Intelligent unmanned aerial vehicle inspection cleaning method and system for photovoltaic power station

Abstract

The invention relates to the technical field of photovoltaic power stations, and particularly discloses an intelligent unmanned aerial vehicle inspection cleaning method and system for a photovoltaic power station. According to the method, after an offset reference value is obtained, the color offset degree of a to-be-detected area is analyzed based on the offset reference value to determine a polluted area, so that uneven illumination caused by inclination angle and illumination change is compensated to eliminate ambient illumination interference; grouping the to-be-cleaned areas of which the pollution risk levels are determined, and sorting the to-be-cleaned areas in combination with weight parameters to generate a cleaning sequence; and integrating the position of the to-be-cleaned area, the height data of the obstacle and the auxiliary measurement points to generate a path set. By optimizing the cleaning sequence, the invalid flight distance and energy consumption of the unmanned aerial vehicle are reduced, and the overall operation efficiency and economical efficiency are improved; and by determining the optimal candidate cleaning path, the collision risk is avoided while the distance is optimal.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power station technology, specifically relating to an intelligent drone inspection and cleaning method and system for photovoltaic power stations. Background Technology

[0002] The surface cleanliness of photovoltaic modules directly affects their photoelectric conversion efficiency, which in turn affects the stable operation of large-scale photovoltaic power plants and the stability of energy supply. Since large-scale photovoltaic power plants are mostly located in vast environments, drones, with their high mobility and convenience, have become one of the means to carry out inspection and cleaning tasks for photovoltaic power plants.

[0003] In existing technologies, the use of drones for photovoltaic power plant operation and maintenance usually treats inspection and cleaning as independent tasks. Cleaning decisions are mostly based on pre-set plans, lacking analysis of the degree of contamination on the photovoltaic panel surface. This leads to unnecessary cleaning operations on photovoltaic modules with only slight or small-area contamination, wasting drone endurance and cleaning resources, and increasing equipment wear and tear. Furthermore, existing technologies prioritize minimizing the cleaning flight distance when planning cleaning paths, neglecting the inspection path the drone is currently executing. This results in the drone needing to return to the inspection interruption point after completing the cleaning task to continue inspection, increasing the total flight time and energy consumption of a single mission, reducing the effective inspection coverage per flight, and making it difficult to improve the operation and maintenance efficiency of photovoltaic power plants.

[0004] In view of this, this application provides an intelligent drone inspection and cleaning method and system for photovoltaic power plants. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent drone inspection and cleaning method and system for photovoltaic power plants. The system can determine whether there are areas in the photovoltaic power plant that need to be cleaned while the drone is conducting inspections, and after determining the areas that need to be cleaned, it can plan the corresponding cleaning path and generate the corresponding cleaning instructions, so that the drone can complete the cleaning operation during the inspection process.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A smart drone inspection and cleaning method for photovoltaic power plants, when the pollution risk level of the area to be inspected is determined to be higher than a preset risk threshold, executes the following: Generate cleaning tasks for drones; deploy cleaning tasks to control drones to perform cleaning operations on the areas to be cleaned according to the cleaning sequence and cleaning path; The process of generating a cleaning task for drones includes: determining the cleaning order of the areas to be cleaned, and determining the cleaning path for the drone to reach the areas to be cleaned. The process of determining the cleaning order of the areas to be cleaned includes: obtaining the three-dimensional center point of the area to be cleaned and calculating the horizontal and vertical distances between the three-dimensional center points; dividing the area to be cleaned into same-group areas and different-group areas based on the horizontal and vertical distances; determining the weight parameters of the area to be cleaned based on the pollution risk level of the area to be cleaned; and sorting the areas to be cleaned based on the weight parameters and their respective groups to generate the cleaning order. The process of determining the cleaning path for the drone to reach the area to be cleaned includes: determining the centroid of the corresponding planar image of the current area to be cleaned as the center point; determining auxiliary measurement points around the area to be cleaned; acquiring the height data of surrounding buildings; generating a path set containing candidate cleaning paths based on the center point, auxiliary measurement points, and height data; and selecting the shortest candidate cleaning path from the path set as the cleaning path.

[0007] Preferably, determining the pollution risk level of the area to be detected includes: Acquire a planar image of the area to be detected; analyze the degree of color shift in the planar image to identify contaminated areas where the color shift degree is higher than a preset color shift threshold; calculate the area of ​​the contaminated area; and determine the contamination risk level of the area to be detected based on the area and a preset area threshold.

[0008] Preferably, the calculation of the area of ​​the contaminated zone includes: Project the contaminated area onto a plane to generate a projected region; divide the projected region into sub-regions; calculate the area of ​​each sub-region and sum them to obtain the area of ​​the contaminated region.

[0009] Preferably, the analysis of the color shift degree of the planar image to determine the contaminated areas where the color shift degree is higher than a preset color shift threshold includes: Obtain the offset reference value; extract the color parameters within the planar image; calculate the color offset of the points based on the color parameters; calculate the ratio of the color offset to the offset reference value to obtain the average color offset amplitude, and define the average color offset amplitude as the degree of color offset.

[0010] Preferably, obtaining the offset reference value includes: Obtain the clean area and designate it as the calibration reference surface; set up comparison surfaces with different tilt angles; measure the average illumination value on the comparison surfaces and calculate the average illumination difference between the comparison surfaces and the calibration reference surface to obtain the color deviation; obtain the light transmission difference at different tilt angles; add the color deviation and the light transmission difference to generate the offset reference value.

[0011] Preferably, the sorting of the areas to be cleaned based on the weight parameters and their respective groups to generate the cleaning order includes: areas to be cleaned with a higher pollution risk level are given priority over areas to be cleaned with a lower pollution risk level, and areas to be cleaned within the same group are given priority over areas to be cleaned in different groups.

[0012] Preferably, determining auxiliary measurement points around the area to be cleaned includes: If the area to be cleaned is above the preset reference horizontal plane, auxiliary measurement points with different offset heights are set based on the vertical plane of the building where the area to be cleaned is located; if the area to be cleaned is below the preset reference horizontal plane, the position of the auxiliary measurement points is determined by different plane elevation angles, and the plane elevation angle is adjusted based on the solar altitude angle.

[0013] A smart drone inspection and cleaning system for photovoltaic power plants includes: The pollution identification module is used to analyze the degree of color shift in the area to be detected in order to determine the pollution risk level of the area to be detected. The cleaning task planning module is used to generate a cleaning task for the drone in response to the pollution risk level determined by the pollution identification module being higher than the preset risk threshold. The cleaning task includes the cleaning sequence of the area to be cleaned and the cleaning path to reach the area to be cleaned. The drone mission deployment module is used to deploy cleaning tasks generated by the cleaning mission planning module, so as to control the drone to perform cleaning operations according to the cleaning sequence and cleaning path.

[0014] Preferably, the analysis of the color shift in the area to be detected to determine the pollution risk level of the area to be detected includes: Acquire 3D point cloud data of the area to be inspected in a photovoltaic power station, and construct a 3D model based on the 3D point cloud data; Within the area defined by the 3D model, a planar image of the area is acquired; the degree of color shift is analyzed based on the planar image to identify contaminated areas where the degree of color shift exceeds a preset color shift threshold; the area of ​​the contaminated area is calculated; and the contamination risk level of the area to be detected is determined based on the area and a preset area threshold.

[0015] Preferably, when the pollution risk level determined by the pollution identification module is higher than a preset risk threshold, a cleaning task for the drone is generated, including: Based on the pollution risk level of the areas to be cleaned and the horizontal and vertical distances between the areas to be cleaned, the cleaning order of the areas to be cleaned is determined; based on the center point of the area to be cleaned, auxiliary measurement points, and the height data of surrounding buildings, a set of paths containing candidate cleaning paths is generated, and the shortest candidate cleaning path is determined as the cleaning path to the area to be cleaned. Beneficial effects

[0016] This invention obtains offset reference values ​​through a calibration step, and analyzes the degree of color shift in the area to be detected based on the offset reference values ​​to determine the contaminated area, compensate for uneven lighting caused by tilt angle and lighting changes, eliminate ambient lighting interference, and improve the accuracy and reliability of determining the contaminated area. This invention determines the pollution risk level of a contaminated area based on its area, groups the determined areas to be cleaned into groups, and sorts them using weight parameters to generate a cleaning sequence, enabling on-demand cleaning. By prioritizing the cleaning of areas with high pollution risk levels and concentrated locations, the cleaning sequence is optimized, reducing the ineffective flight distance and energy consumption of drones, and improving overall operational efficiency and economy. When planning the cleaning path, this invention integrates the location of the area to be cleaned, the height data of obstacles, and auxiliary measurement points to generate a path set, and determines the optimal candidate cleaning path from this path set. This achieves optimal distance while avoiding collision risks, enabling drones to safely perform cleaning operations in complex three-dimensional environments and enhancing the stability and reliability of intelligent drone inspection and cleaning. Attached Figure Description

[0017] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a flowchart illustrating the process of determining the cleaning sequence according to the present invention; Figure 3 This is a flowchart illustrating the process of determining the cleaning path according to the present invention; Figure 4 This is a system module diagram of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely for explaining the invention and are not intended to limit the scope of protection of the invention. Example 1

[0019] Please see Figures 1-3 This embodiment discloses an intelligent drone inspection and cleaning method for photovoltaic power plants, including the following steps: Using a drone-mounted 3D laser scanning device or high-precision oblique photogrammetry device, spatial data is collected from all areas to be inspected in the photovoltaic power station to obtain 3D point cloud data of the areas to be inspected. Through a preset surface reconstruction processing procedure, the discrete points in the 3D point cloud data are topologically connected to generate a continuous surface composed of polygonal meshes, thereby constructing the 3D data structure of the areas to be inspected. On the three-dimensional data structure of the area to be detected, the boundary points of the photovoltaic panel array or building outline are identified by calculating the abrupt changes in surface normal vectors, spatial coordinates or pixel gray values ​​through a preset edge recognition processing procedure. These boundary points together define the range of the area to be detected. Within the area defined by the boundary points, orthophoto correction processing converts the view information on the three-dimensional data structure into a two-dimensional planar image of the area that eliminates perspective distortion. By constructing a three-dimensional data structure and acquiring planar images, the spatial area to be detected is standardized and planarized, providing a uniform scale and visually distortion-free input data foundation for pollution identification and risk decision-making.

[0020] Furthermore, based on the obtained planar image, the contaminated area is identified, and multiple reference points such as the four corner points or geometric center points of the photovoltaic panel are determined in the area to be detected. A reference surface that can represent the visual characteristics of the photovoltaic panel under ideal clean conditions is generated on the planar image, and the reference surface is defined as the verification area. The analysis of the color shift in the verification area includes: obtaining a shift reference value as a quantitative benchmark to distinguish between color changes caused by normal environmental factors such as lighting and viewing angle and color changes caused by real pollutants such as dust, bird droppings, and snow; after obtaining the shift reference value, the verification area is analyzed; based on the analysis of the verification area, the contaminated area is identified; and the coordinate position of the contaminated area on the planar image is recorded as input data for subsequent risk decision-making and task planning. The process of generating an offset reference value through calibration steps includes: obtaining a confirmed clean photovoltaic panel area within the photovoltaic power station as a calibration reference surface; setting up multiple comparison surfaces with different tilt angles within the photovoltaic power station to simulate different lighting conditions; measuring the average lighting value on the comparison surfaces using a light sensor and calculating the average lighting difference between the comparison surfaces and the calibration reference surface to obtain the color deviation caused by different tilt angles; obtaining the difference in transmitted light intensity caused by different refraction and reflection effects when light penetrates the photovoltaic panel's glass cover at different tilt angles, as the light transmission difference; and combining the color deviation and the light transmission difference by weighted summation to generate an offset reference value that integrates the effects of angle and light transmission. The analysis of the verification area includes: extracting color parameters within the verification area, converting image data to color spaces more consistent with human visual perception such as HSV or Lab, and extracting parameters such as hue, saturation, and brightness from multiple sampling points; based on the color parameters, calculating the Euclidean distance between the current point's color parameters and the ideal clean state in a multi-dimensional color space as the color offset to quantify the difference in color parameters, thereby obtaining the color offset of multiple sampling points within the verification area; calculating the ratio of the color offset to the offset reference value to obtain the normalized average color offset amplitude reflecting the color change caused by pollution after eliminating environmental interference; defining the average color offset amplitude as the color offset degree, and identifying areas within the verification area where the color offset degree exceeds a preset color offset threshold as polluted areas.

[0021] Furthermore, for all contaminated areas identified on the planar image, in-plane projection is performed to correct the irregular shapes under the tilted view to the true shapes under the frontal view, thereby generating a projection area that eliminates perspective distortion and avoids inaccurate area calculations due to the tilted angle of the drone shooting. To calculate the area of ​​complex shapes, the projection area is divided into multiple sub-regions with regular square or triangular shapes through geometric processing methods such as meshing or triangulation. The areas of multiple sub-regions are calculated and summed to obtain the accurate area of ​​the contaminated area. After obtaining the precise area of ​​the contaminated area, the pollution risk level and the area to be cleaned are determined, including: based on the precise area of ​​the contaminated area, the pollution risk level of the area to be tested is determined according to the preset area threshold division rules; If the pollution risk level is higher than the preset risk threshold, the area to be detected will be identified as the area to be cleaned, and its identifier and related data will be included in the processing queue. Specifically, the pollution risk level can be divided into three levels: low, medium, and high. Each level corresponds to a different preset area threshold range. Alternatively, the preset risk threshold can be set as the lowest preset area threshold corresponding to the high pollution risk level, thereby identifying the area to be tested corresponding to the high pollution risk level as the area to be cleaned.

[0022] Furthermore, the cleaning sequence and cleaning path are determined. For all areas identified as to be cleaned, task planning is performed, which includes determining the cleaning sequence and the cleaning path for the drone. When determining the cleaning order of the areas to be cleaned, the coordinates of the three-dimensional center points of each area to be cleaned are obtained from the constructed three-dimensional data structure, and the horizontal and vertical distances between any two three-dimensional center points are calculated. Based on the horizontal and vertical distances and in accordance with preset horizontal and vertical distance thresholds, the areas to be cleaned are divided into same-group areas and different-group areas. Then, weight parameters are calculated for all areas to be cleaned. These weight parameters are determined by directly mapping or weighting the values ​​of the pollution risk levels of the areas to be cleaned. Finally, the areas to be cleaned are sorted according to the sorting rules to generate the cleaning order for the drone to perform the cleaning operation. The process of dividing the area to be cleaned into same-group areas and different-group areas includes: areas with a horizontal distance less than a preset horizontal distance threshold and a vertical distance less than a preset vertical distance threshold are divided into same-group areas, which means that the same-group areas are physically adjacent and at the same height level; otherwise, they are different-group areas that are relatively dispersed in space or at different height levels. The sorting rules are as follows: areas with higher weight parameters are prioritized over areas with lower weight parameters; when the weight parameters are the same, areas within the same group are prioritized over areas in different groups, so as to first concentrate on processing high-risk areas and then process them nearby, thereby reducing the large-scale maneuvering of drones; and finally, an ordered list of cleaned areas is generated. When determining the cleaning path for the drone, the location coordinates of the current area to be cleaned in the cleaning area list are obtained, the geometric centroid of the corresponding planar image of the area to be cleaned is determined, and the centroid is defined as the center point as the core reference point for path planning. Around the area to be cleaned, in order to ensure the safety and efficiency of the drone flight, multiple auxiliary measurement points are determined to define the trajectory of the drone for safe approach, hovering operation and withdrawal. Furthermore, to avoid interference with power generation from the drone's shadow on nearby photovoltaic panels, the plane elevation angle needs to be dynamically adjusted based on the current solar altitude angle. By acquiring the height data of surrounding buildings, a three-dimensional obstacle avoidance data structure containing no-fly zone information is established. Based on the center point, auxiliary measurement points, and obstacle avoidance data structure, a path set containing multiple candidate cleaning paths is generated. From the path set, the optimal path is selected as the cleaning path through a preset path evaluation procedure. In determining the location of auxiliary measurement points, it is necessary to determine the position of the area to be cleaned relative to the preset reference horizontal plane. If the area to be cleaned is above the preset reference horizontal plane, multiple auxiliary measurement points with different offset heights are set based on the vertical plane of the building where the area to be cleaned is located; if the area to be cleaned is below the preset reference horizontal plane, the positions of multiple auxiliary measurement points are determined by different plane elevation angles in order to plan a path to approach from below. The path evaluation procedure comprehensively scores each candidate path based on multiple indicators, such as its length, estimated flight time, and minimum distance to obstacles, to select the cleaning path with the lowest overall cost and that meets safety constraints.

[0023] Furthermore, based on the determined cleaning path and cleaning sequence, cleaning task instructions for controlling the UAV are generated; wherein, the cleaning task instructions are a set of structured data, including specific flight path waypoint coordinates, hovering position, fuselage attitude angle during cleaning operation, and control parameters such as water output of the cleaning device. The cleaning task instruction is sent to the drone terminal via wireless communication. The drone terminal's flight control parses the instruction and controls the drone to perform cleaning operations on each area to be cleaned in the cleaning area list in sequence according to the cleaning path and cleaning order. After each area to be cleaned is completed, its status is automatically updated and the drone is guided to fly to the next target in the list. After completing the task of cleaning all areas, the drone can continue to perform subsequent inspection tasks, or automatically return to the preset take-off and landing point according to instructions, so as to carry out efficient inspection and cleaning in the operation and maintenance of photovoltaic power plants. Example 2

[0024] Please see Figure 4 This embodiment discloses an intelligent drone inspection and cleaning system for photovoltaic power plants. The system interacts with a drone equipped with cleaning capabilities via a wireless communication network and issues task commands. The system includes the following modules: The pollution identification module is used to analyze the degree of color shift in the area to be detected in order to determine the pollution risk level of the area to be detected; it acquires three-dimensional point cloud data of the area to be detected in the photovoltaic power station through lidar or multi-view camera mounted on a drone, and constructs a three-dimensional data structure containing precise position and attitude information based on the three-dimensional point cloud data; Within the area defined by the three-dimensional data structure, control a drone or call a fixed camera to acquire planar images of the area, and analyze the degree of color shift in the planar images to accurately quantify the degree of pollution; The calibration process analyzes the degree of color shift in a planar image, including: acquiring a confirmed clean photovoltaic panel area and designating it as the calibration reference surface; setting up comparison surfaces with different tilt angles under similar conditions, measuring the average illumination value on the comparison surfaces, and calculating the average illumination difference between the comparison surfaces and the calibration reference surface to obtain the color deviation amount. Obtain the light transmission difference under different tilt angles, and combine the color deviation and the light transmission difference according to the preset calculation relationship to generate an offset reference value that takes into account the influence of light and angle. After obtaining the offset reference value, the color parameters such as RGB and HSV in the plane image to be analyzed are extracted, the color offset of each point in the image is calculated, and the average color offset amplitude is calculated as the color offset degree. The color offset degree is compared with the set color offset threshold, and the area above the threshold is identified as the pollution area. Calculating the area of ​​the contaminated area to determine the severity of the contamination includes: projecting the identified irregular contaminated area into a two-dimensional plane to generate a projected area; for ease of calculation, the projected area can be divided into multiple regular rectangular or triangular sub-regions; calculating and summing the areas of all sub-regions to obtain the precise area of ​​the contaminated area; and determining the contamination risk level of the area to be detected based on the area and according to a preset area threshold division rule.

[0025] The cleaning task planning module is activated when the pollution risk level of a certain area to be detected is higher than the preset risk threshold. The module generates a cleaning task for the drone. The generation of the cleaning task includes determining the cleaning order of the areas to be cleaned and determining the cleaning path for the drone to reach the areas to be cleaned. When determining the cleaning sequence, obtain the three-dimensional center points of all areas to be cleaned that are determined to need cleaning, and calculate the horizontal and vertical distances between each pair of three-dimensional center points. Based on horizontal and vertical distances, and in accordance with preset horizontal and vertical distance thresholds, the area to be cleaned is divided into same-group areas and different-group areas. Among them, areas with a horizontal distance less than a preset horizontal distance threshold and a vertical distance less than a preset vertical distance threshold are classified as the same group of areas, and vice versa. Weight parameters are determined for each area to be cleaned based on its pollution risk level, with higher weights for more severe pollution. All areas to be cleaned are then sorted according to their weight parameters and groupings to generate the final cleaning order. The cleaning order is determined by the following rules: areas with higher weight parameters are prioritized over areas with lower weight parameters. Furthermore, when weight parameters are similar, areas within the same group are prioritized over areas in different groups to reduce the large-scale movement of drones.

[0026] When determining the cleaning path, path planning is performed for all current areas to be cleaned in the cleaning sequence. It determines the centroid of the corresponding planar image of the area as the center point of the path planning, and determines several auxiliary measurement points around the area to be cleaned to construct a safe flight route. When determining the location of the auxiliary measurement points, their spatial position is considered. If the area to be cleaned is located above a preset ground level, auxiliary measurement points with different offset heights are set based on the vertical plane of the building where the area is located. If the photovoltaic array is on a hillside and the area to be cleaned is below the reference level, different plane elevation angles are used to determine the location of the auxiliary measurement points, and the plane elevation angle can be fine-tuned based on the current solar altitude angle to obtain the best observation and obstacle avoidance effect. The system acquires height data of surrounding buildings to establish a three-dimensional obstacle avoidance data structure that includes no-fly zone information. Based on the center point, auxiliary measurement points, and obstacle avoidance data structure, it uses path planning algorithms such as the RRT algorithm to generate a path set containing multiple candidate cleaning paths. Through a preset path evaluation procedure, it selects the candidate cleaning path with the shortest length that meets all safety constraints from the path set and determines it as the cleaning path.

[0027] The drone mission deployment module receives a complete cleaning mission, which includes a clear list of cleaning areas, a strict cleaning sequence, and a precise cleaning path planned for each area. This logical task is transformed into cleaning task instructions that the UAV flight control system can recognize and execute. The cleaning task instructions specifically include flight task files containing instructions such as waypoint coordinates, flight speed, hovering time, and start / stop of the cleaning device. This task is deployed to a designated drone via a communication link; after receiving the task, the drone performs cleaning operations on the area to be cleaned according to the cleaning sequence and cleaning path until all task points are completed.

[0028] By accurately sensing, making decisions, and handling pollution in photovoltaic power plants, a system for routine maintenance, such as cleaning, has been developed that is applicable to large-scale photovoltaic power plants with complex terrain.

[0029] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligent unmanned aerial vehicle (UAV) inspection and cleaning of photovoltaic power plants, characterized in that, When the pollution risk level of the area to be tested is determined to be higher than the preset risk threshold, the following steps are executed: Generate cleaning tasks for drones; Deploy cleaning tasks to control drones to perform cleaning operations on the areas to be cleaned according to the cleaning sequence and cleaning path; The process of generating a cleaning task for drones includes: determining the cleaning sequence of the areas to be cleaned, and determining the cleaning path for the drone to reach the areas to be cleaned. The process of determining the cleaning order of the areas to be cleaned includes: obtaining the three-dimensional center point of the area to be cleaned and calculating the horizontal and vertical distances between the three-dimensional center points; dividing the area to be cleaned into same-group areas and different-group areas based on the horizontal and vertical distances; determining the weight parameters of the area to be cleaned based on the pollution risk level of the area to be cleaned; and sorting the areas to be cleaned based on the weight parameters and their respective groups to generate the cleaning order. The process of determining the cleaning path for the drone to reach the area to be cleaned includes: determining the centroid of the corresponding planar image of the current area to be cleaned as the center point; determining auxiliary measurement points around the area to be cleaned; acquiring the height data of surrounding buildings; generating a path set containing candidate cleaning paths based on the center point, auxiliary measurement points, and height data; and selecting the shortest candidate cleaning path from the path set as the cleaning path.

2. The intelligent unmanned aerial vehicle (UAV) inspection and cleaning method for photovoltaic power plants according to claim 1, characterized in that, The determination of the pollution risk level of the area to be tested includes: Acquire a planar image of the area to be detected; analyze the degree of color shift in the planar image to identify contaminated areas where the color shift degree is higher than a preset color shift threshold; calculate the area of ​​the contaminated area; and determine the contamination risk level of the area to be detected based on the area and a preset area threshold.

3. The intelligent drone inspection and cleaning method for photovoltaic power plants according to claim 2, characterized in that, The area of ​​the contaminated region being calculated includes: Project the contaminated area onto a plane to generate a projected region; divide the projected region into sub-regions; calculate the area of ​​each sub-region and sum them to obtain the area of ​​the contaminated region.

4. The intelligent unmanned aerial vehicle (UAV) inspection and cleaning method for photovoltaic power plants according to claim 2, characterized in that, The analysis of the color shift in the planar image is used to determine contaminated areas where the color shift exceeds a preset color shift threshold, including: Obtain the offset reference value; extract the color parameters within the planar image; calculate the color offset of the points based on the color parameters; calculate the ratio of the color offset to the offset reference value to obtain the average color offset amplitude, and define the average color offset amplitude as the degree of color offset.

5. A method for intelligent drone inspection and cleaning of photovoltaic power plants according to claim 4, characterized in that, Obtaining the offset reference value includes: Obtain the clean area and designate it as the calibration reference surface; set up comparison surfaces with different tilt angles; measure the average illumination value on the comparison surfaces and calculate the average illumination difference between the comparison surfaces and the calibration reference surface to obtain the color deviation; obtain the light transmission difference at different tilt angles; add the color deviation and the light transmission difference to generate the offset reference value.

6. The intelligent unmanned aerial vehicle (UAV) inspection and cleaning method for photovoltaic power plants according to claim 1, characterized in that, The process of sorting the areas to be cleaned based on weight parameters and their respective groups to generate a cleaning order includes: areas with higher pollution risk levels are prioritized over areas with lower pollution risk levels, and areas within the same group are prioritized over areas in different groups.

7. The intelligent unmanned aerial vehicle (UAV) inspection and cleaning method for photovoltaic power plants according to claim 1, characterized in that, The determination of auxiliary measurement points around the area to be cleaned includes: If the area to be cleaned is above the preset reference horizontal plane, auxiliary measurement points with different offset heights are set based on the vertical plane of the building where the area to be cleaned is located; if the area to be cleaned is below the preset reference horizontal plane, the position of the auxiliary measurement points is determined by different plane elevation angles, and the plane elevation angle is adjusted based on the solar altitude angle.

8. An intelligent unmanned aerial vehicle (UAV) inspection and cleaning system for photovoltaic power plants, characterized in that, include: The pollution identification module is used to analyze the degree of color shift in the area to be detected in order to determine the pollution risk level of the area to be detected. The cleaning task planning module is used to generate a cleaning task for the drone in response to the pollution risk level determined by the pollution identification module being higher than the preset risk threshold. The cleaning task includes the cleaning sequence of the area to be cleaned and the cleaning path to reach the area to be cleaned. The drone mission deployment module is used to deploy cleaning tasks generated by the cleaning mission planning module, so as to control the drone to perform cleaning operations according to the cleaning sequence and cleaning path.

9. A smart drone inspection and cleaning system for photovoltaic power plants according to claim 8, characterized in that, The analysis of color shift in the area to be detected to determine the pollution risk level of the area to be detected includes: Acquire 3D point cloud data of the area to be inspected in a photovoltaic power station, and construct a 3D model based on the 3D point cloud data; Within the area defined by the 3D model, a planar image of the area is acquired; the degree of color shift is analyzed based on the planar image to identify contaminated areas where the degree of color shift exceeds a preset color shift threshold; the area of ​​the contaminated area is calculated; and the contamination risk level of the area to be detected is determined based on the area and a preset area threshold.

10. The intelligent unmanned aerial vehicle (UAV) inspection and cleaning system for photovoltaic power plants according to claim 8, characterized in that, If the pollution risk level determined by the pollution identification module is higher than a preset risk threshold, a cleaning task for the drone is generated, including: Based on the pollution risk level of the areas to be cleaned and the horizontal and vertical distances between the areas to be cleaned, the cleaning order of the areas to be cleaned is determined; based on the center point of the area to be cleaned, auxiliary measurement points, and the height data of surrounding buildings, a set of paths containing candidate cleaning paths is generated, and the shortest candidate cleaning path is determined as the cleaning path to the area to be cleaned.

Images