Unmanned aerial vehicle intelligent positioning and cleaning method and system for photovoltaic pollution

Abstract

This invention provides an intelligent positioning and cleaning method and system for photovoltaic (PV) pollution using drones, relating to the field of PV array cleaning and maintenance technology. First, it acquires a set of pollution-related images, flight parameters, and power generation efficiency-related data collected by the drone while flying over the PV array. Based on this data, it performs pollution area positioning processing to generate spatial positioning information. Then, it plans a pollution cleaning path by combining the spatial positioning information with pollution diffusion trend data and power generation priority data. Following the path, it controls the drone-mounted cleaning device to perform the cleaning operation, simultaneously collecting real-time pollution change images and power generation efficiency change data to generate operation results. Finally, based on the operation results and related data, it evaluates the cleaning effect and generates pollution cleaning assessment information including supplementary cleaning suggestions, achieving intelligent and efficient PV array cleaning.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic array cleaning and maintenance technology, and more specifically, to a drone-based intelligent positioning and cleaning method and system for photovoltaic pollution. Background Technology

[0002] With the increasing widespread application of photovoltaic energy, the cleanliness and maintenance of photovoltaic arrays are crucial to ensuring their power generation efficiency and stability. Due to long-term exposure to the natural environment, the surface of photovoltaic arrays easily accumulates pollutants such as dust, bird droppings, and leaves. These pollutants can block sunlight, reduce the light absorption efficiency of photovoltaic modules, and thus affect the power generation performance of the entire photovoltaic array.

[0003] Currently, cleaning of photovoltaic (PV) contamination mainly relies on manual cleaning or fixed cleaning equipment. While manual cleaning offers flexibility in handling contamination in different areas, it suffers from high labor intensity, low efficiency, and inconsistent cleaning quality. Furthermore, for large PV power plants or areas with complex terrain, the difficulty and cost of manual cleaning increase significantly. Fixed cleaning equipment, although capable of automated cleaning, typically has pre-set cleaning paths and ranges, making it difficult to dynamically adjust based on the actual contamination levels of the PV array. This can easily lead to incomplete or over-cleaning, resulting in water waste and poor cleaning effectiveness. In addition, most existing cleaning methods lack real-time evaluation and feedback mechanisms for cleaning effectiveness, failing to promptly identify and address uncleaned areas, thus hindering the long-term efficient operation of the PV array. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a drone-based intelligent positioning and cleaning method for photovoltaic pollution, the method comprising:

[0005] Acquire a set of pollution-related images collected by a drone while flying over a photovoltaic array, drone flight parameters, and photovoltaic array power generation efficiency-related data. The pollution-related image set contains visual information on surface pollution in each area of ​​the photovoltaic array. The drone flight parameters include the drone's flight altitude and flight angle when the drone collects pollution-related images of each area. The power generation efficiency-related data contains real-time power generation output information of each area of ​​the photovoltaic array.

[0006] Based on the pollution-related image set, UAV flight parameters and power generation efficiency-related data, pollution area localization processing is performed to generate spatial positioning information of the pollution area in the photovoltaic array. The spatial positioning information includes the planar coordinates of the pollution area in the photovoltaic array and a description of the area range.

[0007] By combining the spatial positioning information with pollution diffusion trend data of the polluted area and power generation priority data of the photovoltaic array, a pollution cleaning path for the UAV is planned. The pollution cleaning path includes the flight trajectory of the UAV between different polluted areas and the order of cleaning operations.

[0008] According to the pollution cleaning path, the cleaning device carried by the UAV performs pollution cleaning operation. During the process, real-time pollution change images and real-time power generation efficiency change data of the cleaning area are collected simultaneously to generate pollution cleaning operation results of the UAV. The pollution cleaning operation results include cleaning completion status information of each pollution area.

[0009] Based on the results of the pollution cleaning operation, real-time pollution change images, and real-time power generation efficiency change data, the cleaning effect is evaluated to generate pollution cleaning assessment information for the photovoltaic array. The pollution cleaning assessment information includes supplementary cleaning suggestions for areas that have not been cleaned.

[0010] In another aspect, embodiments of the present invention also provide an intelligent positioning and cleaning system for photovoltaic pollution using unmanned aerial vehicles (UAVs), including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to run the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0011] Based on the above, this embodiment of the invention simultaneously collects pollution-related image sets, UAV flight parameters, and photovoltaic array power generation efficiency-related data. Based on this, it performs pollution area localization processing to generate spatial positioning information, clarifying the location and extent of the pollution area within the photovoltaic array. Combining spatial positioning information with pollution diffusion trend data and power generation priority data, it plans a pollution cleaning path, making the UAV cleaning operation more scientific and rational. This prioritizes cleaning areas with a significant impact on power generation efficiency while effectively preventing further pollution spread. During the cleaning process, real-time pollution change images and real-time power generation efficiency change data are simultaneously collected, and pollution cleaning operation results are generated accordingly, enabling real-time monitoring and dynamic adjustment of the cleaning process. Finally, based on the cleaning operation results and related data, a cleaning effect evaluation is performed. The generated pollution cleaning evaluation information includes supplementary cleaning suggestions for areas that were not properly cleaned, enabling timely identification and resolution of problems during the cleaning process. This significantly improves the cleaning efficiency and quality of the photovoltaic array, ensuring its long-term stable power generation. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the execution flow of the UAV intelligent positioning and cleaning method for photovoltaic pollution provided in an embodiment of the present invention.

[0013] Figure 2This is a schematic diagram of the hardware architecture of the drone intelligent positioning and cleaning system for photovoltaic pollution provided in an embodiment of the present invention. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an embodiment of the intelligent positioning and cleaning method for photovoltaic pollution using drones. The following is a detailed description of this intelligent positioning and cleaning method for photovoltaic pollution using drones.

[0015] Step S110: Obtain a set of pollution-related images collected by the UAV while flying over the photovoltaic array, UAV flight parameters, and power generation efficiency-related data of the photovoltaic array. The pollution-related image set contains visual information on surface pollution in each area of ​​the photovoltaic array. The UAV flight parameters include the UAV flight altitude and flight angle when the UAV collects pollution-related images of each area. The power generation efficiency-related data contains real-time power generation output information of each area of ​​the photovoltaic array.

[0016] This embodiment uses a large-scale grid-connected photovoltaic power station as an application scenario. The photovoltaic array in this power station consists of multiple rows of photovoltaic modules, each row containing multiple photovoltaic panels connected in series, arranged in a matrix. After taking off from the power station's helipad, the drone enters a designated flight area above the photovoltaic array. During the flight, the drone's multispectral camera and high-definition visible light camera work together to continuously capture images of the photovoltaic array surface, forming a pollution-related image set. In these images, some areas show obvious dust coverage, some areas have bird droppings, and some areas may have debris such as leaves; these are all visual information related to surface pollution.

[0017] The GPS module, barometric pressure sensor, and gyroscope installed on the drone record its flight status in real time. When capturing an image of a photovoltaic panel, the barometric pressure sensor records the flight altitude, i.e., the drone's vertical distance from the ground; the gyroscope records the flight angles, including the drone's pitch, roll, and yaw angles. Each area of ​​the photovoltaic array is connected to a micro-inverter, which monitors the power generation in its area in real time, uploading real-time voltage, current, and other information to the power plant's monitoring system to generate real-time power output information, i.e., power generation efficiency data.

[0018] Step S111: Determine the initial flight coverage frame of the UAV based on the overall layout and orientation of the photovoltaic array. The initial flight coverage frame includes the basic flight direction of the UAV over the photovoltaic array.

[0019] In this large-scale grid-connected photovoltaic power station, the photovoltaic array is arranged in a north-south direction, with passageways for maintenance personnel between each row of photovoltaic modules, facing due south to maximize solar radiation. The overall layout information, including the boundary coordinates of the photovoltaic array and the length and width of each row of modules, was obtained through the power station's design drawings and on-site surveys.

[0020] Based on the orientation of the photovoltaic array, the initial flight coverage frame of the UAV is determined. The basic flight direction is set to fly east-west, so that the UAV's flight path is perpendicular to the arrangement direction of the photovoltaic array, ensuring that images of multiple rows of photovoltaic modules can be captured in a single flight. The initial flight coverage frame also includes the start point and the end point. The start point is set above the easternmost edge of the photovoltaic array, and the end point is set above the westernmost edge to ensure complete coverage of the entire photovoltaic array area.

[0021] Step S112: Control the UAV to fly over the photovoltaic array according to the initial flight coverage framework. During the flight, dynamically adjust the UAV's flight altitude according to the tilt angle and illumination angle of the photovoltaic array to maintain a stable distance between the UAV and the photovoltaic array surface.

[0022] The drone began its flight according to the initial flight coverage framework, moving east-west from the starting point towards the destination. The photovoltaic array of this power station has a fixed tilt angle of 30 degrees to adapt to the local latitude. Meanwhile, the angle of sunlight changes over time, with lower angles in the morning and evening, and higher angles at midday.

[0023] The drone's onboard angle sensor measures the tilt angle of the photovoltaic array and the current illumination angle in real time. Based on the measured angle data, the drone's flight control system calculates the appropriate flight altitude. For example, when the illumination angle is low, to avoid glare from the photovoltaic array surface affecting image acquisition, the flight control system controls the drone to increase its flight altitude appropriately; when the illumination angle is high, it decreases its flight altitude appropriately. Through this dynamic adjustment, the drone maintains a stable distance of 6 meters between itself and the photovoltaic array surface, ensuring that the captured images clearly show details while covering a sufficiently large area.

[0024] Step S113: Using the image acquisition device mounted on the UAV, the pollution-related image acquisition angle is synchronously adjusted according to the flight angle of the UAV during flight to acquire initial pollution-related images of each area of ​​the photovoltaic array. The initial pollution-related images contain the original visual information of the photovoltaic array surface.

[0025] The drone is equipped with image acquisition devices including high-definition visible light cameras and multispectral cameras, which are mounted on a rotatable gimbal. During flight, the drone's flight control system transmits real-time flight angle information to the gimbal control system. The gimbal control system then adjusts the shooting angle of the image acquisition devices in real time based on the flight angle, ensuring that the camera lens is always directly facing the photovoltaic array surface.

[0026] For example, when the drone's pitch angle changes, the gimbal will adjust the camera's pitch angle accordingly to ensure that the shooting direction remains perpendicular to the photovoltaic array surface or at the optimal shooting angle. Under the above adjustments, the image acquisition device continuously collects initial pollution-related images of various areas of the photovoltaic array. These initial pollution-related images contain the original visual information of the photovoltaic array surface, such as the blue background color of the photovoltaic panels, the distribution of dust on the surface, the shape and color of bird droppings, and the outline of leaves.

[0027] Step S114: Extract the pollution visual information of the photovoltaic array surface from the initial pollution-associated image. The pollution visual information includes color difference features, texture change features, and morphological contour features of the pollution area.

[0028] The initial pollution-related images were processed to extract the visual pollution information. First, the color difference characteristics were observed. The normal photovoltaic panel surface appeared uniformly blue, while the polluted areas showed different colors. Areas covered by dust were darker, appearing gray; bird droppings areas were mostly white or gray, forming a clear contrast with the surrounding blue.

[0029] Next, the texture change characteristics were analyzed. The surface texture of a normal photovoltaic panel is smooth and uniform, while the texture changes in contaminated areas. Areas with thicker dust accumulation will show a rough texture, and areas with bird droppings will have a more irregular texture due to their sticky nature.

[0030] Then, morphological contour features were identified, and different types of pollution exhibited different morphological contours. Dust pollution often presents as a large, continuous distribution with a relatively blurred shape; bird droppings pollution presents as dot-like or block-like contours with relatively clear boundaries; and debris such as leaves have irregular shapes. Through these analyses, the extracted color difference features, texture variation features, and morphological contour features were used as visual information about pollution.

[0031] Step S115: Record the drone's flight altitude and flight angle when the drone collects each initial pollution-related image as drone flight parameters.

[0032] The drone's flight control system works synchronously with the image acquisition equipment. When the image acquisition equipment captures an initial pollution-related image, the flight control system simultaneously records the flight altitude and flight angle. The flight altitude is calculated using a combination of barometric pressure sensor and GPS data, accurate to 0.1 meters; the flight angle is measured using a gyroscope, including pitch, roll, and yaw angles, accurate to 0.1 degrees.

[0033] These recorded data are linked one-to-one with the corresponding initial contamination-associated images and stored in the drone's storage module. For example, an initial contamination-associated image taken in the central area of ​​a photovoltaic array corresponds to a flight altitude of 6.2 meters, a pitch angle of 5 degrees, a roll angle of 0.3 degrees, and a heading angle of 90 degrees. These data together constitute the drone's flight parameters for that image.

[0034] Step S116: Collect real-time power output information of the corresponding area through the power generation monitoring components of each area of ​​the photovoltaic array, and use the real-time power output information as the power generation efficiency correlation data of the photovoltaic array.

[0035] Each area of ​​the photovoltaic array is equipped with power generation monitoring components, including current sensors and voltage sensors, which measure the real-time current and voltage of that area, respectively. The sensors transmit the measured current and voltage data to a data acquisition unit, which processes this data to calculate the real-time power of that area, i.e., the real-time power generation output information.

[0036] For example, a certain area of ​​a photovoltaic array consists of 20 photovoltaic panels. The power generation monitoring component measures the real-time voltage of this area as 380 volts and the real-time current as 15 amps, and calculates the real-time power as 5700 watts. This data is the real-time power generation output information of this area, and it is stored as the power generation efficiency related data of this area.

[0037] Step S117: Integrate the pollution visual information, initial pollution-related images, UAV flight parameters and power generation efficiency related data corresponding to each region to form a set of pollution-related images and a correspondence between UAV flight parameters and power generation efficiency related data.

[0038] By using the image capture time and location information, the pollution visual information of each area, the initial pollution-related images, the drone flight parameters, and the power generation efficiency data are matched. For example, if an initial pollution-related image was taken at 10:00 AM, corresponding to area A of the photovoltaic array, then the pollution visual information extracted from that image, the flight altitude of 6.1 meters at the time of capture, the flight angles (pitch angle 4.8 degrees, roll angle 0.2 degrees, heading angle 90 degrees), and the real-time power generation output information of 5600 watts in area A at that time are associated and stored.

[0039] Through the above correlation and integration, a complete dataset is formed, in which each image in the pollution-related image set can find corresponding drone flight parameters and power generation efficiency correlation data.

[0040] Step S120: Based on the pollution-related image set, UAV flight parameters and power generation efficiency related data, perform pollution area localization processing to generate spatial positioning information of the pollution area in the photovoltaic array. The spatial positioning information includes the planar coordinates of the pollution area in the photovoltaic array and a description of the area range.

[0041] After obtaining the integrated pollution-related image set, UAV flight parameters, and power generation efficiency correlation data, the pollution area localization process begins. First, each pollution-related image is analyzed, and the photovoltaic array area covered by the image is determined by combining the corresponding UAV flight parameters. Then, based on the power generation efficiency correlation data, it is determined whether the power generation efficiency in that area is abnormal; areas with low power generation efficiency are typically likely to be polluted.

[0042] By comparing the visual information of contaminated areas in the image with that of normal areas, potential contaminated regions are identified. Then, based on the positional information from the UAV's flight parameters, the contaminated regions in the image are mapped onto the actual plane of the photovoltaic array, determining their planar coordinates. Simultaneously, based on the size and shape of the contaminated regions in the image, their extent is described, ultimately generating spatial positioning information containing both planar coordinates and an extent description.

[0043] Step S121: Extract the pollution visual information of each pollution-related image from the pollution-related image set, and convert the pollution visual information into a pollution feature vector. The pollution feature vector is obtained by converting color difference features, texture change features and morphological contour features into numerical form.

[0044] For each image in the pollution-associated image set, image processing algorithms are used to extract the pollution visual information, namely color difference features, texture variation features, and morphological contour features. Then, feature quantization methods are used to convert these features into numerical forms.

[0045] For example, color difference features can be quantified by calculating the color component difference between contaminated and normal regions; texture variation features can be obtained by analyzing the gray-level co-occurrence matrix of pixels in the image; and morphological contour features can be quantified by calculating parameters such as the perimeter, area, and bounding rectangle of the contaminated region. These quantified values ​​are then arranged in a predetermined order to form a multi-dimensional contamination feature vector, with each dimension corresponding to the value of a single feature.

[0046] Step S1211: Perform color space conversion on each pollution-related image, converting the pollution-related image from RGB color space to HSV color space, and separating the hue channel, saturation channel and brightness channel of the pollution-related image.

[0047] For each pollution-related image, a color space conversion algorithm is used for processing. The RGB color space is based on three color channels: red, green, and blue, while the HSV color space is based on three parameters: hue, saturation, and lightness, which is more in line with human eye perception of color.

[0048] During the conversion process, the corresponding hue, saturation, and brightness values ​​are calculated based on the values ​​of the three RGB channels. For example, for a pixel in the image, its RGB values ​​are (R1, G1, B1), and the corresponding HSV value (H1, S1, V1) is calculated using the conversion formula. After the entire image conversion is completed, the hue, saturation, and brightness are separated as independent channels, resulting in three single-channel images: a hue channel image, a saturation channel image, and a brightness channel image.

[0049] Step S1212: Extract the feature differences between the polluted area and the normal photovoltaic surface area in the hue channel, saturation channel and brightness channel respectively, and convert the feature differences into numerical form to obtain a numerical representation of the color difference features.

[0050] In the hue channel image, compare the hue values ​​of the polluted area and the normal area, and calculate the average difference and the maximum difference between them; in the saturation channel image, calculate the difference in the average saturation value between the polluted area and the normal area, as well as the difference in the saturation distribution range; in the brightness channel image, similarly calculate the difference in the average brightness value and the difference in the magnitude of brightness variation between the two.

[0051] The above differences are represented by numerical calculations. For example, the average difference in hue is ΔH, the average difference in saturation is ΔS, and the average difference in brightness is ΔV. These values ​​together constitute the numerical representation of color difference characteristics.

[0052] Step S1213: Extract texture features from the contaminated image. By analyzing the pixel distribution pattern of the contaminated area in the contaminated image, obtain parameters that reflect texture changes, which are then used as numerical results of the texture change features.

[0053] A gray-level co-occurrence matrix (GLCM) method is used to extract texture features from contaminated images. First, the color image is converted to a grayscale image, and then the contaminated region is selected as the region of interest (ROI) within the grayscale image. The GLCM of this region is calculated at different directions and distances, and parameters such as contrast, energy, entropy, and correlation are extracted from the matrix.

[0054] Contrast reflects the degree of difference in gray levels in an image, energy represents the uniformity of gray level distribution in an image, entropy reflects the richness of image information, and correlation represents the degree of association between pixels. These parameters can well reflect the texture changes in polluted areas and can be used as numerical results of texture change features.

[0055] Step S1214: Extract contour features from the pollution-associated image, identify the contour lines of the pollution area in the pollution-associated image, and extract the feature parameters of the contour lines as numerical data of the morphological contour features.

[0056] Edge detection algorithms are used to process contaminated images and identify the contour lines of contaminated areas. These algorithms calculate the rate of change of pixel grayscale values ​​in the image to find locations where grayscale values ​​change drastically; these locations constitute the contours of the contaminated areas.

[0057] Feature parameters of the contour lines are extracted, including the perimeter, the area enclosed, the concavity and convexity of the contour, and the equivalent diameter. The perimeter is the total length of the contour lines, the area is the size of the region enclosed by the contour, the concavity and convexity reflects the degree of irregularity of the contour, and the equivalent diameter is the diameter of a circle with the same area as the contour. These parameters together constitute the numerical data of the morphological contour features.

[0058] Step S1215: Arrange the numerical representation of color difference features, the numerical results of texture change features, and the numerical data of morphological contour features in a fixed order to form a pollution feature vector with fixed dimensions.

[0059] Numerical representations of color difference features (such as ΔH, ΔS, ΔV, etc.), numerical results of texture variation features (such as contrast, energy, entropy, correlation, etc.), and numerical data of morphological contour features (such as perimeter, area, concavity and convexity, equivalent diameter, etc.) are arranged in a preset fixed order.

[0060] For example, first arrange the values ​​of color difference features, then arrange the values ​​of texture variation features, and finally arrange the values ​​of shape and contour features to form an array containing multiple values. This array is the pollution feature vector with fixed dimensions. This pollution feature vector can comprehensively describe the characteristics of the polluted area.

[0061] Step S122: Retrieve the UAV flight parameters corresponding to each pollution-related image, and calculate the field of view parameters when the pollution-related image is acquired based on the UAV flight altitude and UAV flight angle. The field of view parameters reflect the actual size of the area covered by the photovoltaic array by the pollution-related image.

[0062] From the stored correlation data, retrieve the drone flight parameters corresponding to each pollution-related image, including flight altitude and flight angle. Flight altitude is the vertical distance of the drone relative to the photovoltaic array surface, and flight angle includes the camera's downward angle, etc.

[0063] Based on the flight altitude and camera focal length, the field of view angles of the image in the horizontal and vertical directions can be calculated. Combined with the flight angle, the projection range of the image on the photovoltaic array plane can be further calculated. Through these calculations, field of view parameters, such as the length and width of the field of view, are obtained. These parameters can accurately reflect the actual size of the photovoltaic array area covered by the pollution-related image.

[0064] Step S123: Extract the power generation efficiency correlation data corresponding to each pollution-related image, analyze the difference between the real-time power generation output information and the normal power generation output information in the power generation efficiency correlation data, and obtain the power generation efficiency difference data.

[0065] Real-time power generation output information is extracted from the power generation efficiency correlation data corresponding to each pollution-related image. At the same time, normal power generation output information of the area under clean conditions is obtained. Normal power generation output information can be obtained from historical data or power generation data of similar clean areas.

[0066] Calculate the difference between real-time power generation output information and normal power generation output information, and the ratio of this difference to the normal power generation output information. Use these results as power generation efficiency difference data. For example, if the real-time power generation output of a certain area is 4000 watts and the normal power generation output is 5000 watts, then the difference is -1000 watts and the ratio is -0.2. These data are the power generation efficiency difference data for that area.

[0067] Step S124: Establish the correspondence between the pixel coordinates of the pollution-associated image and the actual planar coordinates of the photovoltaic array. The correspondence is obtained by calculating the correlation between the field of view parameter and the pixel size of the pollution-associated image.

[0068] Given the pixel dimensions of the contaminated image (i.e., the number of pixels contained in the image's length and width) and the actual region length and width in the field of view parameters, the pixel resolution is obtained by calculating the length represented by each pixel on the actual plane (i.e., the actual region length divided by the number of pixels in the image length).

[0069] Then, a pixel coordinate system is established with the top left corner of the image as the origin, and a real-plane coordinate system is established with a fixed point of the photovoltaic array as the origin. Based on the image's shooting position and angle, a coordinate transformation algorithm is used to convert each pixel coordinate in the pixel coordinate system into coordinates in the real-plane coordinate system, thereby establishing the correspondence between the two.

[0070] Step S125: Based on the pollution feature vector and the power generation efficiency difference data, determine the set of pollution-related image pixels belonging to the pollution area in each pollution-related image. Transform the set of pollution-related image pixels into the actual planar coordinates of the photovoltaic array through the correspondence to obtain the initial planar coordinates of the pollution area.

[0071] The pollution feature vector is input into a trained pollution identification model, which outputs the probability that each pixel in the image belongs to a polluted area. Combined with power generation efficiency difference data, when the power generation efficiency difference data for a certain area exceeds a preset threshold, that area is considered potentially polluted, and the probability weight of pixels in that area belonging to a polluted area is increased.

[0072] By setting a probability threshold, pixels with probabilities exceeding the threshold are identified as pixels in the contaminated area, forming a set of contaminated image pixels. Then, based on the previously established correspondence between pixel coordinates and actual planar coordinates, the above pixel set is converted into the actual planar coordinates of the photovoltaic array, obtaining the initial planar coordinates of the contaminated area.

[0073] Step S126: Perform region integration processing on the initial plane coordinates, integrate adjacent initial plane coordinate sets into a complete contaminated region, determine the boundary coordinates of each contaminated region, and form spatial positioning information containing the plane coordinates of the contaminated region in the photovoltaic array and a description of the region's extent.

[0074] Cluster analysis is performed on the obtained initial planar coordinates to calculate the distance between each coordinate point. When the distance is less than a preset neighbor distance threshold, these coordinate points are considered to belong to the same contaminated area. Adjacent initial planar coordinate sets are then merged to form a complete contaminated area.

[0075] Next, the outermost coordinates of each complete contaminated area are identified; these points constitute the boundary of the contaminated area, and the boundary coordinates are determined. The extent of the contaminated area is described based on these boundary coordinates; for example, if the contaminated area has an irregular shape, its east-west span is from x1 to x2, and its north-south span is from y1 to y2, etc. The boundary coordinates and the extent description are then integrated to form spatial positioning information.

[0076] Step S130: Combine the spatial positioning information with the pollution diffusion trend data of the polluted area and the power generation priority data of the photovoltaic array to plan the pollution cleaning path of the UAV. The pollution cleaning path includes the flight trajectory of the UAV between different polluted areas and the order of cleaning operations.

[0077] In this large-scale grid-connected photovoltaic power station, after obtaining spatial positioning information, it is necessary to combine it with other data to plan the cleaning path. The spatial positioning information clarifies the location and extent of each contaminated area. For example, the planar coordinate range of a certain contaminated area is from x3 to x4 in the east-west direction and from y3 to y4 in the north-south direction, forming an irregular polygon.

[0078] Pollution diffusion trend data is derived by analyzing historical pollution data and current environmental conditions of the power plant. For example, dust pollution in a certain area tends to spread southeastward under the influence of wind. Power generation priority data is determined based on the importance of each area in the photovoltaic array. For example, areas located in the center of the array and with a higher power generation share have higher priority.

[0079] Based on this data, we first determine which contaminated areas need to be cleaned first, then plan the flight paths of drones between these areas to avoid duplicate flights and omissions, and at the same time determine the cleaning sequence of each area to form a complete contamination cleaning path.

[0080] Step S131: Analyze the planar coordinates and area descriptions of each polluted area in the spatial positioning information, and extract the area data and center coordinates of each polluted area.

[0081] The spatial positioning information is analyzed to extract the coordinate values ​​of boundary points from the planar coordinates of each polluted area. For example, the boundary coordinates of a polluted area might be (x5, y5), (x6, y6), (x7, y7), etc. The area of ​​this area is calculated using these coordinate points. When calculating the area, a polygon area calculation formula is used. The boundary points are connected sequentially to form a polygon, and then the area data is calculated based on the coordinates of each point.

[0082] The center coordinates are calculated by taking the average of the coordinates of all boundary points in the polluted area. That is, the x-value of the center coordinate is the average of the x-coordinates of all boundary points, and the y-value is the average of the y-coordinates of all boundary points, thus obtaining the center coordinates of each polluted area, such as (xm1, ym1), (xm2, ym2), etc.

[0083] Step S132: Obtain environmental data of the area where the photovoltaic array is located, analyze the pollution diffusion trend of each polluted area based on the environmental data, and generate pollution diffusion trend data of the polluted area. The pollution diffusion trend data reflects the future diffusion direction of the polluted area.

[0084] Environmental monitoring equipment installed within the photovoltaic power station acquires environmental data for the area, including wind speed, wind direction, temperature, and humidity. For example, it may detect a current wind speed of a certain value, a wind direction of northeast, a temperature of a certain degree Celsius, and a humidity percentage.

[0085] For each polluted area, an analysis is conducted based on its pollution type (e.g., dust, bird droppings) and environmental data. Dust pollution, under the influence of northeasterly winds, may spread southwestward; while solid pollutants such as bird droppings have poor diffusion, but may spread to lower elevations under the washout of rainwater. Based on these analyses, the future diffusion direction and possible diffusion rate of each polluted area are determined, generating pollution diffusion trend data, such as "a polluted area will spread southwestward in the next 48 hours, with a slow diffusion rate," etc.

[0086] Step S133: Analyze the power generation efficiency correlation data of each region of the photovoltaic array, and determine the power generation priority data of each region in combination with the overall power generation demand of the photovoltaic array. The power generation priority data reflects the degree of influence of the power generation output of each region on the overall power generation efficiency.

[0087] Analyze the correlation data of power generation efficiency in different regions, i.e., real-time power generation output information, and compare the power generation contribution of each region. For example, the real-time power generation output of one region accounts for 10% of the total output of the entire photovoltaic array, while that of another region accounts for only 3%.

[0088] Meanwhile, considering the overall power generation needs of the photovoltaic array, such as the need for certain key areas to maintain high-efficiency power generation during peak electricity consumption periods, the regions are prioritized based on these factors. Regions that contribute significantly to power generation and have a significant impact on overall power generation efficiency are assigned higher priority values, while those that contribute less are assigned lower values, forming power generation priority data. For example, region A has a priority value of 8, region B has a priority value of 3, and so on.

[0089] Step S134: Based on the center coordinates of each polluted area, pollution diffusion trend data, and power generation priority data, all polluted areas are grouped together. Polluted areas with diffusion direction angles less than a preset angle threshold, Euclidean distances between center points less than a preset distance threshold, and the same power generation priority level are divided into the same cleaning group.

[0090] Calculate the angle between the diffusion directions of each polluted area. For example, the diffusion direction of area C is southwest, and the diffusion direction of area D is southwest-southwest. Calculate the angle between them. Simultaneously, calculate the Euclidean distance between the center coordinates of each polluted area, i.e., the straight-line distance between the two points.

[0091] The preset angle threshold is 30 degrees, and the preset distance threshold is 5 meters. When the angle between the diffusion directions of two polluted areas is less than 30 degrees, the Euclidean distance between their center points is less than 5 meters, and they have the same power generation priority level (e.g., both are high priority), they are assigned to the same cleaning group. For example, areas E and F meet the above conditions and are assigned to cleaning group 1; areas G and H meet the conditions and are assigned to cleaning group 2, etc.

[0092] Step S135: Determine the priority of cleaning operations based on the pollution diffusion trend data and power generation priority data of each cleaning group. Cleaning groups with a diffusion rate exceeding the preset rate threshold and a higher power generation priority value in the pollution diffusion trend data will be given priority in cleaning operations.

[0093] The diffusion rate in the pollution diffusion trend data of each cleaning group was analyzed, with a preset rate threshold. Cleaning groups with diffusion rates exceeding this threshold indicate rapid pollution diffusion and require timely intervention. Simultaneously, the power generation priority values ​​of each cleaning group were compared; higher values ​​indicate greater importance for power generation.

[0094] Taking both factors into account, cleaning groups with diffusion rates exceeding a preset threshold and higher power generation priority values ​​are determined to have higher cleaning operation priority. For example, cleaning group 3 has a diffusion rate exceeding the threshold and a power generation priority value of 9; cleaning group 4 has a diffusion rate not exceeding the threshold and a power generation priority value of 8. Therefore, cleaning group 3 has a higher cleaning operation priority than cleaning group 4.

[0095] Step S136: For each contaminated area in the cleaning group, determine the cleaning order within the group based on the area data of the contaminated area and the power generation priority data. Contaminated areas with a contaminated area exceeding the preset area threshold and a higher power generation priority value are cleaned first.

[0096] Within each cleaning group, a preset area threshold is set. By comparing the area data of each contaminated area within the group with the power generation priority data, if the area of ​​a contaminated area exceeds the preset area threshold and its power generation priority value is higher within the group, that area will be cleaned earlier in the group.

[0097] For example, in cleaning group 1 there are areas E and F. The area of ​​area E exceeds the preset area threshold, and the power generation priority value is 8; the area of ​​area F does not exceed the threshold, and the power generation priority value is 7. Then the cleaning order in the group is area E before area F.

[0098] Step S137: Based on the center coordinates of each cleaning group, the priority of cleaning operations, and the cleaning sequence of contaminated areas within the group, plan the flight trajectory of the drone between different cleaning groups, and at the same time plan the contamination cleaning path of the drone in a single contaminated area, forming a contamination cleaning path that includes the flight trajectory of the drone between different contaminated areas and the cleaning sequence of operations.

[0099] Based on the center coordinates and cleaning priority of each cleaning group, the flight order of the drones between different cleaning groups is determined. For example, cleaning group 3 has the highest priority, followed by cleaning group 1, then cleaning group 2. The drone first flies from its current position to the vicinity of the center coordinates of cleaning group 3, completes the cleaning of that group, and then flies to cleaning group 1, and so on.

[0100] When planning the flight path, it is necessary to avoid obstacles in the photovoltaic array, such as brackets and junction boxes. For a single contaminated area, a cleaning path is planned within the area based on its description and the cleaning sequence within the group, such as using a serpentine route to cover the entire contaminated area. The flight path and cleaning sequence described above are then integrated to form a complete contamination cleaning path.

[0101] Step S1371: Taking the center coordinates of the current cleaning group as the starting point and the center coordinates of the next priority cleaning group as the ending point, obtain the photovoltaic array structure data between the starting point and the ending point. The photovoltaic array structure data includes the arrangement of photovoltaic modules and the distribution of module connection nodes.

[0102] The center coordinates (xstar, ystar) of the current cleaning group and the center coordinates (xend, yend) of the next priority cleaning group are determined, and these are used as the start and end points of the flight trajectory. The photovoltaic array structure data between the start and end points is obtained by retrieving the design drawings and structural database of the photovoltaic power station.

[0103] These data record in detail whether the photovoltaic modules within this range are arranged in a row-column or other manner, as well as the specific locations of the connection nodes between the modules, such as the planar coordinates (xnode1, ynode1) and (xnode2, ynode2) of each connection node.

[0104] Step S1372: Plan the flight trajectory of the UAV between the two cleaning groups based on the photovoltaic array structure data, so that the flight trajectory avoids the protruding parts of the photovoltaic modules and the module connection nodes.

[0105] By analyzing the acquired photovoltaic array structure data, protruding parts of the photovoltaic modules, such as the location of junction boxes, and the distribution of module connection nodes, are identified. When planning the flight trajectory from the starting point to the end point, a path planning algorithm is used to calculate a path that avoids these protruding parts and connection nodes.

[0106] For example, if the drone were to fly in a straight line and pass through a connecting node, it could adjust its trajectory to bypass the node from the side, ensuring that the drone does not collide with these structures during flight while maintaining a smooth flight path and the shortest possible distance.

[0107] Step S1373: For a single contaminated area, determine the cleaning boundary based on the area description and contamination distribution density data of the contaminated area. Starting from the center of the contaminated area, plan the contamination cleaning path of the UAV within the contaminated area in the direction of expansion from the area with more concentrated contamination to the area with less dispersed contamination.

[0108] Obtain a description of the area of ​​a single polluted region and clarify its boundaries. At the same time, obtain the pollution distribution density data of the region by analyzing the pollution-related images, such as the dense pollution particles in some sub-regions and the relatively sparse pollution particles in some sub-regions.

[0109] Based on this data, the cleaning boundary is determined, which is the maximum area that needs to be cleaned. Starting from the center of the contaminated area, the area with a higher concentration of contamination is cleaned first, and then the cleaning is gradually expanded to areas with a lower concentration. A spiral or step-by-step cleaning path is planned to ensure that high-density contaminated areas are thoroughly cleaned.

[0110] Step S1374: Set path markers on the planned pollution cleaning path. The path markers are used for the UAV to adjust its position during flight so that the flight trajectory is consistent with the pollution cleaning path.

[0111] Along the planned contamination cleaning path, a path marker is placed at predetermined intervals. The coordinates of the markers are precisely calculated based on the path trajectory. For example, a marker is placed every 2 meters on straight sections, and the density of markers is appropriately increased at bends.

[0112] The coordinates of these markers are stored in the drone's navigation system. During flight, the drone uses real-time positioning to continuously compare its coordinates with those of the markers, adjusting its flight position and attitude to ensure that the actual flight trajectory remains consistent with the planned pollution cleaning path.

[0113] Step S1375: Obtain the flight speed data of the drone and the cleaning speed data of the cleaning device, and combine them with the length data of the pollution cleaning path to calculate the time required for the drone to complete each segment of flight and each area cleaning.

[0114] The drone's flight speed data is provided by its flight control system, such as a certain cruise speed; the cleaning speed data of the cleaning device is obtained through testing, such as a certain area that can be cleaned per unit time. The length data of the contamination cleaning path is obtained by calculating the sum of the distances of each segment of the trajectory.

[0115] When calculating the time required for each flight segment, divide the length of that segment by the flight speed; when calculating the time required for cleaning each area, divide the contaminated area of ​​that area by the cleaning speed of the cleaning device. Apply these time data to the respective flight segments and cleaning processes for each area.

[0116] Step S1376: Integrate the time data with the trajectory data of the contamination cleaning path to form a complete contamination cleaning path, wherein the complete contamination cleaning path includes the correspondence between the time data and the trajectory data.

[0117] The calculated time data is correlated with the corresponding trajectory data. For example, on a certain trajectory, the flight time from marker point A to marker point B is t1, and the time to clean a certain area is t2, etc.

[0118] Through the above integration, the resulting complete pollution cleaning path not only includes the drone's flight path and cleaning sequence, but also clarifies the drone's location and operations at each point in time.

[0119] Step S140: Control the cleaning device carried by the UAV to perform pollution cleaning operation according to the pollution cleaning path. During the process, real-time pollution change images and real-time power generation efficiency change data of the cleaning area are collected simultaneously to generate pollution cleaning operation results of the UAV. The pollution cleaning operation results include cleaning completion status information of each pollution area.

[0120] After receiving complete pollution cleaning path data, the drone's flight control system controls the drone to fly according to the trajectory and time requirements in the path. When it reaches the designated pollution area, it activates the onboard cleaning device, such as a high-pressure water spray device or a rotating brush device, to clean the pollution area.

[0121] During the cleaning process, the image acquisition equipment on the drone continuously captures images of the cleaning area, obtaining real-time images of pollution changes and recording the process of pollution from its presence to its gradual disappearance. Simultaneously, the photovoltaic array's power generation monitoring components continue to collect real-time power output information for the area, generating real-time power generation efficiency change data.

[0122] After all contaminated areas have been cleaned, the cleaning time and post-cleaning status of each area are integrated to generate a contamination cleaning operation result, which clearly indicates whether each area has been cleaned.

[0123] Step S141: The contamination cleaning path is sent to the flight control system of the UAV. The flight control system adjusts the flight attitude and speed of the UAV according to the trajectory data in the contamination cleaning path, and controls the UAV to fly along the contamination cleaning path.

[0124] The contamination cleaning path data is transmitted to the drone's flight control system via a wireless communication module. The flight control system analyzes the trajectory data to obtain the coordinates and corresponding time points of each path marker.

[0125] During flight, the flight control system adjusts the drone's pitch angle, roll angle, and other flight attitudes based on the deviation between the current position and the target marker point, and adjusts the flight speed according to time requirements to ensure that the drone flies accurately along the pollution cleaning path.

[0126] Step S142: When the drone arrives at the designated contaminated area, it sends a start signal to the cleaning device using the flight control system. After receiving the start signal, the cleaning device starts to operate. During operation, the cleaning device adjusts the cleaning pressure and the amount of cleaning medium sprayed according to the contamination level data of the contaminated area.

[0127] When the drone's positioning system displays that it has reached the center coordinates of the designated contaminated area, the flight control system generates a start signal containing relevant parameters of the contaminated area, such as the area's extent, and sends it to the cleaning device.

[0128] After receiving the signal, the cleaning device starts its internal power system and begins operation. At the same time, based on pre-acquired data on the degree of contamination in the area, such as heavy or light contamination, it adjusts the cleaning pressure and the spray volume of the cleaning medium (such as water or detergent) accordingly.

[0129] Step S1421: When the UAV flies to the center coordinates of the designated contaminated area, the flight control system transmits a start signal to the cleaning device. Upon receiving the start signal, the cleaning device activates its internal contamination level detection component. The contamination level detection component detects the designated contaminated area and generates contamination level data for the area. The start signal includes the area range data of the contaminated area. The cleaning device internally stores the correspondence data between contamination level and cleaning parameters. The correspondence data includes cleaning pressure data and cleaning medium spraying volume data corresponding to different contamination levels.

[0130] When the drone flies precisely to the center coordinates (xm3, ym3) of the designated contaminated area, the flight control system sends a start signal to the cleaning device, which clearly defines the boundary coordinate range of the contaminated area.

[0131] After receiving a signal, the cleaning device activates its internal contamination detection components, such as optical sensors, to scan and detect the designated area. By analyzing parameters such as the intensity of reflected light, it determines the severity of the contamination and generates contamination level data, such as "heavy contamination," "moderate contamination," or "light contamination." The cleaning device's storage module pre-stores corresponding relationship data; for example, heavy contamination corresponds to higher cleaning pressure and a larger spray volume, while light contamination corresponds to lower pressure and a smaller spray volume.

[0132] Step S1422: Based on the pollution level data of the polluted area, query the corresponding relationship data to determine the cleaning pressure data and cleaning medium spraying volume data corresponding to the current polluted area.

[0133] The control module of the cleaning device compares and queries the detected pollution level data with the stored corresponding data. For example, if a certain area is detected to be moderately polluted, the corresponding cleaning pressure data is determined to be P1 and the cleaning medium spraying volume data is determined to be Q1 after querying.

[0134] By checking the above parameters, you can ensure that the cleaning parameters match the degree of contamination and avoid insufficient cleaning or waste of resources.

[0135] Step S1423: Adjust the pressure regulating component inside the cleaning device to achieve the determined cleaning pressure data, and adjust the spraying component of the cleaning device to achieve the determined cleaning medium spraying volume data.

[0136] The pressure regulating component of the cleaning device, such as a pressure regulating valve, adjusts the valve opening degree according to the determined cleaning pressure data P1, so that the pressure of the sprayed medium reaches P1. The spraying component, such as a flow control valve, controls the volume of medium flowing out per unit time to be Q1 according to the cleaning medium spraying volume data Q1.

[0137] During the adjustment process, sensors inside the device monitor pressure and flow in real time to ensure that preset data are achieved.

[0138] Step S1424: After the cleaning device completes parameter adjustment, it begins to clean the contaminated area. During the operation, the cleaning pressure data and the amount of cleaning medium sprayed are dynamically adjusted based on the real-time contamination change image.

[0139] Once the parameters are adjusted, the cleaning device begins cleaning the contaminated area; for example, the high-pressure water sprayer starts spraying water, and the rotating brushes begin to turn. Simultaneously, the drone transmits real-time images of the changing contamination to the cleaning device's control module.

[0140] The control module analyzes the residual pollution in the image. If it finds that there is still significant pollution in a certain area, it appropriately increases the cleaning pressure or spray volume for that area. If most of the pollution has been removed, it appropriately reduces the pressure and spray volume to achieve dynamic fine-tuning.

[0141] Step S143: During the operation of the cleaning device, the image acquisition device carried by the UAV is used to synchronously acquire real-time pollution change images of the cleaning area to obtain real-time pollution change images, which contain visual change information of the polluted area during the cleaning process.

[0142] The high-definition camera on the drone takes pictures of the cleaning area at fixed time intervals when the cleaning device is running, such as taking one picture per second. The above pictures continuously record the visual changes of the polluted area from obvious stains to gradual cleanliness, such as the process of dust being washed away by water and bird droppings being removed by a brush.

[0143] Real-time images of pollution changes are temporarily stored in the drone's storage module and transmitted to the ground control center at an appropriate time.

[0144] Step S144: Simultaneously, real-time power generation output information is collected by the power generation monitoring component of the photovoltaic array in the designated polluted area to obtain real-time power generation efficiency change data. The real-time power generation efficiency change data includes the changes in real-time power generation output in the designated polluted area during the cleaning process.

[0145] During the cleaning process, current and voltage sensors in the designated contaminated area work continuously, measuring the current and voltage values ​​at set intervals to calculate the real-time power. The real-time power output information is continuously updated to form real-time power generation efficiency change data, reflecting the change in power generation efficiency from low to high as the contamination is removed.

[0146] The data acquisition device uploads the above data in real time, corresponding to the timestamps of the real-time pollution change images.

[0147] Step S145: Continuously collect real-time pollution change images and real-time power generation efficiency change data until the cleaning device completes the pollution cleaning operation in the polluted area, and then stop collecting real-time pollution change images and real-time power generation efficiency change data.

[0148] The cleaning device thoroughly cleans the contaminated area according to the planned cleaning path. Throughout the process, image acquisition and power generation data acquisition are continuously performed. Once the cleaning device has completed the cleaning operation for the area, that is, after covering the entire contaminated area according to the path and reaching the preset cleaning time, the flight control system sends a stop acquisition signal to the image acquisition equipment and power generation monitoring components.

[0149] Upon receiving the signal, the image acquisition device stops capturing images, and the power generation monitoring component returns to its normal data acquisition frequency, ceasing high-density real-time acquisition.

[0150] Step S146: Record the start time of cleaning, the end time of cleaning, the acquisition record of real-time pollution change images and the acquisition record of real-time power generation efficiency change data for each polluted area, and integrate them to generate a pollution cleaning operation result containing the cleaning completion status information of each polluted area.

[0151] For each contaminated area, the time at which the cleaning device starts is recorded as the start time of cleaning; the time at which the cleaning device stops cleaning the area is recorded as the end time of cleaning. The real-time pollution change image acquisition record includes the capture time, storage path, and corresponding image thumbnail information for each image, facilitating quick retrieval and viewing later. The real-time power generation efficiency change data acquisition record includes the time point of each data acquisition, the corresponding current value, voltage value, and the calculated power value.

[0152] The above information is categorized and organized according to the polluted area. For example, the cleaning of area A starts at time T1 and ends at time T2, with N real-time pollution change images and M sets of real-time power generation efficiency change data collected. Simultaneously, based on whether the cleaning covered all polluted areas and the initial visual assessment after cleaning, the cleaning completion status of the area is determined, such as "cleaning completed" or "cleaning incomplete." Finally, all this information from all areas is integrated to form the pollution cleaning operation result.

[0153] Step S150: Based on the results of the pollution cleaning operation, real-time pollution change images, and real-time power generation efficiency change data, perform cleaning effect evaluation processing to generate pollution cleaning evaluation information for the photovoltaic array. The pollution cleaning evaluation information includes supplementary cleaning suggestions for areas that have not been cleaned.

[0154] After obtaining the results of the pollution cleaning operation, the cleaning effect of each polluted area is evaluated by combining real-time pollution change images and real-time power generation efficiency change data. By comparing real-time pollution change images before and after cleaning, it is observed whether the pollution has completely disappeared or still has residues; by analyzing real-time power generation efficiency change data, it is seen whether the power generation efficiency of the area has recovered to the normal level or reached the expected recovery degree.

[0155] For areas that were not properly cleaned, supplementary cleaning recommendations are developed based on the residual contamination and the recovery of power generation efficiency. The assessment results and supplementary cleaning recommendations for all areas are then integrated to generate a contamination cleaning assessment report for the photovoltaic array.

[0156] Step S151: Analyze the cleaning completion status information of each polluted area in the pollution cleaning operation results, and extract the real-time pollution change image and real-time power generation efficiency change data corresponding to the cleaning end time of each polluted area.

[0157] The cleaning completion status information of each contaminated area is extracted from the pollution cleaning operation results. For each area, regardless of whether its status is "cleaned completed" or "cleaned incomplete", the cleaning end time is found. Based on the cleaning end time, the corresponding image is found in the real-time pollution change image acquisition record, that is, the image(s) taken closest to the cleaning end time; several sets of data before and after the cleaning end time are found in the real-time power generation efficiency change data acquisition record, usually selecting the last set of data before the end time and the first set of data after the end time to reflect the power generation efficiency immediately after cleaning.

[0158] For example, if the cleaning of region B ends at time T3, find the image taken at time T3±Δt in the real-time pollution change image, and find the data sets collected at time T3-Δt1 and T3+Δt2 in the real-time power generation efficiency change data.

[0159] Step S152: Extract residual pollution features from the real-time pollution change image corresponding to the end of cleaning. The residual pollution features include color features, texture features, and morphological features of the residual pollution area.

[0160] The real-time contamination change image corresponding to the end of the cleaning process is processed. First, an image segmentation algorithm is used to separate areas where residual contamination may exist from the background. For the separated residual contamination areas, their color characteristics are analyzed, namely the hue, saturation, and lightness values ​​of the area in the HSV color space, and these values ​​are compared with those of the normal photovoltaic surface area to determine the color differences.

[0161] Next, the texture features of the residual contamination area are analyzed. By calculating the gray-level co-occurrence matrix of the area, parameters such as contrast, energy, and entropy are extracted to describe its texture roughness, uniformity, and other characteristics. Then, the morphological features of the residual contamination area are identified, including the shape (e.g., circular, irregular), perimeter, area, and aspect ratio of the circumscribed rectangle. The above color features, texture features, and morphological features are integrated as the residual contamination features of the area.

[0162] Step S153: Analyze the real-time power generation efficiency change data corresponding to the end of the cleaning, compare the real-time power generation output information of the polluted area before and after cleaning, and obtain the power generation efficiency recovery data.

[0163] Acquire real-time power generation efficiency change data at the end of the cleaning process, and extract the power value of the contaminated area after the cleaning. Simultaneously, find the power value of the area before the start of the cleaning process from the real-time power generation efficiency change data before cleaning, and use this as the real-time power output information before cleaning.

[0164] The power difference between the power value after cleaning and the power value before cleaning, as well as the ratio of this difference to the power value before cleaning, together constitute the power generation efficiency recovery data. For example, if the power value before cleaning in region C is Pbefore and the power value after cleaning is Pafter, then the power generation efficiency recovery data includes ΔP = Pafter - Pbefore and ΔP / Pbefore.

[0165] Step S154: Compare the pollution residue characteristics with the characteristic data of the normal area of ​​the photovoltaic array, and combine them with the power generation efficiency recovery data to obtain the pollution residue data of each polluted area.

[0166] The characteristic data of the normal area of ​​the photovoltaic array refers to the color, texture, and morphological characteristics of the uncontaminated or completely cleaned areas. This data can be obtained by pre-collecting and analyzing images of the normal area. The residual pollution characteristics of each contaminated area are compared item by item with the characteristic data of the normal area to calculate the degree of difference between the two, such as the difference in hue value in color characteristics and the difference in contrast in texture characteristics.

[0167] By combining power generation efficiency recovery data, if the differences between the pollution residue characteristics and the normal area characteristics are small, and the power generation efficiency recovery data shows good power recovery, it indicates that the pollution residue in that area is low; conversely, it indicates that the pollution residue is high. Based on these comparison results and analysis, the pollution residue status data for each polluted area is determined, such as "no residue," "slight residue," or "severe residue."

[0168] Step S155: Collect data on the residual pollution in all polluted areas, calculate the number and area percentage of areas without residual pollution after cleaning, calculate the number and area percentage of areas with residual pollution, and collect data on the number and area percentage of areas where power generation efficiency recovery reaches the expected level, thus forming statistical data on the cleaning effect.

[0169] The pollution residue data of all polluted areas are classified and statistically analyzed. First, the number of areas without residual pollution is counted, and then the total area of ​​these areas is calculated as the proportion of the total area of ​​all cleaned polluted areas. That is, the area proportion = (total area of ​​areas without residual pollution / total area of ​​all cleaned polluted areas) × 100%.

[0170] Similarly, the number of areas with residual pollution and their area percentage are statistically analyzed. For power generation efficiency recovery data, an expected recovery level is preset, such as recovering to more than 90% of normal power generation efficiency. The number of areas with power generation efficiency recovery reaching this level and their area percentage are statistically analyzed. The above statistical results are compiled into statistical data on cleaning effect to intuitively reflect the overall cleaning effect.

[0171] Step S156: For areas with residual pollution, generate supplementary cleaning recommendations based on the pollution residue characteristics and power generation efficiency recovery data. The supplementary cleaning recommendations include adjusted cleaning pressure data, cleaning medium spray volume data, and supplementary pollution cleaning path data.

[0172] For areas with residual contamination, a thorough analysis of the contamination characteristics and power generation efficiency recovery data is necessary. If the residual contamination is dark in color and has a rough texture, it indicates potentially stubborn contamination. Given that power generation efficiency recovery has not met expectations, adjustments to the cleaning pressure and cleaning medium spray volume are required. Simultaneously, based on the distribution of residual contamination, supplementary contamination cleaning paths should be planned. The adjusted parameters and paths should be integrated to form supplementary cleaning recommendations.

[0173] For example, step S1561: extract the pollution residue characteristics of the area where residual pollution exists, analyze the distribution pattern of residual pollution, and determine the concentrated area and distribution range of residual pollution.

[0174] From the residual pollution characteristics of areas with residual pollution, information related to distribution, such as area and location coordinates in morphological features, is extracted. Analysis of this information identifies areas with concentrated residual pollution, for example, a corner of region D where residual pollution is densely concentrated. Simultaneously, based on the boundary coordinates of these residual pollution areas, their distribution range is determined, such as x8 to x9 in the east-west direction and y8 to y9 in the north-south direction.

[0175] Step S1562: Analyze the power generation efficiency recovery data of the area with residual pollution, and determine the correspondence between the areas where the power generation efficiency recovery did not meet expectations and the areas with concentrated residual pollution.

[0176] Examine the power generation efficiency recovery data for areas with residual pollution to identify the regions where recovery has fallen short of expectations. For example, a portion of this region may still have power values ​​below the expected level. Compare this region with the distribution range of areas with concentrated residual pollution to determine if they overlap or contain each other. If the region where power generation efficiency recovery falls short of expectations coincides with an area with concentrated residual pollution, it indicates that residual pollution is the primary cause of the unmet expectations for power generation efficiency.

[0177] Step S1563: Based on the concentrated areas and distribution range of residual pollution and their correspondence with areas where power generation efficiency recovery has not met expectations, plan supplementary pollution cleaning path data. The supplementary pollution cleaning path data includes the flight trajectory and cleaning sequence of the UAV in the residual pollution area, prioritizing coverage of concentrated residual pollution areas where power generation efficiency recovery has not met expectations.

[0178] When planning supplementary pollution cleaning paths, the focus is on areas with concentrated residual pollution, and the drone flight paths are designed based on their distribution range. For example, a dense serpentine trajectory is used to cover areas with concentrated residual pollution before expanding to the entire distribution range. Simultaneously, based on the correspondence with areas where power generation efficiency recovery has not met expectations, priority is given to cleaning those areas with concentrated residual pollution where power generation efficiency recovery has not met expectations, ensuring these areas are cleaned first. The flight paths in the supplementary pollution cleaning path data are accurate to specific coordinate points, and the cleaning order clearly defines the order of cleaning for each sub-region.

[0179] Step S1564: Analyze the characteristic intensity of residual contamination, which reflects the stubbornness of residual contamination. Combined with the degree to which the power generation efficiency recovery data did not meet expectations, determine the adjusted cleaning pressure data.

[0180] The intensity of residual pollution can be judged by the depth of its color and the roughness of its texture. The darker the color and the rougher the texture, the higher the intensity of the pollution and the more persistent it is. The degree to which power generation efficiency recovery data falls short of expectations is measured by the ratio of the difference between the actual recovered power and the expected recovered power to the expected recovered power. The larger the ratio, the greater the degree of failure to meet expectations.

[0181] For residual pollution areas with high characteristic intensity and low expected power generation efficiency recovery, the cleaning pressure data needs to be increased; for areas with low characteristic intensity and low expected power generation efficiency recovery, the cleaning pressure data can be appropriately reduced or the original pressure data can be maintained. For example, if area E has high residual pollution characteristic intensity and low expected power generation efficiency recovery, the cleaning pressure data should be adjusted from the original P to P+ΔP.

[0182] Step S1565: Based on the distribution range, characteristic intensity, and power generation efficiency recovery requirements of residual pollution, determine the adjusted cleaning medium spraying volume data so that the cleaning medium can cover the residual pollution area and meet the power generation efficiency recovery requirements.

[0183] The wider the distribution range of residual pollution, the more cleaning media is needed; the higher the characteristic intensity, i.e., the more stubborn the pollution, the more cleaning media may be required to assist in cleaning. At the same time, considering the need for power generation efficiency recovery, for areas with high demand for power generation efficiency recovery, ensure that there is enough cleaning media to thoroughly remove pollution in order to promote the recovery of power generation efficiency.

[0184] Taking all these factors into account, the adjusted cleaning medium spraying volume data is calculated. For example, the residual pollution distribution range in region F is large, the characteristic intensity is moderate, and the demand for power generation efficiency recovery is high. Therefore, the cleaning medium spraying volume data is adjusted from the original Q to Q+ΔQ.

[0185] Step S1566: Integrate the planned supplementary contamination cleaning path data, the determined adjusted cleaning pressure data, and the cleaning medium spraying volume data to form a supplementary cleaning recommendation for areas with residual contamination.

[0186] The flight trajectory, cleaning sequence, adjusted cleaning pressure data, and cleaning medium spray volume data in the supplementary contamination cleaning path data are associated by region. For example, the supplementary cleaning suggestion for region G includes its corresponding supplementary contamination cleaning path, cleaning pressure data P1, and cleaning medium spray volume data Q1.

[0187] Through the above integration, the resulting supplementary cleaning recommendations are complete and well-organized, and can directly guide drones in carrying out supplementary cleaning operations.

[0188] Step S157: Integrate the statistical data of cleaning effect with the supplementary cleaning recommendations to generate contamination cleaning assessment information for the photovoltaic array, which includes supplementary cleaning recommendations for areas that have not been cleaned.

[0189] The statistical data on cleaning effectiveness, such as the number and area percentage of areas without residual pollution, the number and area percentage of areas with residual pollution, and the number and area percentage of areas where power generation efficiency has recovered to the expected level, are summarized together with supplementary cleaning recommendations for all areas with residual pollution.

[0190] This information is arranged according to regional order or assessment importance to form a clearly structured contamination cleaning assessment report for the photovoltaic array. This contamination cleaning assessment report not only allows staff to fully understand the overall effect of the cleaning but also provides specific operational guidance for subsequent supplementary cleaning work.

[0191] Throughout the implementation process, sensitive data potentially related to power plant operation, such as real-time power generation output information from various areas of the photovoltaic array, and equipment data like drone flight parameters, were involved. To protect the privacy and security of this data, encrypted transmission technology was employed. All data underwent encryption during transmission, and only authorized recipients could decrypt and read it. Simultaneously, an access control mechanism was implemented for data storage, assigning different access permissions to different personnel based on their responsibilities to prevent unauthorized access and leakage. Furthermore, data was regularly backed up and audited to ensure its integrity and security.

[0192] Figure 2 The present invention illustrates the hardware structure of a drone intelligent positioning and cleaning system 100 for implementing the above-described drone intelligent positioning and cleaning method for photovoltaic pollution, as provided in an embodiment of the present invention. Figure 2 As shown, the drone intelligent positioning and cleaning system 100 for photovoltaic pollution may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0193] In one possible design, the drone-based intelligent positioning and cleaning system 100 for photovoltaic pollution can be a single server or a group of servers. The server group can be centralized or distributed (e.g., the drone-based intelligent positioning and cleaning system 100 for photovoltaic pollution can be a distributed system). In some embodiments, the drone-based intelligent positioning and cleaning system 100 for photovoltaic pollution can be local or remote. For example, the drone-based intelligent positioning and cleaning system 100 for photovoltaic pollution can access information and / or data stored in machine-readable storage medium 120 via a network. As another example, the drone-based intelligent positioning and cleaning system 100 for photovoltaic pollution can directly connect to machine-readable storage medium 120 to access the stored information and / or data.

[0194] Machine-readable storage medium 120 may store data and / or instructions. In some embodiments, machine-readable storage medium 120 may store data acquired from an external terminal. In some embodiments, machine-readable storage medium 120 may store data and / or instructions used by the drone intelligent positioning and cleaning system 100 for photovoltaic pollution to perform or complete the exemplary methods described in this invention.

[0195] In the specific implementation process, one or more processors 110 execute computer-executable instructions stored in machine-readable storage medium 120, so that processor 110 can execute the UAV intelligent positioning and cleaning method for photovoltaic pollution as described in the above method embodiment. The processor 110, machine-readable storage medium 120 and communication unit 140 are connected through bus 130. The processor 110 can be used to control the sending and receiving actions of communication unit 140.

[0196] The specific implementation process of processor 110 can be found in the various method embodiments executed by the above-mentioned UAV intelligent positioning and cleaning system 100 for photovoltaic pollution. The implementation principle and technical effect are similar, and will not be repeated here.

[0197] Furthermore, this embodiment of the invention also provides a readable storage medium containing computer-executable instructions. When the processor runs the computer-executable instructions, the above-mentioned intelligent positioning and cleaning method for photovoltaic pollution by drones is implemented.

[0198] It should be noted that, in order to simplify the description of this invention and thus aid in the understanding of one or more embodiments, the foregoing description of the embodiments of this invention sometimes combines multiple features into a single embodiment, drawing, or description thereof. Similarly, it should be noted that, in order to simplify the description of this invention and thus aid in the understanding of one or more embodiments, the foregoing description of the embodiments of this invention sometimes combines multiple features into a single embodiment, drawing, or description thereof.

Claims

1. A drone-based intelligent positioning and cleaning method for photovoltaic pollution, characterized in that, The method includes: Acquire a set of pollution-related images collected by a drone while flying over a photovoltaic array, drone flight parameters, and photovoltaic array power generation efficiency-related data. The pollution-related image set contains visual information on surface pollution in each area of ​​the photovoltaic array. The drone flight parameters include the drone's flight altitude and flight angle when the drone collects pollution-related images of each area. The power generation efficiency-related data contains real-time power generation output information of each area of ​​the photovoltaic array. Based on the pollution-related image set, UAV flight parameters and power generation efficiency-related data, pollution area localization processing is performed to generate spatial positioning information of the pollution area in the photovoltaic array. The spatial positioning information includes the planar coordinates of the pollution area in the photovoltaic array and a description of the area range. By combining the spatial positioning information with pollution diffusion trend data of the polluted area and power generation priority data of the photovoltaic array, a pollution cleaning path for the UAV is planned. The pollution cleaning path includes the flight trajectory of the UAV between different polluted areas and the order of cleaning operations. According to the pollution cleaning path, the cleaning device carried by the UAV performs pollution cleaning operation. During the process, real-time pollution change images and real-time power generation efficiency change data of the cleaning area are collected simultaneously to generate pollution cleaning operation results of the UAV. The pollution cleaning operation results include cleaning completion status information of each pollution area. Based on the results of the pollution cleaning operation, real-time pollution change images, and real-time power generation efficiency change data, the cleaning effect is evaluated to generate pollution cleaning evaluation information for the photovoltaic array. The pollution cleaning evaluation information includes supplementary cleaning suggestions for areas that have not been cleaned. The method of planning the pollution cleaning path for the UAV by combining the spatial positioning information with the pollution diffusion trend data of the polluted area and the power generation priority data of the photovoltaic array includes: The planar coordinates and area descriptions of each polluted area in the spatial positioning information are analyzed to extract the area data and center coordinates of each polluted area. Acquire environmental data of the area where the photovoltaic array is located, analyze the pollution diffusion trend of each polluted area based on the environmental data, and generate pollution diffusion trend data of the polluted area, which reflects the future diffusion direction of the polluted area. Analyze the correlation data of power generation efficiency in each region of the photovoltaic array, and combine it with the overall power generation demand of the photovoltaic array to determine the power generation priority data of each region. The power generation priority data reflects the degree of influence of the power generation output of each region on the overall power generation efficiency. Based on the center coordinates of each polluted area, pollution diffusion trend data, and power generation priority data, all polluted areas are grouped. Polluted areas with diffusion direction angles less than a preset angle threshold, Euclidean distance between center points less than a preset distance threshold, and the same power generation priority level are divided into the same cleaning group. The priority of cleaning operations is determined based on the pollution diffusion trend data and power generation priority data of each cleaning group. Cleaning groups with a diffusion rate exceeding the preset rate threshold and a higher power generation priority value are given priority in cleaning operations. For each contaminated area within a cleaning group, the cleaning order within the group is determined based on the area data of the contaminated area and the power generation priority data. Contaminated areas with a contaminated area exceeding a preset area threshold and a higher power generation priority value are cleaned first. Based on the center coordinates of each cleaning group, the priority of cleaning operations, and the cleaning sequence of contaminated areas within the group, the flight trajectory of the drone between different cleaning groups is planned, and the contamination cleaning path of the drone in a single contaminated area is also planned, forming a contamination cleaning path that includes the flight trajectory of the drone between different contaminated areas and the cleaning sequence of operations. The process involves planning the flight paths of drones between different cleaning groups based on the center coordinates of each cleaning group, the priority of cleaning operations, and the cleaning sequence of contaminated areas within each group. Simultaneously, it plans the contamination cleaning path of the drone within a single contaminated area, including: Starting from the center coordinates of the current cleaning group and ending at the center coordinates of the next priority cleaning group, obtain the photovoltaic array structure data between the starting point and the ending point. The photovoltaic array structure data includes the arrangement of photovoltaic modules and the distribution of module connection nodes. Based on the photovoltaic array structure data, the flight trajectory of the UAV between the two cleaning groups is planned so that the flight trajectory avoids the protruding parts of the photovoltaic modules and the module connection nodes. For a single contaminated area, the cleaning boundary is determined based on the description of the area's extent and the data on the density of the contaminated area. Starting from the center of the contaminated area, the contamination cleaning path of the UAV within the contaminated area is planned in the direction of expansion from the area with a higher concentration of contamination density to the area with a lower concentration of contamination density. Path markers are set on the planned pollution cleaning path. These path markers are used by the UAV to adjust its position during flight so that its flight trajectory remains consistent with the pollution cleaning path. By acquiring the flight speed data of the drone and the cleaning speed data of the cleaning device, and combining them with the length data of the pollution cleaning path, the time required for the drone to complete each segment of flight and each area cleaning is calculated. By integrating time data with trajectory data of the pollution cleaning path, a complete pollution cleaning path is formed, which includes the correspondence between time data and trajectory data.

2. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 1, characterized in that, The acquisition of the pollution-related image set collected by the UAV while flying over the photovoltaic array, the UAV flight parameters, and the photovoltaic array power generation efficiency related data includes: The initial flight coverage frame of the UAV is determined based on the overall layout and orientation of the photovoltaic array. The initial flight coverage frame includes the basic flight direction of the UAV over the photovoltaic array. The drone is controlled to fly over the photovoltaic array according to the initial flight coverage framework. During the flight, the drone's flight altitude is dynamically adjusted according to the tilt angle and illumination angle of the photovoltaic array to maintain a stable distance between the drone and the surface of the photovoltaic array. Using image acquisition equipment mounted on a drone, the pollution-related image acquisition angle is synchronously adjusted according to the drone's flight angle during flight to acquire initial pollution-related images of each area of ​​the photovoltaic array. The initial pollution-related images contain the original visual information of the photovoltaic array surface. Visual pollution information of the photovoltaic array surface is extracted from the initial pollution-associated image. The visual pollution information includes color difference features, texture change features, and morphological contour features of the polluted area. The drone's flight altitude and angle when acquiring each initial pollution-related image were recorded as drone flight parameters. Real-time power output information of each region is collected by power generation monitoring components in each region of the photovoltaic array, and the real-time power output information is used as the power generation efficiency correlation data of the photovoltaic array. The pollution visual information, initial pollution-related images, UAV flight parameters, and power generation efficiency-related data corresponding to each region are linked and integrated to form a set of pollution-related images and a corresponding relationship between UAV flight parameters and power generation efficiency-related data.

3. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 1, characterized in that, The step of performing pollution area localization processing based on the pollution-related image set, UAV flight parameters, and power generation efficiency correlation data to generate spatial location information of the pollution area in the photovoltaic array includes: The pollution visual information of each pollution-related image is extracted from the pollution-related image set, and the pollution visual information is transformed into a pollution feature vector. The pollution feature vector is obtained by converting color difference features, texture change features and morphological contour features into numerical form. Retrieve the drone flight parameters corresponding to each pollution-related image, and calculate the field of view parameters when the pollution-related image is acquired based on the drone's flight altitude and flight angle. The field of view parameters reflect the actual size of the area covered by the photovoltaic array by the pollution-related image. Extract the power generation efficiency correlation data corresponding to each pollution-related image, analyze the difference between real-time power generation output information and normal power generation output information in the power generation efficiency correlation data, and obtain power generation efficiency difference data; A correspondence is established between the pixel coordinates of the pollution-associated image and the actual planar coordinates of the photovoltaic array. This correspondence is calculated by associating the field of view parameter with the pixel size of the pollution-associated image. Based on the pollution feature vector and the power generation efficiency difference data, the set of pollution-related image pixels belonging to the pollution area in each pollution-related image is determined. The set of pollution-related image pixels is then converted into the actual planar coordinates of the photovoltaic array through the correspondence relationship, thus obtaining the initial planar coordinates of the pollution area. The initial planar coordinates are processed by region integration, and adjacent sets of initial planar coordinates are integrated into a complete contaminated region. The boundary coordinates of each contaminated region are determined, forming spatial positioning information that includes the planar coordinates of the contaminated region in the photovoltaic array and a description of the region's extent.

4. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 3, characterized in that, The step of extracting pollution visual information from each pollution-associated image in the pollution-associated image set and converting the pollution visual information into a pollution feature vector includes: For each pollution-related image, a color space conversion is performed, converting the pollution-related image from the RGB color space to the HSV color space, and separating the hue channel, saturation channel, and lightness channel of the pollution-related image. In the hue channel, saturation channel and brightness channel, the feature differences between the polluted area and the normal photovoltaic surface area are extracted respectively, and the feature differences are converted into numerical form to obtain a numerical representation of the color difference features. Texture features are extracted from contaminated images. By analyzing the pixel distribution patterns of contaminated areas in the contaminated images, parameters reflecting texture changes are obtained and used as numerical results of texture change features. Contour feature extraction is performed on pollution-associated images to identify the contour lines of pollution areas in the pollution-associated images and extract the feature parameters of the contour lines as numerical data of morphological contour features. The numerical representations of color difference features, the numerical results of texture change features, and the numerical data of morphological contour features are arranged in a fixed order to form a pollution feature vector with fixed dimensions.

5. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 1, characterized in that, The process involves controlling the cleaning device mounted on the UAV to perform a pollution cleaning operation based on the pollution cleaning path. During the operation, real-time images of pollution changes and real-time power generation efficiency changes in the cleaning area are collected simultaneously to generate pollution cleaning operation results from the UAV, including: The contamination cleaning path is sent to the UAV's flight control system. The flight control system adjusts the UAV's flight attitude and speed based on the trajectory data in the contamination cleaning path, and controls the UAV to fly along the contamination cleaning path. When the drone reaches the designated contaminated area, it sends a start signal to the cleaning device using the flight control system. After receiving the start signal, the cleaning device starts to operate. During operation, the cleaning device adjusts the cleaning pressure and the amount of cleaning medium sprayed according to the contamination level data of the contaminated area. During the operation of the cleaning device, the image acquisition equipment on the drone is used to synchronously acquire real-time pollution change images of the cleaning area to obtain real-time pollution change images, which contain visual change information of the polluted area during the cleaning process. Simultaneously, real-time power output information is collected by the power generation monitoring components of the photovoltaic array in the designated polluted area to obtain real-time power generation efficiency change data, which includes the changes in real-time power generation output in the designated polluted area during the cleaning process. The system continuously collects real-time pollution change images and real-time power generation efficiency change data until the cleaning device completes the pollution cleaning operation in the polluted area, at which point it stops collecting real-time pollution change images and real-time power generation efficiency change data. Record the start and end times of cleaning for each contaminated area, collect real-time images of pollution changes, and collect real-time data on power generation efficiency changes. Integrate these records to generate a pollution cleaning operation result that includes information on the completion status of each contaminated area.

6. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 5, characterized in that, When the drone reaches the designated contaminated area, the flight control system sends a start signal to the cleaning device. Upon receiving the start signal, the cleaning device begins operation. During operation, the cleaning device adjusts the cleaning pressure and the amount of cleaning medium sprayed based on the contamination level data of the contaminated area, including: When the drone flies to the center coordinates of the designated contaminated area, the flight control system transmits a start signal to the cleaning device. Upon receiving the start signal, the cleaning device activates its internal contamination level detection component. The contamination level detection component detects the designated contaminated area and generates contamination level data for the area. The start signal includes the area range data of the contaminated area. The cleaning device internally stores data on the correspondence between contamination levels and cleaning parameters. This correspondence data includes cleaning pressure data and cleaning medium spraying volume data corresponding to different contamination levels. Based on the pollution level data of the polluted area, query the corresponding relationship data to determine the cleaning pressure data and cleaning medium spraying volume data corresponding to the current polluted area; Adjust the pressure regulating component inside the cleaning device to achieve the specified cleaning pressure data, and adjust the spraying component of the cleaning device to achieve the specified cleaning medium spraying volume data. After the cleaning device completes parameter adjustments, it begins cleaning the contaminated area. During the operation, the cleaning pressure data and the amount of cleaning medium sprayed are dynamically adjusted based on real-time images of changes in contamination.

7. The intelligent positioning and cleaning method for photovoltaic pollution using unmanned aerial vehicles according to claim 1, characterized in that, The process of evaluating the cleaning effect based on the results of the pollution cleaning operation, real-time pollution change images, and real-time power generation efficiency change data generates pollution cleaning evaluation information for the photovoltaic array, including: Analyze the cleaning completion status information of each polluted area in the pollution cleaning operation results, and extract the real-time pollution change image and real-time power generation efficiency change data corresponding to the cleaning end time of each polluted area; Extract residual pollution features from the real-time pollution change image corresponding to the end of the cleaning process. The residual pollution features include the color features, texture features, and morphological features of the residual pollution area. Analyze the real-time power generation efficiency change data at the end of the cleaning process, compare the real-time power generation output information of the polluted area before and after cleaning, and obtain the power generation efficiency recovery data. By comparing the pollution residue characteristics with the characteristic data of the normal area of ​​the photovoltaic array, and combining them with the power generation efficiency recovery data, the pollution residue data of each polluted area is obtained. The data on residual pollution in all polluted areas are collected. The number and area percentage of areas with no residual pollution after cleaning are calculated, as well as the number and area percentage of areas with residual pollution. At the same time, the number and area percentage of areas where the power generation efficiency recovery data reaches the expected recovery level are collected to form statistical data on the cleaning effect. For areas with residual pollution, supplementary cleaning recommendations are generated based on the characteristics of residual pollution and power generation efficiency recovery data. The supplementary cleaning recommendations include adjusted cleaning pressure data, cleaning medium spray volume data, and supplementary pollution cleaning path data. By integrating statistical data on cleaning effectiveness with recommendations for supplementary cleaning, a contamination cleaning assessment of the photovoltaic array is generated, which includes recommendations for supplementary cleaning of areas that were not properly cleaned.

8. A drone-based intelligent positioning and cleaning system for photovoltaic pollution, characterized in that, The device includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to run the programs, instructions or code in the memory to implement the UAV intelligent positioning and cleaning method for photovoltaic pollution as described in any one of claims 1-7.

Images