge and known non-installable areas such as around the skylight are subtracted from these polygons. Next, the preliminary effective area is subdivided using a grid partitioning algorithm based on the size parameters of the photovoltaic panel. For each grid cell, check whether it is completely within the preliminary effective area, and consider the spacing requirements with adjacent photovoltaic panels. The grid cells that meet the conditions constitute the final effective installation area. This method ensures that each divided area can actually install a photovoltaic panel while maximizing the use of available space.

[0104] S420: Calculate the shadow coverage range and shadow coverage duration of each effective installation area according to the blocking object information.

[0105] In this embodiment, the shadow coverage refers to the shape and position of the shadow cast by the shielding object on the effective installation area at different times of the day. The shadow coverage duration refers to the cumulative time that each effective installation area is covered by the shadow.

[0106] S430: Determine a power generation model of a single photovoltaic panel according to the input photovoltaic panel parameters.

[0107] The photovoltaic panel parameters include but are not limited to technical indicators such as rated power, conversion efficiency, and temperature coefficient. The power generation model is a mathematical expression used to describe the expected power generation of a single photovoltaic panel under given environmental conditions. It is generally calculated using a fixed model, and then calibrated and verified based on experimental data or performance curves provided by the manufacturer.

[0108] S440, calculating a photovoltaic panel layout plan that maximizes annual power generation based on the effective installation area, the shadow coverage range of each effective installation area, the shadow coverage duration, and the power generation model.

[0109] In this embodiment, the photovoltaic panel layout scheme that maximizes annual power generation refers to a photovoltaic panel arrangement method that can generate the highest annual total power generation under given roof conditions and constraints.

[0110] In one embodiment, referring to Figure 5 In step S420, the shadow coverage range and shadow coverage duration of each effective installation area are calculated according to the shielding information, which specifically includes the following steps:

[0111] S421. Based on the solar trajectory model, calculate the solar azimuth and altitude angle at different time points throughout the year.

[0112] In this embodiment, the sun trajectory model is a mathematical model that describes the apparent motion of the sun in the sky. The solar azimuth is the angle between the sun's projection and the true north direction, while the altitude is the angle between the sun and the horizontal plane. The azimuth and altitude change with time and geographic location, and are the basis for calculating shadow distribution.

[0113] S422. According to the three-dimensional position and shape of each shielding object and the sun's trajectory, simulate the shadow distribution at each time point throughout the year.

[0114] The three-dimensional position and shape of the occluder refers to the specific position and geometric features of the occluder in the roof coordinate system. Shadow distribution refers to the set of shadow areas cast by all occluders on the roof plane at a specific time point.

[0115] S423: Map the shadow distribution to each effective installation area to obtain the shadow coverage of each effective installation area.

[0116] In this embodiment, shadow distribution mapping refers to the process of spatially superimposing the calculated shadow contour with the predetermined effective installation area. The shadow coverage range represents the specific part of each effective installation area covered by the shadow at different time points.

[0117] Specifically, a rasterization method is used to achieve the mapping of shadow distribution. First, the entire roof plane is divided into high-resolution grids. Then, for the shadow distribution at each time point, the grid cells in the shadow area are marked as 1, and other areas are marked as 0 to form a binary image. For each effective installation area, its corresponding grid cell is extracted and a pixel-by-pixel "AND" operation is performed with the shadow binary image. In this way, the shadow coverage of each effective installation area at different time points is obtained.

[0118] S424. Count the cumulative duration of each effective installation area being covered by shadows throughout the year to obtain the shadow coverage duration.

[0119] In one embodiment, referring to Figure 6 In step S440, a photovoltaic panel layout scheme for maximizing annual power generation is calculated based on the effective installation area, the shadow coverage range of each effective installation area, the shadow coverage duration and the power generation model, which specifically includes the following steps:

[0120] S441, calling a preset objective function for maximizing annual power generation.

[0121] Specifically, the objective function is: F(x) = ∑(P i *E i *(1-S i )*H i ), where x represents the location and orientation of the photovoltaic panel, P i is the rated power of the ith photovoltaic panel, E i is the effective sunshine coefficient, S i is the shadow loss factor, S i Determined according to the shadow coverage range and shadow coverage duration, H i It is the annual effective working hours of photovoltaic panels.

[0122] S442: Generate constraint conditions according to the boundary restrictions of each effective installation area and the predetermined minimum spacing requirements between photovoltaic panels.

[0123] Among them, the boundary restriction means that the photovoltaic panels must be completely located within the effective installation area and must not exceed the edge of the roof or enter the non-installable area. The predetermined minimum spacing requirement is to ensure that there is enough space between photovoltaic panels for installation and maintenance, and to reduce mutual shading. These constraints together define the feasible solution space.

[0124] S443. Using an improved particle swarm optimization algorithm, generate multiple initial candidate layout solutions that meet the constraint conditions.

[0125] In this embodiment, the improved particle swarm optimization algorithm is an optimization method based on swarm intelligence, and the initial candidate layout scheme refers to a group of photovoltaic panel layouts that meet the constraint conditions, which serves as the starting point of the optimization algorithm.

[0126] S444. Calculate the objective function value of each candidate layout solution according to the shadow coverage range, shadow coverage duration and power generation model.

[0127] The objective function value is a quantitative evaluation of the total annual power generation of each candidate layout scheme. This step combines the shadow information obtained previously with the power generation model to assign a specific performance indicator to each scheme.

[0128] S445. According to the objective function value and the constraint conditions, the position and orientation of the photovoltaic panels are iteratively optimized to obtain a photovoltaic panel layout plan.

[0129] In this embodiment, iterative optimization refers to gradually improving the candidate layout solutions through multiple rounds of calculations and adjustments to obtain a higher objective function value.

[0130] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0131] In the second aspect, the present application provides an automatic exploration system based on drone aerial photography of roofs. The automatic exploration system based on drone aerial photography of roofs of the present application is described below in combination with the above-mentioned automatic exploration method based on drone aerial photography of roofs.

[0132] Reference Figure 7 , an automatic exploration system based on drone aerial photography of roofs, including:

[0133] A data acquisition module is used to acquire 3D point cloud data and multi-angle 2D images of the target roof through a 3D visual sensor carried by the UAV;

[0134] The model building module is used to build a three-dimensional roof model of the target roof based on the three-dimensional point cloud data and the multi-angle two-dimensional images; the recognition and classification module is used to identify and classify the plane area and obstruction information on the target roof based on the three-dimensional roof model;

[0135] The layout generation module is used to calculate the photovoltaic panel layout plan based on the plane area and obstruction information combined with the input photovoltaic panel parameters;

[0136] The terminal display module is used to generate a three-dimensional rendering image according to the photovoltaic panel layout plan and display it on the user terminal for the user to confirm or make further adjustments.

[0137] In one embodiment, the recognition and classification module includes:

[0138] The basic structure acquisition unit is used to extract the main plane area according to the three-dimensional roof model to obtain the basic structure of the target roof; the occlusion area acquisition unit is used to cluster the remaining non-planar points according to the basic structure to obtain the potential occlusion area;

[0139] The type and position determination unit is used to classify the occluder area according to the clustered geometric features through a preset decision tree algorithm to determine the occluder type and position corresponding to the occluder area.

[0140] In one embodiment, the type and position determination unit comprises:

[0141] A preliminary classification subunit, used to preliminarily classify the occluder area into high-level occluders and low-level occluders according to the geometric features of the clusters and a preset height threshold;

[0142] A high-level occluder classification subunit, used to refine the classification of high-level occluders according to their volume and shape factors;

[0143] A low-level occluder classification subunit, used to refine the classification of low-level occluders according to their area and shape factors;

[0144] The position coordinate recording subunit is used to determine the type of each shielding object according to the refined classification result, and record the position coordinates of each shielding object in the three-dimensional model of the roof.

[0145] In one embodiment, the layout generation module includes:

[0146] An effective installation area determination unit is used to determine a preliminary effective area where photovoltaic panels can be installed based on the plane area information, and to subdivide the preliminary effective area based on the input photovoltaic panel size parameters to obtain a final effective installation area;

[0147] A shadow coverage information determination unit, used to calculate the shadow coverage range and shadow coverage duration of each effective installation area according to the shielding information;

[0148] A power generation model determination unit, used to determine a power generation model of a single photovoltaic panel according to input photovoltaic panel parameters;

[0149] The layout scheme calculation unit is used to calculate the photovoltaic panel layout scheme that maximizes the annual power generation according to the effective installation area, the shadow coverage range of each effective installation area, the shadow coverage time and the power generation model.

[0150] In one embodiment, the shadow coverage information determining unit includes:

[0151] The sun angle calculation subunit is used to calculate the sun's azimuth and altitude at different time points throughout the year based on the sun trajectory model; the shadow distribution simulation subunit is used to simulate the shadow distribution at each time point throughout the year based on the three-dimensional position and shape of each shielding object combined with the sun trajectory;

[0152] A shadow distribution mapping subunit is used to map the shadow distribution to each effective installation area to obtain the shadow coverage of each effective installation area;

[0153] The coverage duration statistics subunit is used to count the cumulative duration of each part of each effective installation area being covered by shadows throughout the year to obtain the shadow coverage duration.

[0154] In one embodiment, the layout solution calculation unit includes:

[0155] A function calling subunit, used to call a preset objective function of maximizing annual power generation;

[0156] A constraint condition generating subunit, for generating constraint conditions according to boundary restrictions of each effective installation area and predetermined minimum spacing requirements between photovoltaic panels;

[0157] A candidate layout scheme generating subunit is used to generate multiple initial candidate layout schemes satisfying constraint conditions by using an improved particle swarm optimization algorithm;

[0158] An objective function value calculation subunit is used to calculate the objective function value of each candidate layout scheme according to the shadow coverage range, shadow coverage duration and power generation model;

[0159] The layout scheme generating subunit is used to iteratively optimize the position and orientation of the photovoltaic panels according to the objective function value and the constraint conditions to obtain the photovoltaic panel layout scheme.

[0160] In one embodiment, the objective function is: F(x)=∑(P i *E i *(1-S i )*H i ), where x represents the location and orientation of the photovoltaic panel, P i is the rated power of the ith photovoltaic panel, E i is the effective sunshine coefficient, S i is the shadow loss factor, S i Determined according to the shadow coverage range and shadow coverage duration, H i It is the annual effective working hours of photovoltaic panels.

[0161] In one embodiment, the present application provides an electronic device, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown. The electronic device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store data. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an automatic exploration method based on drone aerial photography of roofs is implemented.

[0162] Those skilled in the art will understand that Figure 8 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0163] In one embodiment, an electronic device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0164] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the above-mentioned computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0165] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. An automatic exploration method based on drone aerial photography of roofs, characterized in that: The steps include: The 3D point cloud data and multi-angle 2D images of the target roof are obtained through the 3D visual sensor carried by the drone; Constructing a three-dimensional roof model of the target roof according to the three-dimensional point cloud data and the multi-angle two-dimensional images; According to the three-dimensional roof model, identifying and classifying the plane area and obstruction information on the target roof; Calculate the photovoltaic panel layout plan according to the plane area and the obstruction information and the input photovoltaic panel parameters; According to the photovoltaic panel layout plan, a three-dimensional rendering image is generated and displayed on a user terminal for user confirmation or further adjustment; According to the three-dimensional roof model, identifying and classifying the plane area and obstruction information on the target roof specifically includes the following steps: According to the three-dimensional model of the roof, extract the main plane area to obtain the basic structure of the target roof; According to the basic structure, cluster analysis is performed on the remaining non-planar points to obtain potential occluder areas; According to the clustered geometric features, the occluder area is classified by a preset decision tree algorithm to determine the occluder type and position corresponding to the occluder area; According to the plane area and the obstruction information, combined with the input photovoltaic panel parameters, the photovoltaic panel layout plan is calculated, which specifically includes the following steps: Determine a preliminary effective area where photovoltaic panels can be installed based on the plane area information, and subdivide the preliminary effective area based on the input photovoltaic panel size parameters to obtain a final effective installation area; Calculate the shadow coverage range and shadow coverage duration of each effective installation area according to the obstruction information; According to the input photovoltaic panel parameters, determine the power generation model of a single photovoltaic panel; Call the preset objective function of maximizing annual power generation; generating constraint conditions according to the boundary limits of each effective installation area and the predetermined minimum spacing requirements between photovoltaic panels; Using an improved particle swarm optimization algorithm, a plurality of initial candidate layout solutions satisfying the constraint conditions are generated; Calculating the objective function value of each candidate layout scheme according to the shadow coverage range, shadow coverage duration and power generation model; Iteratively optimize the position and orientation of the photovoltaic panels according to the objective function value and the constraint conditions to obtain a photovoltaic panel layout plan; Wherein, the objective function is: F(x)=∑(P i *E i *(1-S i )*H i ), where x represents the location and orientation of the photovoltaic panel, P i is the rated power of the ith photovoltaic panel, E i is the effective sunshine coefficient, S i is the shadow loss factor, S i Determined according to the shadow coverage range and the shadow coverage duration, H i It is the annual effective working hours of photovoltaic panels.

2. The automatic exploration method based on drone aerial photography of roofs according to claim 1 is characterized in that: According to the geometric features of the clusters, the occluder region is classified by a preset decision tree algorithm to determine the occluder type and position corresponding to the occluder region, which specifically includes the following steps: According to the geometric features of the cluster and a preset height threshold, the occluder area is preliminarily divided into high-level occluders and low-level occluders; Classifying the high-rise obstructions according to their volumes and shape factors; According to the area and shape factor of the low-level obstructions, the low-level obstructions are classified into detailed categories; According to the refined classification result, the type of each shielding object is determined, and the position coordinates of each shielding object in the three-dimensional model of the roof are recorded.

3. The automatic exploration method based on drone aerial photography of roofs according to claim 1 is characterized in that: According to the obstruction information, the shadow coverage range and shadow coverage duration of each effective installation area are calculated, which specifically includes the following steps: Based on the sun trajectory model, calculate the sun's azimuth and altitude at different times throughout the year; Based on the three-dimensional position and shape of each shielding object and the sun's trajectory, the shadow distribution at each time point throughout the year is simulated; Mapping the shadow distribution to each effective installation area to obtain the shadow coverage of each effective installation area; The cumulative duration of each effective installation area being covered by shadows throughout the year is counted to obtain the shadow coverage duration.

4. An automatic exploration system based on drone aerial photography of roofs, characterized in that: The automatic exploration method based on drone aerial photography of roofs according to any one of claims 1 to 3 comprises: A data acquisition module is used to acquire 3D point cloud data and multi-angle 2D images of the target roof through a 3D visual sensor carried by the UAV; A model building module, used to build a three-dimensional roof model of the target roof according to the three-dimensional point cloud data and the multi-angle two-dimensional image; An identification and classification module, used to identify and classify the plane area and obstruction information on the target roof according to the three-dimensional roof model; A layout generation module, used to calculate a photovoltaic panel layout plan based on the plane area and the obstruction information and the input photovoltaic panel parameters; The terminal display module is used to generate a three-dimensional rendering image according to the photovoltaic panel layout plan and display it on the user terminal for the user to confirm or make further adjustments.

5. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the automatic exploration method based on aerial photography of roofs by unmanned aerial vehicles according to any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the automatic exploration method based on drone aerial photography of roofs described in any one of claims 1 to 3 are implemented.

Images