Photovoltaic fault location method and system based on drone

Abstract

The present invention discloses a photovoltaic fault location method and system based on an unmanned aerial vehicle (UAV), which specifically relates to the technical field of fault location. The method involves collecting images of photovoltaic modules along a preset trajectory using a thermal imaging camera and a visible light camera carried by the UAV, obtaining thermal radiation images and visible light images; registering and fusing the thermal radiation images with the visible light images to generate multimodal image data; performing superpixel segmentation on the multimodal images to form multiple analysis sub-regions; identifying temperature anomaly and color anomaly regions based on the temperature gradient distribution and color feature differences of the sub-regions; analyzing the propagation path of the anomaly features through spatial neighborhood topology to determine the diffusion range of the propagation path; comprehensively scoring the temperature anomaly and color anomaly regions based on the diffusion range of the propagation path, marking the fault region, and outputting spatial positioning information, thereby achieving high-precision positioning of photovoltaic module faults and improving the automation and intelligence level of fault detection.

Description

Technical Field

[0001] The present invention relates to the technical field of fault location, and more specifically, to a photovoltaic fault location method and system based on an unmanned aerial vehicle. Background Art

[0002] Existing technologies primarily rely on manual inspections or localized testing with ground-based infrared equipment for photovoltaic module fault detection. This makes it difficult to accurately locate fault locations in complex environments. Traditional methods often rely on data from only a single modality, making it difficult to comprehensively identify multiple types of faults.

[0003] In order to solve the above problems, a technical solution is now provided. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a photovoltaic fault location method and system based on a drone to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The photovoltaic fault location method based on drone includes the following steps:

[0007] S1: Use the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory to obtain corresponding thermal radiation image data and visible light image data;

[0008] S2: Register and fuse the thermal radiation image data and the visible light image data to generate multimodal image data in a unified spatial coordinate system;

[0009] S3: Perform superpixel segmentation on the multimodal image data to form multiple analysis sub-regions;

[0010] S4: Identify abnormal temperature and color areas based on the analysis of temperature gradient distribution characteristics and color feature differences in the sub-areas;

[0011] S5: Perform spatial neighborhood topology analysis on the temperature anomaly area and the color anomaly area to determine the propagation path of the anomaly feature and the diffusion range of the propagation path of the anomaly feature;

[0012] S6: Based on the diffusion range of the propagation path of the abnormal characteristics, the temperature abnormal area and the color abnormal area are comprehensively scored respectively;

[0013] S7: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a photovoltaic module fault area, and the spatial positioning information of the fault area is output.

[0014] In a preferred embodiment, S1 is specifically:

[0015] Using the thermal imaging camera and visible light camera onboard the drone, the drone will capture images of the surface area of ​​the photovoltaic module at a fixed altitude along a pre-planned flight path.

[0016] During the flight, thermal radiation images and visible light images of the photovoltaic module surface are captured at fixed intervals;

[0017] Marking the thermal radiation image and visible light image obtained by shooting according to the geographical location information at the time of shooting;

[0018] The thermal radiation image and the visible light image are respectively associated with the corresponding geographical location coordinates and flight altitude information to obtain thermal radiation image data and visible light image data.

[0019] In a preferred embodiment, S2 is specifically:

[0020] Perform coordinate system conversion on thermal radiation image data and visible light image data respectively, and uniformly convert the geographical location coordinates corresponding to the time of shooting into a standard geographic space coordinate system;

[0021] Perform spatial interpolation processing on thermal radiation image data and visible light image data in a standard geographic space coordinate system;

[0022] Perform spatial feature matching based on the corresponding photovoltaic module edge contour features and installation position features in the thermal radiation image data and the visible light image data, and extract spatial registration feature point pairs;

[0023] Based on the spatial registration feature point pairs, the coordinate transformation matrix between the thermal radiation image data and the visible light image data is calculated, and the spatial registration of the thermal radiation image data and the visible light image data is completed according to the coordinate transformation matrix;

[0024] The spatially registered thermal radiation image data and visible light image data are fused at the pixel level to generate multimodal image data in a unified spatial coordinate system.

[0025] In a preferred embodiment, S3 is specifically:

[0026] Based on the multimodal image data, a predetermined number of initial cluster centers are set to divide the multimodal image data into a number of initial cluster regions;

[0027] Calculate the distance between all pixels in the initial clustering area in terms of spatial position and pixel features;

[0028] Based on the weighted combination of the spatial position distance of pixels and the pixel feature distance, the position of the initial cluster center is adjusted and the cluster area is re-divided;

[0029] Repeat the adjustment of cluster center positions and the redivision of cluster areas until the sum of the spatial position distance and pixel feature distance of pixels in the cluster area no longer changes;

[0030] The boundary outline of the cluster area is determined as the outline of the analysis sub-area to form a plurality of analysis sub-areas.

[0031] In a preferred embodiment, S4 is specifically:

[0032] The temperature value corresponding to the thermal radiation image data is extracted for each pixel point in the analysis sub-area, and the spatial gradient change of the temperature values ​​of all pixels in the analysis sub-area is calculated to obtain the temperature gradient distribution characteristics of the analysis sub-area;

[0033] Extract the color value corresponding to the visible light image data for each pixel point in the analysis sub-area, perform statistics on the color values ​​of the pixels in the analysis sub-area, and obtain the color distribution characteristics of the analysis sub-area;

[0034] Based on the temperature gradient distribution characteristics of the analysis sub-region, determine whether the temperature gradient of the analysis sub-region is higher or lower than that of the adjacent analysis sub-region, and mark it as a temperature abnormal region;

[0035] Based on the color distribution characteristics of the analysis sub-regions, the analysis sub-regions whose color values ​​deviate from the adjacent analysis sub-regions are determined and marked as color abnormal regions.

[0036] In a preferred embodiment, the color values ​​include red channel values, green channel values, and blue channel values.

[0037] In a preferred embodiment, S5 is specifically:

[0038] Establish spatial neighborhood topological structures for temperature anomaly areas and color anomaly areas respectively, and determine the spatial neighboring relationship between temperature anomaly areas and color anomaly areas according to the spatial neighborhood topological structures;

[0039] Determine the propagation path of abnormal features in the temperature anomaly area and the color anomaly area based on the spatial adjacent relationship; the propagation path includes a spatial area chain consisting of multiple temperature anomaly areas or multiple color anomaly areas adjacent to each other in spatial position;

[0040] For the propagation path of the abnormal feature, the variation range of the temperature value gradient and the variation range of the color value difference between adjacent areas in the propagation path are calculated to obtain the diffusion range of the abnormal feature on the propagation path.

[0041] In a preferred embodiment, S6 is specifically:

[0042] Based on the diffusion range of the abnormal characteristics on the propagation path, the temperature value gradient change range of the temperature abnormal area and the color value difference change range of the color abnormal area are calculated respectively;

[0043] The scoring value of the temperature abnormality area is set according to the temperature value gradient change range, and the scoring value of the color abnormality area is set according to the color value difference change range;

[0044] Assign weight coefficients to the scoring values ​​of the temperature anomaly area and the color anomaly area respectively;

[0045] The score values ​​of the temperature anomaly area and the score values ​​of the color anomaly area are weighted by weight coefficients to obtain the comprehensive scores of the temperature anomaly area and the color anomaly area.

[0046] In a preferred embodiment, S7 is specifically:

[0047] Setting fault level thresholds, which include fault level thresholds for temperature abnormality areas and fault level thresholds for color abnormality areas;

[0048] Compare the comprehensive scores of the temperature abnormality area and the color abnormality area with the corresponding fault level threshold respectively to determine whether the comprehensive scores are higher than the corresponding fault level threshold;

[0049] When the comprehensive score of the temperature abnormality area or color abnormality area is higher than the corresponding fault level threshold, the temperature abnormality area or color abnormality area higher than the corresponding fault level threshold is marked as a photovoltaic module fault area;

[0050] Output is marked as the spatial positioning information of the fault area of ​​the photovoltaic module.

[0051] In another aspect, the present invention provides a photovoltaic fault location system based on a drone, comprising:

[0052] Image acquisition module: uses the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory, and obtains the corresponding thermal radiation image data and visible light image data;

[0053] Image fusion module: aligns and fuses thermal radiation image data and visible light image data to generate multimodal image data in a unified spatial coordinate system;

[0054] Region segmentation module: performs superpixel segmentation on multimodal image data to form multiple analysis sub-regions;

[0055] Abnormal identification module: Based on the analysis of the temperature gradient distribution characteristics and color feature differences of the sub-regions, it identifies the abnormal temperature and color areas;

[0056] Topological analysis module: performs spatial neighborhood topological analysis on abnormal temperature and color areas to determine the propagation path of abnormal features and the diffusion range of the propagation path of abnormal features;

[0057] Scoring calculation module: Based on the diffusion range of the propagation path of the abnormal characteristics, comprehensive scores are given to the temperature abnormality area and the color abnormality area respectively;

[0058] Fault marking module: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a PV module fault area, and the spatial positioning information of the fault area is output.

[0059] The technical effects and advantages of the photovoltaic fault location method and system based on drones of the present invention are as follows:

[0060] By using drones equipped with thermal imaging cameras and visible light cameras to obtain image data on the surface of photovoltaic modules, it is possible to efficiently collect images of photovoltaic modules over a large area, achieve rapid coverage and remote monitoring, and improve detection efficiency and operational safety; through the registration and fusion of thermal radiation images and visible light images, the image alignment accuracy and information complementarity capabilities are improved; through superpixel segmentation of multimodal images, fine-grained analysis of local areas is achieved; abnormal areas are identified based on temperature gradients and color features, enhancing the accuracy of identifying multiple types of fault features; the diffusion range of abnormal propagation paths is analyzed in combination with spatial neighborhood topological relationships, effectively capturing the fault propagation trend; the abnormal areas are comprehensively scored based on the diffusion range of the abnormal propagation paths to achieve a quantitative assessment of the severity of the fault; the fault area is determined and spatial positioning output is completed based on the comprehensive score, improving the accuracy of fault detection as well as the operation and maintenance capabilities and operational reliability of the photovoltaic system. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 Schematic diagram of the photovoltaic fault location method based on drone of the present invention;

[0062] Figure 2 The figure is a schematic structural diagram of the photovoltaic fault location system based on drone of the present invention. DETAILED DESCRIPTION

[0063] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0064] Example 1, Figure 1 The present invention provides a photovoltaic fault location method based on a drone, which includes the following steps:

[0065] S1: Use the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory to obtain corresponding thermal radiation image data and visible light image data;

[0066] S2: Register and fuse the thermal radiation image data and the visible light image data to generate multimodal image data in a unified spatial coordinate system;

[0067] S3: Perform superpixel segmentation on the multimodal image data to form multiple analysis sub-regions;

[0068] S4: Identify abnormal temperature and color areas based on the analysis of temperature gradient distribution characteristics and color feature differences in the sub-areas;

[0069] S5: Perform spatial neighborhood topology analysis on the temperature anomaly area and the color anomaly area to determine the propagation path of the anomaly feature and the diffusion range of the propagation path of the anomaly feature;

[0070] S6: Based on the diffusion range of the propagation path of the abnormal characteristics, the temperature abnormal area and the color abnormal area are comprehensively scored respectively;

[0071] S7: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a photovoltaic module fault area, and the spatial positioning information of the fault area is output.

[0072] S1: Use the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory, and obtain the corresponding thermal radiation image data and visible light image data, including:

[0073] Using the thermal imaging camera and visible light camera onboard the drone, the drone will capture images of the surface area of ​​the photovoltaic module at a fixed altitude along a pre-planned flight path.

[0074] Before the drone performs its image acquisition mission, the ground control equipment sets the drone's flight path. The flight path is set by using the boundary position of the area where the photovoltaic panels are located as the planning basis for the flight path, dividing the flight trajectory into multiple adjacent straight segments in a manner that covers all photovoltaic panels. At the same time, the drone's flight altitude on each straight segment is set to a fixed level, so that the drone's height relative to the ground remains stable during flight, thereby ensuring the consistency of the spatial resolution of the acquired images. For example, when setting the actual flight path, the path planning function of the drone's flight control software can be used to set the flight trajectory to multiple parallel straight lines with constant spacing, completely covering the entire area where the photovoltaic panels are located, avoiding missed or duplicate shots during flight.

[0075] During the flight, thermal radiation images and visible light images of the photovoltaic module surface are captured at fixed intervals;

[0076] As the drone flies along a set flight path, it uses a thermal imaging camera and a visible light camera to capture images of the surface of the photovoltaic panels below at preset intervals. The images captured by the thermal imaging camera are thermal radiation images reflecting the surface temperature distribution of the photovoltaic panels, while the images captured by the visible light camera are visible light images reflecting the color and outline of the photovoltaic panels. To ensure that the image coverage areas between adjacent shooting positions overlap and avoid blind spots, the interval between the thermal radiation and visible light images is calculated based on the drone's flight speed and the camera's field of view. The interval is calculated as the ratio of the ground width covered by the camera's field of view to the drone's flight speed. Specifically, the interval is calculated by dividing the ground width covered by the camera's field of view by the drone's flight speed.

[0077] Marking the thermal radiation image and visible light image obtained by shooting according to the geographical location information at the time of shooting;

[0078] When a drone captures each thermal radiation image and visible light image, its onboard positioning device simultaneously records the drone's geographic coordinates at the time of capture. These coordinates consist of longitude, latitude, and altitude, and each location is associated with the corresponding thermal radiation image and visible light image. For example, by recording the longitude, latitude, and altitude at the moment a drone captures a photo, these coordinates can be stored in the data associated with the image file. This allows the spatial origin of each image to be clearly identified, ensuring the accurate correspondence between image and location.

[0079] Associating the thermal radiation image and the visible light image with the corresponding geographical location coordinates and flight altitude information respectively to obtain thermal radiation image data and visible light image data;

[0080] After image capture and location tagging, each thermal radiation image and visible light image is linked to its corresponding longitude and latitude values, as well as the drone's flight altitude, to form an image dataset with spatial information. The thermal radiation image data includes information on the spatial distribution of PV module surface temperature, along with the corresponding geographic coordinates and drone's flight altitude. The visible light image data includes information on the spatial distribution of PV module surface color, along with the corresponding geographic coordinates and drone's flight altitude.

[0081] S2: Register and fuse the thermal radiation image data and the visible light image data to generate multimodal image data in a unified spatial coordinate system, including:

[0082] Perform coordinate system conversion on thermal radiation image data and visible light image data respectively, and uniformly convert the geographical location coordinates corresponding to the time of shooting into a standard geographic space coordinate system;

[0083] The geographic location coordinates recorded by the drone when capturing thermal radiation and visible light images are converted into a unified standard geospatial coordinate system. The standard geospatial coordinate system is defined as a geographic coordinate system based on the Earth's reference ellipsoid, using longitude, latitude, and altitude to determine spatial locations. The conversion process includes the following steps: Using known coordinate conversion methods, the geographic location coordinates recorded by the drone are converted from the coordinate system provided by the drone's onboard positioning device (e.g., the coordinate system provided by a global satellite navigation and positioning system) into a standard geospatial coordinate system. The coordinate conversion method involves first determining the spatial position difference and rotation angle difference between the original coordinate system and the standard geospatial coordinate system, and then performing spatial coordinate translation and rotation operations to achieve the conversion. The calculation method of spatial coordinate translation is: add or subtract the value of the original geographic location coordinate on each coordinate axis and the difference between the coordinate systems to obtain the geographic location coordinate value in the standard geographic space coordinate system; the calculation method of spatial coordinate rotation is: according to the size of the rotation angle between the two coordinate systems, use the mathematical formula of spatial coordinate rotation to calculate, specifically, perform a combined operation of the sine and cosine of the rotation angle on each axial value of the original geographic location coordinate to obtain the coordinate value of the standard geographic space coordinate system.

[0084] Perform spatial interpolation processing on thermal radiation image data and visible light image data in a standard geographic space coordinate system;

[0085] After completing the coordinate system transformation, spatial interpolation is performed on the thermal radiation image data and visible light image data, ensuring that they are identical in spatial resolution and position. The spatial interpolation method is as follows: first, determine the original spatial resolution values ​​of the thermal radiation image data and the visible light image data; then, use the spatial resolution of the image data with the higher spatial resolution as the target resolution and perform interpolation on the image data with the lower spatial resolution. The spatial interpolation calculation method is as follows: The value of each pixel position in the interpolated image data is calculated using a distance-weighted average method based on the relative positional relationship between the pixel position at the target resolution and the original pixel position. This method uses the inverse of the distance between the pixel values ​​of neighboring pixels and the interpolated position as the weight to calculate the interpolated pixel value at the target position. Through interpolation, the thermal radiation image data and the visible light image data are ensured to have exactly the same spatial size and position.

[0086] Perform spatial feature matching based on the corresponding photovoltaic module edge contour features and installation position features in the thermal radiation image data and the visible light image data, and extract spatial registration feature point pairs;

[0087] Feature points reflecting the edge contour and installation position characteristics of the photovoltaic module are extracted from the interpolated thermal radiation image data and visible light image data, respectively. The feature point extraction method is as follows: the edge contour lines of the photovoltaic module are extracted respectively by an image edge detection method. The edge detection method compares the numerical differences of adjacent pixels in the image pixel by pixel, and marks them as edge points when the differences exceed a preset threshold. The feature points of the installation position are determined by an image feature recognition method. The image feature recognition method analyzes the color distribution, brightness distribution, and temperature gradient changes in the image to determine special points that can reflect the installation position of the photovoltaic module. Feature point sets are established for the feature points extracted from the thermal radiation image data and the visible light image data, and by comparing the similarity of the feature distributions in the feature point sets of the two images, feature points with consistent spatial positions are matched into spatial registration feature point pairs to form feature point correspondence relationships. For example, the relative distance relationship between the feature point positions can be selected as the matching basis. That is, if the relative position relationship between a feature point in the thermal radiation image data and a feature point in the visible light image data satisfies spatial consistency, they are determined as spatial registration feature point pairs.

[0088] Based on the spatial registration feature point pairs, the coordinate transformation matrix between the thermal radiation image data and the visible light image data is calculated, and the spatial registration of the thermal radiation image data and the visible light image data is completed according to the coordinate transformation matrix;

[0089] The coordinate transformation matrix between the two image data sets is determined using the identified spatial registration feature point pairs. This matrix is ​​determined by establishing a corresponding set of spatial point coordinates between the two images based on the spatial registration feature point pairs. The positional offset, scaling ratio, and rotation angle between these sets are then calculated, and the coordinate transformation matrix is ​​constructed based on these offsets, scaling ratios, and rotation angles. The coordinate transformation matrix accurately converts one image data set to the spatial position of another. Once the coordinate transformation matrix is ​​obtained, it is applied to the thermal radiation image data or the visible light image data, resulting in precise spatial registration of the two image data sets.

[0090] The spatially registered thermal radiation image data and visible light image data are fused at the pixel level to generate multimodal image data in a unified spatial coordinate system;

[0091] After completing the spatial registration, the two image data are fused in a pixel-by-pixel fusion manner, that is, the temperature information and color information of each corresponding pixel point are integrated into a single pixel to form multimodal image data containing both temperature features and color features; the fusion method includes: storing the temperature value and color value of the corresponding pixel point at the same pixel position respectively, to form multimodal image data in a unified spatial coordinate system that can simultaneously perform temperature and color feature analysis.

[0092] S3: Perform superpixel segmentation on the multimodal image data to form multiple analysis sub-regions, including:

[0093] Based on the multimodal image data, a predetermined number of initial cluster centers are set to divide the multimodal image data into a number of initial cluster regions;

[0094] A clustering operation is performed on the multimodal image data in a unified spatial coordinate system, and the positions of multiple initial cluster centers are set. The positions of the initial cluster centers are pre-set and evenly distributed in the image data space so that they can reasonably cover the entire spatial area where the multimodal image data is located. For example, the initial cluster centers can be set in the following manner: the spatial area corresponding to the multimodal image data is divided into a number of grid units of equal size, and the position of the center point of each grid unit is used as the position of an initial cluster center; according to the position of each initial cluster center, each pixel in the multimodal image data is respectively assigned to the nearest initial cluster center, thereby forming several initial cluster areas. The above division method can ensure that all pixel points belong to one and only one cluster area during the initial clustering process, avoiding omissions or repeated divisions.

[0095] Calculate the distance between all pixels in the initial clustering area in terms of spatial position and pixel features;

[0096] For each initial clustering region, the spatial position distance and pixel feature distance of each pixel in the clustering region relative to the initial clustering center are calculated. The spatial position distance refers to the geometric straight-line distance between the position of a pixel in the spatial coordinate system and the position of the initial clustering center of the clustering region. The calculation method is as follows: first, the position differences between the pixel and the initial clustering center in the longitude, latitude, and altitude directions are calculated, the position differences are squared and added, and finally the square root of the sum is taken to obtain the spatial position distance. The pixel feature distance is calculated as follows: pixel features include temperature values ​​and color values, where color values ​​specifically include red channel values, green channel values, and blue channel values. During the calculation, the difference between the temperature value of a pixel and the temperature value corresponding to the initial clustering center, as well as the difference between each color channel in the pixel color value and the color channel value corresponding to the initial clustering center, are calculated, the above differences are squared and added, and finally the square root of the sum is taken to obtain the pixel feature distance.

[0097] Based on the weighted combination of the spatial position distance of pixels and the pixel feature distance, the position of the initial cluster center is adjusted and the cluster area is re-divided;

[0098] After completing the distance calculation, for each initial clustering area, the spatial position distance and pixel feature distance are weighted and combined in a certain ratio to obtain the comprehensive distance value of each pixel relative to the initial cluster center. The weighted combination calculation method is as follows: set the weight values ​​corresponding to the spatial position distance and pixel feature distance respectively; multiply the spatial position distance value by the corresponding weight value, and multiply the pixel feature distance value by its corresponding weight value, and finally add them together to obtain the comprehensive distance value of each pixel. Based on the comprehensive distance values ​​of all pixels in each initial clustering area, the position of the new cluster center is recalculated. The method for determining the position of the new cluster center is as follows: the spatial position coordinates of all pixels in the clustering area are weighted by the inverse of the comprehensive distance value corresponding to each pixel point, and the spatial position coordinates are weighted averaged to obtain the spatial position coordinates of the new cluster center; the weighted average of the pixel feature values ​​corresponding to all pixels in the clustering area is calculated in the same way to obtain the pixel feature value of the new cluster center. Based on the newly calculated cluster center position and pixel features, each pixel point in the multimodal image data is divided again, and the pixel points are reclassified into the cluster area corresponding to the new cluster center with the smallest comprehensive distance, completing a new round of cluster area division.

[0099] Repeat the adjustment of cluster center positions and the redivision of cluster areas until the sum of the spatial position distance and pixel feature distance of pixels in the cluster area no longer changes;

[0100] According to the above method, the cluster center position adjustment and pixel point re-division are repeatedly performed. After each re-division of the cluster area, the sum of the spatial position distance and the pixel feature distance of the pixels in each cluster area is calculated and compared to determine whether the cluster area is stable. The determination method is as follows: based on the current cluster area division result, the cluster center position is adjusted again and a new cluster area division is performed. Then, the sum of the spatial position distance and the pixel feature distance of the pixels in the newly divided cluster area is calculated. The sum of the distance calculated this time is compared with the sum of the distances corresponding to the previous division result. If the difference between the two calculated distance sums is less than a predetermined threshold value, it is determined that the cluster area has reached stability; otherwise, if the difference is greater than the predetermined threshold value, the next cluster center position adjustment and pixel point re-division step is continued until the cluster area is stable.

[0101] Determine the boundary outline of the cluster area as the outline of the analysis sub-area to form multiple analysis sub-areas;

[0102] After the cluster regions stabilize, the pixel distribution within each cluster region is clearly defined, with distinct boundaries. The outer boundary pixels of each cluster region are connected to form a closed contour line, which is then defined as the boundary outline of the analysis subregion. Specifically, the pixels at the edge of the cluster region are identified by examining all pixels within the region one by one. If a pixel touches at least one adjacent pixel belonging to another cluster region, it is considered a boundary pixel. The identified boundary pixels are connected in a spatially continuous sequence to obtain a complete closed contour line. The spatial relationship between all pixels within the cluster region and the boundary contour line is clearly expressed, i.e., pixels within the cluster region are located inside the contour line, while pixels outside the contour line belong to other cluster regions or the image boundary. Through this method, several spatially continuous and well-defined analysis subregions are ultimately formed.

[0103] S4: Based on the analysis of the temperature gradient distribution characteristics and color feature differences of the sub-regions, identify the temperature abnormality areas and color abnormality areas, including:

[0104] The temperature value corresponding to the thermal radiation image data is extracted for each pixel point in the analysis sub-area, and the spatial gradient change of the temperature values ​​of all pixels in the analysis sub-area is calculated to obtain the temperature gradient distribution characteristics of the analysis sub-area;

[0105] All pixels within each analysis subregion are processed one by one, and the temperature value corresponding to each pixel in the thermal radiation image data is extracted. The temperature value is the temperature information in a unified spatial coordinate system after spatial coordinate transformation, interpolation, and registration. Each pixel corresponds to a unique temperature value. A spatial gradient calculation is performed on the temperature values ​​of all pixels within each analysis subregion. The spatial gradient calculation method is as follows: With a pixel as the center, the temperature difference between the central pixel and its adjacent pixels is calculated, forming temperature difference values ​​in multiple directions. The temperature difference values ​​are squared and summed, and the square root of the summed result is taken to obtain the temperature spatial gradient value at the central pixel. This method is repeated for all pixels within the analysis subregion, and the spatial gradient distribution of the temperature values ​​within the entire analysis subregion is finally obtained. The obtained spatial gradient value represents the intensity of temperature changes within the analysis subregion and the consistency of the temperature change direction, reflecting the characteristics of the temperature gradient distribution within the analysis subregion.

[0106] Extract the color value corresponding to the visible light image data for each pixel point in the analysis sub-area, perform statistics on the color values ​​of the pixels in the analysis sub-area, and obtain the color distribution characteristics of the analysis sub-area;

[0107] For each pixel in the same analysis sub-region, the corresponding color value in the visible light image data is extracted. The color value consists of three components: the red channel value, the green channel value, and the blue channel value. Each pixel corresponds to a unique set of color values. Color value statistics are performed on all pixels in the analysis sub-region, that is, the distribution of each color channel value in the analysis sub-region is statistically analyzed. The statistical method is: the red channel value, the green channel value, and the blue channel value are separately summarized in the entire analysis sub-region, and the frequency of occurrence of each color channel value in the analysis sub-region and the average difference of each channel value relative to the color values ​​of all pixels are calculated to reflect the degree of concentration or dispersion of the distribution of color values ​​in the analysis sub-region. The statistical results reflect the color distribution characteristics of the analysis sub-region. The color distribution characteristics can reflect the uniformity or abnormality of the color characteristics of the analysis sub-region.

[0108] Based on the temperature gradient distribution characteristics of the analysis sub-region, determine whether the temperature gradient of the analysis sub-region is higher or lower than that of the adjacent analysis sub-region, and mark it as a temperature abnormal region;

[0109] After calculating the temperature gradient distribution characteristics of all analysis sub-regions, a comparative analysis of the temperature gradient characteristics between the analysis sub-regions is performed. The comparative analysis method is as follows: the average temperature gradient value of all pixels in each analysis sub-region is calculated, and the average value is used as an indicator representing the temperature gradient characteristics of the analysis sub-region; for each analysis sub-region, its temperature gradient characteristic index is compared with the temperature gradient characteristic index of the adjacent analysis sub-regions one by one. If the temperature gradient characteristic index of the analysis sub-region is significantly higher or significantly lower than the temperature gradient characteristic index of its adjacent analysis sub-region, the analysis sub-region is marked as a temperature anomaly area. By comparing all analysis sub-regions one by one through the above method, the temperature anomaly areas are marked.

[0110] Based on the color distribution characteristics of the analysis sub-region, determine the analysis sub-region where the color value deviates from the adjacent analysis sub-region, and mark it as a color abnormal region;

[0111] The color distribution characteristics of all analysis subregions are analyzed and compared to identify analysis subregions whose color values ​​significantly deviate from those of adjacent analysis subregions. The specific method is as follows: for each analysis subregion, the average value of each color channel (including the red channel value, the green channel value, and the blue channel value) is calculated to form a color characteristic index; the color characteristic index of the analysis subregion is compared with the color characteristic index of the corresponding color channels of the adjacent analysis subregions, and the difference between the color channel values ​​is calculated; if a difference is found between the average color channel value of a certain analysis subregion and the average color channel value of the corresponding color channel of its adjacent analysis subregion, the color value of the analysis subregion is determined to deviate from the adjacent analysis subregion. All analysis subregions whose color values ​​deviate from adjacent analysis subregions are marked as color anomaly regions. For example, if the average red channel value of a certain analysis subregion is much higher or much lower than the average red channel value of the adjacent analysis subregion, the analysis subregion is determined to have a color anomaly and is marked as a color anomaly region.

[0112] S5: Perform spatial neighborhood topology analysis on the temperature anomaly area and the color anomaly area to determine the propagation path of the anomaly feature and the diffusion range of the propagation path of the anomaly feature, including:

[0113] Establish spatial neighborhood topological structures for temperature anomaly areas and color anomaly areas respectively, and determine the spatial neighboring relationship between temperature anomaly areas and color anomaly areas according to the spatial neighborhood topological structures;

[0114] Based on the boundary contours of multiple determined temperature anomaly areas and color anomaly areas, the positional relationship of each anomaly area is determined one by one in a unified spatial coordinate system; the spatial adjacent relationship between the anomaly areas is established by using a spatial neighborhood topological structure: the boundary contours of each anomaly area are analyzed through spatial position to determine the pairs of anomaly areas with common boundaries or common vertices; the pairs of anomaly areas with common boundaries or common vertices are defined as pairs of anomaly areas with a spatial neighborhood relationship and recorded as adjacent areas; a neighborhood topological relationship table is established, in which the spatial adjacency relationship of each temperature anomaly area and color anomaly area with its adjacent anomaly areas is recorded; the adjacency relationship is recorded as follows: the unique identifier of each anomaly area is listed in the neighborhood topological relationship table, and the unique identifiers of other anomaly areas with which it has a spatial adjacent relationship are listed separately to form a topological relationship record; the topological relationship record can accurately reflect the spatial neighborhood connection between each anomaly area and other anomaly areas, providing a basis for determining the propagation path of anomaly features.

[0115] Determine the propagation paths of abnormal features in temperature anomaly areas and color anomaly areas based on spatial adjacent relationships;

[0116] Based on the spatial neighborhood topology, the spatial anomaly characteristic propagation paths of temperature and color anomaly regions are analyzed and determined. The propagation path determination method is as follows: starting with each temperature anomaly region and color anomaly region, the neighborhood topology is examined, and the directly adjacent anomaly region is found in sequence. The next adjacent anomaly region is examined, and so on, expanding until no further extension is possible. Ultimately, a continuous chain of one or more spatially adjacent anomaly regions is formed, which is the anomaly characteristic propagation path. For example, for a temperature anomaly region, its neighborhood topology is first examined. If one or more spatially adjacent temperature anomaly regions exist, these adjacent regions are sequentially added to the temperature anomaly region's propagation path. The newly added adjacent region is then used as the new starting region for examination, and this process is repeated until no new adjacent temperature anomaly regions are added. This completes the determination of the temperature anomaly region's propagation path. The propagation path determination method for color anomaly regions is the same as that for temperature anomaly regions. Ultimately, a temperature anomaly characteristic propagation path consisting of multiple spatially adjacent temperature anomaly regions and a color anomaly characteristic propagation path consisting of multiple spatially adjacent color anomaly regions are formed. The propagation path includes a spatial region chain consisting of a plurality of temperature anomaly regions or a plurality of color anomaly regions that are adjacent in spatial position.

[0117] For the propagation path of the abnormal feature, calculate the variation range of the temperature value gradient and the variation range of the color value difference between adjacent areas in the propagation path to obtain the diffusion range of the abnormal feature on the propagation path;

[0118] For each pair of adjacent temperature anomaly regions along the identified propagation path of the temperature anomaly characteristic, a detailed calculation and analysis of the temperature gradient variation range is performed. Specifically, the average difference in temperature values ​​at the boundary between each pair of adjacent temperature anomaly regions within the propagation path is calculated. First, several boundary locations are uniformly selected along the boundary line of the adjacent regions. At each boundary location, the temperature difference between the corresponding regions on both sides is calculated. After calculating the temperature difference of all selected boundary locations, the maximum and minimum values ​​of the difference are calculated. The difference between the maximum and minimum values ​​is used as the temperature gradient variation range, representing the spatial diffusion intensity of the temperature anomaly characteristic between adjacent temperature anomaly regions. The above steps are performed one by one for all pairs of adjacent temperature anomaly regions within the entire propagation path to obtain the temperature gradient variation range between each pair of adjacent regions. The overall set of temperature gradient variation ranges is the diffusion range of the temperature anomaly characteristic along the entire propagation path.

[0119] For the color anomaly feature propagation path, a calculation method similar to that for the temperature anomaly feature propagation path is used to calculate the range of color value differences between adjacent color anomaly areas within the path one by one. Specifically, for each pair of adjacent color anomaly areas within the propagation path, multiple evenly distributed boundary pixels are selected, and the red channel, green channel, and blue channel values ​​of the boundary pixels are extracted respectively. For each pair of boundary pixels, the value differences of the three color channels are calculated respectively, and then the value differences of the three channels are squared respectively. The squared results are added together and then squared to obtain the overall difference value of the color value of each boundary pixel. The maximum and minimum values ​​of the color value differences of all boundary pixels are counted, and the difference between the maximum and minimum values ​​is calculated to obtain the range of color value differences, which reflects the color difference diffusion intensity of the color anomaly feature between adjacent color anomaly areas. The above steps are repeated to calculate the range of color value differences between all adjacent color anomaly areas within the propagation path one by one. The range of all color value differences together constitutes the diffusion range of the color anomaly feature on the entire propagation path.

[0120] S6: Based on the diffusion range of the propagation path of the abnormal characteristics, the temperature abnormality area and the color abnormality area are comprehensively scored, including:

[0121] Based on the diffusion range of the abnormal characteristics on the propagation path, the temperature value gradient change range of the temperature abnormal area and the color value difference change range of the color abnormal area are calculated respectively;

[0122] According to the diffusion range of the determined abnormal characteristic propagation path, quantitative analysis and calculation of temperature anomaly areas and color anomaly areas are performed respectively; first, the temperature gradient variation range between adjacent areas in the propagation path of the temperature anomaly area is statistically analyzed and summarized; the statistical method of the temperature numerical gradient variation range is as follows: for each pair of spatially adjacent temperature anomaly areas, the temperature difference data of all boundary positions on its boundary are obtained; the temperature difference values ​​of all boundary positions are statistically summarized separately; the maximum temperature difference and the minimum temperature difference are found from the summarized values; the difference between the maximum and minimum temperature difference is calculated as the temperature numerical gradient variation range; all adjacent area pairs on the entire temperature anomaly characteristic propagation path are statistically calculated one by one to obtain the temperature gradient variation range set between each pair of adjacent areas; the values ​​in the temperature gradient variation range set of all adjacent areas are uniformly sorted out to form the total range of temperature numerical gradient variation on the propagation path of the temperature anomaly area.

[0123] A statistical analysis is performed on the range of color value difference changes between adjacent areas in the propagation path of the color abnormality area; the statistical method for the range of color value difference changes is as follows: for each boundary position between each pair of spatially adjacent color abnormality areas, the color values ​​of the red channel, green channel and blue channel are extracted respectively; for each boundary position, the color value difference between the pixels on both sides of the boundary of the adjacent areas of the three color channels is calculated respectively, the calculated color value differences are squared respectively, and then the squared results are added, and finally the square root operation is performed on the added results to obtain the color difference value; the color difference values ​​of all boundary positions are statistically summarized respectively, and the maximum and minimum color difference values ​​are found from the summarized values; the difference between the maximum and minimum color difference values ​​is taken as the range of color value difference changes; the above statistical calculation is repeated for all adjacent area pairs on the entire color abnormality feature propagation path, and finally the complete total range of color value difference changes is obtained.

[0124] The scoring value of the temperature abnormality area is set according to the temperature value gradient change range, and the scoring value of the color abnormality area is set according to the color value difference change range;

[0125] For the temperature value gradient change range, the corresponding temperature anomaly area score value is set respectively; the score value setting method is: based on the size of the temperature gradient change range value, the score value of each temperature anomaly area is determined according to the principle that the larger the value, the higher the score value; the way to determine the score value is: first set the maximum allowable threshold and the minimum allowable threshold of the temperature gradient change range; if the temperature gradient change range of a certain temperature anomaly area reaches or exceeds the maximum allowable threshold, the highest score value is given to the temperature anomaly area; if the temperature gradient change range is lower than the maximum allowable threshold but higher than the minimum allowable threshold, a score value corresponding to the proportion is determined based on the ratio of the temperature gradient change range to the maximum and minimum allowable thresholds; if the temperature gradient change range is lower than the minimum allowable threshold, the lowest score value is given to the temperature anomaly area.

[0126] Corresponding color abnormality area score values ​​are set for the color value difference variation range respectively; the scoring value setting method is: based on the size of the color value difference variation range, the scoring value is determined according to the principle that the larger the value, the higher the score value; specifically: first set the maximum allowable threshold and the minimum allowable threshold of the color value difference variation range; when the color value difference variation range of a certain color abnormality area reaches or exceeds the maximum allowable threshold, the highest score value is given to the color abnormality area; when the color value difference variation range is between the maximum allowable threshold and the minimum allowable threshold, the score value is determined by the ratio of the color value difference variation range value to the maximum and minimum allowable thresholds; when the color value difference variation range is lower than the minimum allowable threshold, the lowest score value is given to the color abnormality area.

[0127] Assign weight coefficients to the scoring values ​​of the temperature anomaly area and the color anomaly area respectively;

[0128] The method for determining the weight coefficient is as follows: the weight coefficient is determined according to the importance of temperature anomaly characteristics and color anomaly characteristics to the diagnosis of photovoltaic module faults; specifically, the importance ratio of temperature anomaly characteristics and color anomaly characteristics in fault judgment is first determined, and the corresponding weight coefficients are given according to the contribution of temperature anomaly characteristics and color anomaly characteristics to the failure of photovoltaic modules; the sum of the weight coefficient of the temperature anomaly area and the weight coefficient of the color anomaly area is one, and the size of each coefficient directly reflects the proportion of the corresponding abnormal feature in fault identification.

[0129] The score values ​​of the temperature anomaly area and the score values ​​of the color anomaly area are weighted by weight coefficients to obtain the comprehensive score of the temperature anomaly area and the color anomaly area;

[0130] The scores for the temperature anomaly area and the color anomaly area are weighted and calculated separately with their respective weight coefficients. Specifically, the temperature anomaly area score is multiplied by the corresponding weight coefficient; the color anomaly area score is multiplied by the corresponding weight coefficient; the results of these weighted calculations are recorded as the weighted score for the temperature anomaly area and the weighted score for the color anomaly area, respectively. The weighted scores for the temperature anomaly area and the color anomaly area are added together to obtain a comprehensive score for each anomaly area. The comprehensive score reflects the overall degree of anomaly and the likelihood of failure for each anomaly area, providing a basis for identifying and determining faulty areas in PV modules.

[0131] S7: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a PV module fault area, and the spatial location information of the fault area is output, including:

[0132] Setting fault level thresholds, which include fault level thresholds for temperature abnormality areas and fault level thresholds for color abnormality areas;

[0133] The fault level threshold of the temperature abnormality area is set; the fault level threshold of the temperature abnormality area means that when the comprehensive score of the temperature abnormality area reaches the fault level threshold, it is regarded as a temperature abnormality that affects the operating status of the photovoltaic module; in the actual setting process, based on historical data analysis and expert experience, the critical value representing the degree of abnormal risk in the temperature abnormality area score value is determined, which is the fault level threshold.

[0134] The fault level threshold for the color anomaly area indicates that when the combined score of the color anomaly area reaches the threshold, the color anomaly characteristics on the PV module surface have reached a noticeable abnormality. Based on historical experience in actual photography and testing, and through long-term observation and manual review, the critical score value for actual failures has been determined. When the combined score of the color anomaly area reaches or exceeds the critical score value, the PV module will show visual performance degradation. This critical score value is the fault level threshold for the color anomaly area.

[0135] Compare the comprehensive scores of the temperature abnormality area and the color abnormality area with the corresponding fault level threshold respectively to determine whether the comprehensive scores are higher than the corresponding fault level threshold;

[0136] Compare the comprehensive score of the temperature abnormality area with the corresponding set fault level threshold to determine whether the comprehensive score exceeds the fault level threshold of the temperature abnormality area: subtract the fault level threshold of the temperature abnormality area from the comprehensive score of the temperature abnormality area, and determine whether the temperature abnormality area is a fault area based on the positive or negative value of the calculation result; if the calculation result is a positive value, it means that the comprehensive score of the temperature abnormality area exceeds the fault level threshold of the temperature abnormality area and is determined to be a PV module fault area; if the calculation result is a positive value or zero, it means that the temperature abnormality area is not determined to be a PV module fault area;

[0137] The comprehensive score of the color abnormality area is compared with the fault level threshold of the corresponding color abnormality area, that is, the comprehensive score of the color abnormality area is subtracted from the fault level threshold of the corresponding color abnormality area, and the positive or negative value of the calculated result is used to determine whether the color abnormality area meets the standard of the fault area; when the calculated result is a positive value, the color abnormality area is determined to be a fault area; when the calculated result is zero or a negative value, the color abnormality area is determined not to be a fault area.

[0138] When the comprehensive score of the temperature abnormality area or color abnormality area is higher than the corresponding fault level threshold, the temperature abnormality area or color abnormality area higher than the corresponding fault level threshold is marked as a photovoltaic module fault area;

[0139] After completing the comparison and judgment of the comprehensive score and the threshold, mark all temperature anomaly areas or color anomaly areas whose comprehensive scores are higher than the corresponding fault level threshold; specifically, establish a record table of fault areas; the content of the record table includes the unique number, comprehensive score and spatial coordinate information of the temperature anomaly area or color anomaly area determined to be a fault area.

[0140] Outputting spatial positioning information of the fault area of ​​the photovoltaic module;

[0141] Spatial positioning information includes the longitude, latitude, and altitude coordinates of each fault area within a unified spatial coordinate system. This positioning information is output by displaying the marked faulty PV panels as an image, map, or 3D model, indicating the geographic location of each area, on a ground control device display or other human-computer interaction device. This positioning data is also recorded and stored as an electronic file or report, and transmitted to the terminal equipment of ground maintenance personnel.

[0142] Example 2: The difference between Example 2 of the present invention and Example 1 is that this example introduces a photovoltaic fault location system based on a drone.

[0143] Figure 2 The present invention provides a schematic structural diagram of a photovoltaic fault location system based on a drone. The photovoltaic fault location system based on a drone includes:

[0144] Image acquisition module: uses the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory, and obtains the corresponding thermal radiation image data and visible light image data;

[0145] Image fusion module: aligns and fuses thermal radiation image data and visible light image data to generate multimodal image data in a unified spatial coordinate system;

[0146] Region segmentation module: performs superpixel segmentation on multimodal image data to form multiple analysis sub-regions;

[0147] Abnormal identification module: Based on the analysis of the temperature gradient distribution characteristics and color feature differences of the sub-regions, it identifies the abnormal temperature and color areas;

[0148] Topological analysis module: performs spatial neighborhood topological analysis on abnormal temperature and color areas to determine the propagation path of abnormal features and the diffusion range of the propagation path of abnormal features;

[0149] Scoring calculation module: Based on the diffusion range of the propagation path of the abnormal characteristics, comprehensive scores are given to the temperature abnormality area and the color abnormality area respectively;

[0150] Fault marking module: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a PV module fault area, and the spatial positioning information of the fault area is output.

[0151] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0152] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0153] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0154] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0156] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0157] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0158] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0159] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0160] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. The photovoltaic fault location method based on drone is characterized by: The steps include: S1: Use the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory to obtain corresponding thermal radiation image data and visible light image data; S2: Register and fuse the thermal radiation image data and the visible light image data to generate multimodal image data in a unified spatial coordinate system; S3: Perform superpixel segmentation on the multimodal image data to form multiple analysis sub-regions; Based on the multimodal image data, a predetermined number of initial cluster centers are set to divide the multimodal image data into a number of initial cluster regions; Calculate the distance between all pixels in the initial clustering area in terms of spatial position and pixel features; Based on the weighted combination of the spatial position distance of pixels and the pixel feature distance, the position of the initial cluster center is adjusted and the cluster area is re-divided; Repeat the adjustment of cluster center positions and the redivision of cluster areas until the sum of the spatial position distance and pixel feature distance of pixels in the cluster area no longer changes; Determine the boundary outline of the cluster area as the outline of the analysis sub-area to form multiple analysis sub-areas; S4: Identify abnormal temperature and color areas based on the analysis of temperature gradient distribution characteristics and color feature differences in the sub-areas; S5: Perform spatial neighborhood topology analysis on the temperature anomaly area and the color anomaly area to determine the propagation path of the anomaly feature and the diffusion range of the propagation path of the anomaly feature; Establish spatial neighborhood topological structures for temperature anomaly areas and color anomaly areas respectively, and determine the spatial neighboring relationship between temperature anomaly areas and color anomaly areas according to the spatial neighborhood topological structures; Determine the propagation path of abnormal features in the temperature anomaly area and the color anomaly area based on the spatial adjacent relationship; the propagation path includes a spatial area chain consisting of multiple temperature anomaly areas or multiple color anomaly areas adjacent to each other in spatial position; For the propagation path of the abnormal feature, calculate the variation range of the temperature value gradient and the variation range of the color value difference between adjacent areas in the propagation path to obtain the diffusion range of the abnormal feature on the propagation path; S6: Based on the diffusion range of the propagation path of the abnormal characteristics, the temperature abnormal area and the color abnormal area are comprehensively scored respectively; S7: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a photovoltaic module fault area, and the spatial positioning information of the fault area is output.

2. The photovoltaic fault location method based on drone according to claim 1 is characterized in that: S1, specifically: Using the thermal imaging camera and visible light camera onboard the drone, the drone will capture images of the surface area of ​​the photovoltaic module at a fixed altitude along a pre-planned flight path. During the flight, thermal radiation images and visible light images of the photovoltaic module surface are captured at fixed intervals; Marking the thermal radiation image and visible light image obtained by shooting according to the geographical location information at the time of shooting; The thermal radiation image and the visible light image are respectively associated with the corresponding geographical location coordinates and flight altitude information to obtain thermal radiation image data and visible light image data.

3. The photovoltaic fault location method based on drone according to claim 2, characterized in that: S2, specifically: Perform coordinate system conversion on thermal radiation image data and visible light image data respectively, and uniformly convert the geographical location coordinates corresponding to the time of shooting into a standard geographic space coordinate system; Perform spatial interpolation processing on thermal radiation image data and visible light image data in a standard geographic space coordinate system; Perform spatial feature matching based on the corresponding photovoltaic module edge contour features and installation position features in the thermal radiation image data and the visible light image data, and extract spatial registration feature point pairs; Based on the spatial registration feature point pairs, the coordinate transformation matrix between the thermal radiation image data and the visible light image data is calculated, and the spatial registration of the thermal radiation image data and the visible light image data is completed according to the coordinate transformation matrix; The spatially registered thermal radiation image data and visible light image data are fused at the pixel level to generate multimodal image data in a unified spatial coordinate system.

4. The photovoltaic fault location method based on drone according to claim 3 is characterized in that: S4, specifically: The temperature value corresponding to the thermal radiation image data is extracted for each pixel point in the analysis sub-area, and the spatial gradient change of the temperature values ​​of all pixels in the analysis sub-area is calculated to obtain the temperature gradient distribution characteristics of the analysis sub-area; Extract the color value corresponding to the visible light image data for each pixel point in the analysis sub-area, perform statistics on the color values ​​of the pixels in the analysis sub-area, and obtain the color distribution characteristics of the analysis sub-area; Based on the temperature gradient distribution characteristics of the analysis sub-region, determine whether the temperature gradient of the analysis sub-region is higher or lower than that of the adjacent analysis sub-region, and mark it as a temperature abnormal region; Based on the color distribution characteristics of the analysis sub-regions, the analysis sub-regions whose color values ​​deviate from the adjacent analysis sub-regions are determined and marked as color abnormal regions.

5. The photovoltaic fault location method based on drone according to claim 4, characterized in that: The color value includes the red channel value, the green channel value, and the blue channel value.

6. The photovoltaic fault location method based on drone according to claim 5, characterized in that: S6, specifically: Based on the diffusion range of the abnormal characteristics on the propagation path, the temperature value gradient change range of the temperature abnormal area and the color value difference change range of the color abnormal area are calculated respectively; The scoring value of the temperature abnormality area is set according to the temperature value gradient change range, and the scoring value of the color abnormality area is set according to the color value difference change range; Assign weight coefficients to the scoring values ​​of the temperature anomaly area and the color anomaly area respectively; The score values ​​of the temperature anomaly area and the score values ​​of the color anomaly area are weighted by weight coefficients to obtain the comprehensive scores of the temperature anomaly area and the color anomaly area.

7. The photovoltaic fault location method based on drone according to claim 6, characterized in that: S7, specifically: Setting fault level thresholds, which include fault level thresholds for temperature abnormality areas and fault level thresholds for color abnormality areas; Compare the comprehensive scores of the temperature abnormality area and the color abnormality area with the corresponding fault level threshold respectively to determine whether the comprehensive scores are higher than the corresponding fault level threshold; When the comprehensive score of the temperature abnormality area or color abnormality area is higher than the corresponding fault level threshold, the temperature abnormality area or color abnormality area higher than the corresponding fault level threshold is marked as a photovoltaic module fault area; Output is marked as the spatial positioning information of the fault area of ​​the photovoltaic module.

8. A photovoltaic fault location system based on a drone, used to implement the photovoltaic fault location method based on a drone according to any one of claims 1 to 7, characterized in that: include: Image acquisition module: uses the thermal imaging camera and visible light camera carried by the drone to collect image data of the photovoltaic module surface along the set trajectory, and obtains the corresponding thermal radiation image data and visible light image data; Image fusion module: aligns and fuses thermal radiation image data and visible light image data to generate multimodal image data in a unified spatial coordinate system; Region segmentation module: performs superpixel segmentation on multimodal image data to form multiple analysis sub-regions; Abnormal identification module: Based on the analysis of the temperature gradient distribution characteristics and color feature differences of the sub-regions, it identifies the abnormal temperature and color areas; Topological analysis module: performs spatial neighborhood topological analysis on abnormal temperature and color areas to determine the propagation path of abnormal features and the diffusion range of the propagation path of abnormal features; Scoring calculation module: Based on the diffusion range of the propagation path of the abnormal characteristics, comprehensive scores are given to the temperature abnormality area and the color abnormality area respectively; Fault marking module: When the comprehensive score exceeds the set fault level threshold, the corresponding area is marked as a PV module fault area, and the spatial positioning information of the fault area is output.

Images