Photovoltaic module hot spot detection method and system, computer equipment and storage medium

Abstract

The invention discloses a photovoltaic module hot spot detection method and system, computer equipment and a storage medium. The method comprises the following steps: carrying out cutting preprocessing on a visible light image and an infrared thermal imaging image; extracting the position information of the photovoltaic module sub-region of the cut visible light image; extracting edge feature points of the cut visible light image and the cut infrared thermal imaging image, and performing feature point matching; the feature points are resampled, a transformation matrix is calculated, and alignment of the visible light image and the infrared thermal imaging image is achieved; mapping the position information of the sub-regions of the photovoltaic module into an infrared thermal imaging image through a transformation matrix, and segmenting to obtain sub-images of the infrared thermal imaging image; training a plurality of classifiers according to the infrared thermal imaging image sub-graphs; and inputting a to-be-detected infrared thermal imaging image of the photovoltaic module into the plurality of trained classifiers to realize hot spot detection of the photovoltaic module. According to the method, whether the hot spot defect exists in the photovoltaic module or not can be accurately judged at a relatively high speed and relatively high accuracy.

Description

Technical Field

[0001] The present invention relates to a method, system, computer device and storage medium for detecting hot spots of photovoltaic modules, and belongs to the technical field of image processing. Background Art

[0002] With the rapid development of the photovoltaic industry, the quality inspection of photovoltaic modules has become particularly important. As the core component of a distributed photovoltaic power generation system, compared with traditional fossil energy, photovoltaic modules not only emit zero emissions and are pollution-free during use, but also greatly reduce the energy cost of the region, improve the energy utilization efficiency, and have been widely used due to their unique advantages. In addition, the installation of photovoltaic modules is flexible and diverse, and can adapt to complex terrain and climate conditions, such as being installed on street lamp poles, rooftops, mountain tops, etc., effectively utilizing the abundant idle space and bringing new opportunities for sustainable economic benefits and green development to most regions.

[0003] Against the backdrop of the country's active promotion of the development of renewable energy, the field of photovoltaic power generation has witnessed rapid growth. However, this development trend has also brought new challenges to the operation and maintenance of photovoltaic power stations, especially the problem of hot spot defects in photovoltaic modules has become increasingly prominent. Hot spot defects not only reduce the energy conversion efficiency of photovoltaic modules, but may also cause the local temperature of the modules to be too high. If not discovered and processed in time, it is very likely to cause serious consequences such as fires. To meet the demand for detecting hot spots of photovoltaic modules, the industry has explored various detection methods, including infrared thermal imager detection, current-voltage characteristic curve analysis, visual detection, and deep learning technology. Traditional machine vision detection methods, such as threshold processing, although can achieve hot spot detection to a certain extent, but since image acquisition is often carried out in a natural environment, the change of lighting conditions will significantly affect the detection effect, resulting in the lack of stability and reliability of such methods. Summary of the Invention

[0004] In view of this, the present invention provides a method, system, computer device and storage medium for detecting hot spots of photovoltaic modules, which can accurately determine whether there are hot spot defects in photovoltaic modules at a relatively fast speed and with a high accuracy.

[0005] The first object of the present invention is to provide a method for detecting hot spots of photovoltaic modules.

[0006] The second object of the present invention is to provide a system for detecting hot spots of photovoltaic modules.

[0007] The third object of the present invention is to provide a computer device.

[0008] The fourth object of the present invention is to provide a computer-readable storage medium.

[0009] The first object of the present invention can be achieved by adopting the following technical solutions:

[0010] A method for detecting hot spots of a photovoltaic module, the method comprising:

[0011] Obtaining a visible light image and an infrared thermal imaging image of the photovoltaic module;

[0012] Performing cropping preprocessing on the visible light image and the infrared thermal imaging image so that the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image is the same, obtaining a cropped visible light image and a cropped infrared thermal imaging image;

[0013] Extracting the position information of the photovoltaic module sub-regions of the cropped visible light image;

[0014] Based on edge detection, extracting edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and performing feature point matching;

[0015] Performing resampling on the feature points, calculating a transformation matrix, and realizing the alignment of the visible light image and the infrared thermal imaging image;

[0016] Mapping the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix, and segmenting to obtain sub-images of the infrared thermal imaging image;

[0017] Training multiple classifiers according to the sub-images of the infrared thermal imaging image to obtain multiple trained classifiers;

[0018] Inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to realize the detection of hot spots of the photovoltaic module.

[0019] Further, the extracting the position information of the photovoltaic module sub-regions of the cropped visible light image specifically includes:

[0020] Converting the color space of the cropped visible light image from RGB to HSV, and separating the natural environment background and the region of the photovoltaic module using a threshold method in the S channel;

[0021] Processing the overall region of the photovoltaic module using Gaussian filtering and opening operation to remove noise points and smooth the boundary;

[0022] Using connected regions to label each sub-region of the photovoltaic module and calculating the rectangular similarity characteristics to screen out the sub-regions belonging to the photovoltaic module, thereby obtaining the position and region information of several complete sub-regions of the photovoltaic module in the cropped visible light image.

[0023] Further, the based on edge detection, extracting edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and performing feature point matching specifically includes:

[0024] Edge extraction is performed on the cropped visible light image and the cropped infrared thermal imaging image to obtain the edge information of the visible light image and the edge information of the infrared thermal imaging image;

[0025] Calculate the grayscale histograms of the edge information of the visible light image and the edge information of the infrared thermal imaging image, and statistically calculate the maximum frequency grayscale value μ and the grayscale standard deviation σ in the grayscale histogram. Use the 3σ principle to extract the points with grayscale values in the range of μ + 2σ to 255 in the grayscale histogram as edge feature points;

[0026] Grayscale the visible light image and the infrared thermal imaging image, and use the grayscale matching method of normalized cross-correlation to calculate the similarity between the feature points of the visible light image and the infrared thermal imaging image, and determine the corresponding relationship of the feature points on the visible light image and the infrared thermal imaging image.

[0027] Further, resampling the feature points and calculating the transformation matrix to align the visible light image and the infrared thermal imaging image specifically include:

[0028] Resample the feature points, and eliminate abnormal feature points through multiple random sampling and verification processes;

[0029] According to the resampled feature points, calculate the transformation matrix and perform an affine transformation on the grayscaled visible light image to align the visible light image and the infrared thermal imaging image.

[0030] Further, mapping the position information of the photovoltaic module sub-region into the infrared thermal imaging image through the transformation matrix and segmenting to obtain sub-images of the infrared thermal imaging image specifically include:

[0031] Map the position information of the photovoltaic module sub-region into the infrared thermal imaging image through the transformation matrix, and calculate the intersection of each sub-region and the infrared thermal imaging image respectively to obtain several sub-images of the infrared thermal imaging image segmented according to the region position information.

[0032] Further, training multiple classifiers based on the sub-images of the infrared thermal imaging image to obtain multiple trained classifiers specifically include:

[0033] According to the sub-images of the infrared thermal imaging image, divide the training set, and set the label of the photovoltaic module sub-region with hot spot defects to 0, otherwise 1;

[0034] Calculate multiple features of the images in the training set, input them into multiple classifiers for training, and obtain multiple trained classifiers.

[0035] Further, inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to implement hot spot detection of the photovoltaic module specifically includes:

[0036] Input the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers, and use a voting strategy to determine whether there are hot spot defects in the photovoltaic module;

[0037] If the number of classifiers that determine the existence of hot spot defects is more than the number of classifiers that determine the non-existence of hot plate defects, it is determined that there are hot spot defects in the sub-region of the photovoltaic module;

[0038] For the sub-region with hot spot defects, calculate the actual position of the sub-region in the infrared thermal imaging image to be measured, and make a frame selection mark to indicate the existence of hot spot defects.

[0039] The second object of the present invention can be achieved by adopting the following technical solutions:

[0040] A hot spot detection system for a photovoltaic module, the system includes:

[0041] An acquisition module, configured to acquire a visible light image and an infrared thermal imaging image of the photovoltaic module;

[0042] A cropping module, configured to perform cropping preprocessing on the visible light image and the infrared thermal imaging image, so that the number of complete sub-regions of the photovoltaic module in the visible light image and the infrared thermal imaging image is the same, and obtain a cropped visible light image and a cropped infrared thermal imaging image;

[0043] A first extraction module, configured to extract the position information of the sub-region of the photovoltaic module in the cropped visible light image;

[0044] A second extraction module, configured to extract edge feature points of the cropped visible light image and the cropped infrared thermal imaging image based on edge detection, and perform feature point matching;

[0045] A resampling module, configured to resample the feature points, calculate a transformation matrix, and realize the alignment of the visible light image and the infrared thermal imaging image;

[0046] A mapping module, configured to map the position information of the sub-region of the photovoltaic module into the infrared thermal imaging image through the transformation matrix, and segment the infrared thermal imaging image into sub-images;

[0047] A training module, configured to train multiple classifiers according to the sub-images of the infrared thermal imaging image, and obtain multiple trained classifiers;

[0048] A detection module, configured to input the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to realize the hot spot detection of the photovoltaic module.

[0049] The third object of the present invention can be achieved by adopting the following technical solutions:

[0050] A computer device includes a processor and a memory for storing programs executable by the processor. When the processor executes the programs stored in the memory, the above-mentioned photovoltaic module hot spot detection method is implemented.

[0051] The fourth object of the present invention can be achieved by adopting the following technical solutions:

[0052] A computer-readable storage medium stores a program, and when the program is executed by a processor, the above-mentioned photovoltaic module hot spot detection method is implemented.

[0053] The present invention has the following beneficial effects compared with the prior art:

[0054] Based on machine vision detection technology, the present invention introduces a multi-source information matching strategy, aiming to make full use of the rich texture and color information in visible light images to make up for the deficiency of infrared thermal imaging images in detail resolution, thereby improving the robustness of the method. In the process of extracting image matching points, an adaptive edge threshold algorithm and a feature point sampling algorithm are combined to find the projective transformation matrix to achieve more accurate image information matching; at the same time, the present invention proposes a multi-classifier voting mechanism, which can reduce the error rate, overfitting risk of a single classifier and the dependence on specific data or features, effectively reduce the misdetection situation, and thus significantly improve the accuracy and reliability of hot spot detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on the structures shown in these drawings without creative efforts.

[0056] Figure 1 It is a flowchart of the photovoltaic module hot spot detection method according to Embodiment 1 of the present invention.

[0057] Figure 2a It is a schematic diagram of the visible light image after cropping and preprocessing in Embodiment 1 of the present invention.

[0058] Figure 2b It is a schematic diagram of the infrared thermal imaging image after cropping and preprocessing in Embodiment 1 of the present invention.

[0059] Figure 3 It is a diagram of the position information of the photovoltaic module sub-region for extracting the visible light image in Embodiment 1 of the present invention.

[0060] Figure 4a It is a schematic diagram of the gradient magnitude image in Embodiment 1 of the present invention.

[0061] Figure 4b Schematic diagram of edge feature points in Embodiment 1 of the present invention.

[0062] Figure 5 Schematic diagram of the result of feature edge matching in Embodiment 1 of the present invention.

[0063] Figure 6 Schematic diagram of the result of feature point resampling in Embodiment 1 of the present invention.

[0064] Figure 7 Schematic diagram of an infrared thermal imaging image for matching the position information of the visible photon region in Embodiment 1 of the present invention.

[0065] Figure 8 Schematic diagram of the hot spot detection result in Embodiment 1 of the present invention.

[0066] Figure 9 Block diagram of the photovoltaic module hot spot detection system according to Embodiment 2 of the present invention.

[0067] Figure 10 Block diagram of the computer device according to Embodiment 3 of the present invention. Detailed implementation manners

[0068] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0069] Embodiment 1:

[0070] As Figure 1 shown, this embodiment provides a method for detecting hot spots in a photovoltaic module, and the method includes the following steps:

[0071] S101. Obtain a visible light image and an infrared thermal imaging image of the photovoltaic module.

[0072] In one embodiment, images of the photovoltaic module are respectively captured by a visible light camera and an infrared light camera to obtain a visible light image and an infrared thermal imaging image of the photovoltaic module.

[0073] S102. Perform cropping preprocessing on the visible light image and the infrared thermal imaging image to make the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image the same, and obtain a cropped visible light image and a cropped infrared thermal imaging image.

[0074] In one embodiment, since there are differences in the field of view when the visible light camera and the infrared light camera capture images, which results in the problem that the number of photovoltaic panels in the visible light image and the infrared thermal imaging image is inconsistent. This may cause significant errors in the subsequent image alignment step, thereby having a greater negative impact on the effect of hot spot detection and leading to a decrease in detection accuracy. Therefore, preprocessing by cropping the obtained visible light image and infrared thermal imaging image is performed. The cropped and preprocessed images are as Figures 2a to 2b shown, ensuring that the number of complete photovoltaic component sub-regions in the images is the same and making the two images have better correspondence in content.

[0075] S103. Extract the position information of the photovoltaic component sub-regions of the cropped visible light image.

[0076] In one embodiment, the color space of the cropped visible light image is converted from RGB to HSV, and the threshold method is used in the S channel to separate the natural environment background and the region of the photovoltaic component; Gaussian filtering and opening operation are used to process the overall region of the photovoltaic component to remove noise and smooth the boundaries; connected regions are used to label each sub-region of the photovoltaic component and the rectangular similarity characteristics are calculated to screen out the sub-regions belonging to the photovoltaic component, thereby obtaining the position and region information of several complete sub-regions of the photovoltaic component in the cropped visible light image. The extraction effect is as Figure 3 shown.

[0077] In one embodiment, the rectangular similarity is calculated to accurately extract the region belonging to the photovoltaic component and obtain the position information of the photovoltaic component region of the visible light image, as shown in the following formula:

[0078]

[0079] where R is the rectangular similarity, S0 is the area of the region, and S1 is the area of the minimum bounding rectangle.

[0080] S104. Based on edge detection, extract the edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and perform feature point matching.

[0081] In one embodiment, the process of extracting the edge feature points of the cropped visible light image and the cropped infrared thermal imaging image is as follows: Edge extraction is performed on the cropped visible light image and the cropped infrared thermal imaging image to obtain the visible light image edge information and the infrared thermal imaging image edge information; The gray level histograms of the visible light image edge information and the infrared thermal imaging image edge information are calculated, as Figure 4a shown. The maximum frequency gray level value μ and the gray standard deviation σ in the gray level histogram are statistically analyzed, and the points with gray level values in the gray level histogram between μ + 2σ and 255 are extracted as edge feature points according to the 3σ principle, as Figure 4b shown.

[0082] In one embodiment, the specific steps of extracting edge feature points of the cropped visible light image and the cropped infrared thermal imaging image are as follows:

[0083] S1041, edge extraction is performed on the cropped visible light image and the cropped infrared thermal imaging image. The edge extraction calculation formula is as follows:

[0084] G x =G1*A,G y =G2*A

[0085] G=(thin(|G x |)+thin(|G y |)) / 4

[0086] Among them, G1 is the horizontal convolution kernel, G2 is the vertical convolution kernel, and G x and G y are horizontal and vertical images respectively. The thinning calculation thin(x) means that for a given image gradient, a specific mask is applied to identify the vertical maximum value and the horizontal maximum value, and these maximum values are retained as the original values, and other values are set to 0.

[0087] S1042, calculating the grayscale histogram of the visible light image edge information and the infrared thermal imaging image edge information, the grayscale calculation formula is as follows:

[0088] Gray=0.299*r+0.587*g+0.114*b

[0089] Among them, Gray is the grayscale value of the grayscale image, r, g, and b are the pixel values of the red channel, green channel, and blue channel of the RGB image respectively.

[0090] S1043. When extracting feature points, the grayscale value distribution of the grayscale histogram is approximated to a Gaussian distribution. The 3σ principle in statistics is used to eliminate most of the points in the grayscale histogram, and a very small number of points with obvious edge features are retained. The maximum frequency grayscale value μ and the grayscale standard deviation σ of the grayscale histogram are statistically calculated. The probability calculation formula for the pixel value x appearing in the image is as follows:

[0091] P(|x-μ|>3σ)≤0.003

[0092] Among them, P(x) is the probability that the pixel value x appears in the image. The demarcation threshold for extracting the grayscale histogram is calculated to be μ+2σ~255. The points in the grayscale histogram with grayscale values in the range of μ+2σ~255 are the characteristic edge points of the visible light image and the infrared thermal imaging image.

[0093] In one embodiment, in order to eliminate the brightness and contrast differences between the visible light image and the infrared thermal imaging image, reduce the influence of noise and interference, improve the accuracy of matching, and accelerate the matching efficiency, the feature point matching process is as follows: The visible light image and the infrared thermal imaging image are grayscaled, and the normalized cross-correlation grayscale matching method is used to calculate the similarity between the feature points of the visible light image and the infrared thermal imaging image, and the corresponding relationship of the feature points on the visible light image and the infrared thermal imaging image is determined. The result of the feature point matching is as Figure 5 shown.

[0094] In one embodiment, the specific steps of the feature point matching are as follows:

[0095] S1044. Gray-scale the visible light image and the infrared thermal imaging image, and calculate the corresponding normalized images V and I of the grayscaled visible light image and the grayscaled infrared thermal imaging image. The image normalization formula is as follows:

[0096]

[0097] where G VMax , G VMin are respectively the maximum and minimum values of the pixels of the grayscaled visible light image, and G IMax , G IMin are respectively the maximum and minimum values of the pixels of the grayscaled infrared thermal imaging image.

[0098] S1045. Calculate the correlation degree NCC(x, y) of the gray values of the surrounding mask windows of the input points in image V and image I. The calculation formula is as follows:

[0099]

[0100] where V(x, y) and I(x, y) are respectively the pixel values of the surrounding mask windows in image V and image I, are respectively the average pixel values of the pixel values of the surrounding mask windows in image V and image I.

[0101] S105. Resample the feature points, calculate the transformation matrix, and realize the alignment of the visible light image and the infrared thermal imaging image.

[0102] In one embodiment, for resampling the feature points, through multiple random sampling and verification processes, abnormal feature points are eliminated. By sampling processing, the number of feature points to be processed is reduced, the calculation complexity of the transformation matrix is reduced, the running efficiency of the overall algorithm is improved, and at the same time, the mis-matched points are eliminated, and the robustness of the feature point matching is improved. The result of the feature point resampling is as Figure 6 shown; According to the resampled feature points, calculate the transformation matrix, and perform an affine transformation on the grayscaled visible light image to realize the alignment of the visible light image and the infrared thermal imaging image.

[0103] In one embodiment, the specific process of step S105 is as follows:

[0104] S1051. Randomly select four groups of corresponding feature points from the set of feature points obtained from the visible light image and the infrared thermal image, where (x, y) represents the feature point in image V, and (x′, y′) represents the feature point in image I. Let h33 = 1 for the normalization matrix, and calculate the 3×3 optimal homography matrix H. The calculation formula for the optimal homography matrix H is as follows:

[0105]

[0106] For each point pair in the set of feature points, calculate the projection error between the position after transformation by the preliminarily estimated homography matrix H and the actual position. If the projection error is less than the set threshold, then consider this point pair as an inlier.

[0107] S1052. Continuously iterate and repeat step S1051, and select the model with the most inliers in the iteration times as the final model, and record the new set of feature inliers S V =(X, Y) and S I =(X′, Y′).

[0108] S1053. Calculate the affine transformation matrix Trans between the point sets S V , S I . Xi and Yi represent the horizontal and vertical coordinates of the i-th index in the point set SV, Xi′ and Yi′ represent the horizontal and vertical coordinates of the i-th index in the point set SI, and M represents minimizing the distance between the correspondence of (Xi, Yi) and the transformed point (X′, Y′), and then determine an optimal rigid body affine transformation matrix T, as shown in the following formula:

[0109]

[0110] S1054. Perform an affine transformation on the grayscale visible light image to achieve the alignment of the visible light image and the infrared thermal image, as shown in the following formula:

[0111]

[0112] Among them, X and Y represent the horizontal and vertical coordinates of the visible light image, and X trans , Y trans represent the coordinates after the affine transformation.

[0113] S106. Map the position information of the photovoltaic module sub-region into the infrared thermal image through the transformation matrix, and segment the sub-image of the infrared thermal image.

[0114] In one embodiment, map the position information of the photovoltaic module sub-region into the infrared thermal image through the transformation matrix, such asFigure 7 As shown, the matching of multi-source information is realized; the intersections of each sub-region and the infrared thermal imaging image are calculated respectively to obtain several sub-images of the infrared thermal imaging image segmented according to the regional position information, and they are numbered sequentially starting from zero.

[0115] S107. Train multiple classifiers based on the sub-images of the infrared thermal imaging image to obtain multiple trained classifiers.

[0116] In one embodiment, according to the sub-images of the infrared thermal imaging image, 80% is randomly selected as the training set, and the rest is the test set. The label of the sub-region of the photovoltaic module with hot spot defects is set to 0, otherwise it is 1; calculate multiple features of the images in the training set, input them into multiple classifiers for training, and obtain multiple trained classifiers.

[0117] S108. Input the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to realize the hot spot detection of the photovoltaic module.

[0118] In one embodiment, the test set is used as the infrared thermal imaging image to be measured of the photovoltaic module and input into multiple trained classifiers, and a voting strategy is used to judge whether there are spot defects in the photovoltaic module; if the number of classifiers that judge the existence of hot spot defects is more than the number of classifiers that judge the non-existence of hot plate defects, it is determined that there are hot spot defects in the sub-region of the photovoltaic module; for the sub-region with hot spot defects, return the label of the corresponding sub-image of the infrared thermal imaging image of this sub-region, calculate the actual position of this sub-region in the infrared thermal imaging image to be measured (the corresponding sub-image of the infrared thermal imaging image in the test set), and make a frame selection mark to indicate the existence of hot spot defects. The hot spot detection result is as Figure 8 shown.

[0119] It should be noted that although the method operations of the above embodiments are described in a specific order, this does not require or imply that these operations must be performed in this specific order, or that all the shown operations must be performed to achieve the desired result. On the contrary, the described steps can be changed in the execution order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step for execution, and / or one step can be decomposed into multiple steps for execution.

[0120] Embodiment 2:

[0121] As Figure 9 shown, this embodiment provides a hot spot detection system for a photovoltaic module. The system includes an acquisition module 901, a cropping module 902, a first extraction module 903, a second extraction module 904, a resampling module 905, a mapping module 906, a training module 907, and a detection module 908. The specific descriptions of each module are as follows:

[0122] An acquisition module 901, configured to acquire a visible light image and an infrared thermal imaging image of a photovoltaic module;

[0123] A cropping module 902, configured to perform cropping preprocessing on the visible light image and the infrared thermal imaging image, so that the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image is the same, and obtain a cropped visible light image and a cropped infrared thermal imaging image;

[0124] A first extraction module 903, configured to extract the position information of the photovoltaic module sub-region of the cropped visible light image;

[0125] A second extraction module 904, configured to extract edge feature points of the cropped visible light image and the cropped infrared thermal imaging image based on edge detection, and perform feature point matching;

[0126] A resampling module 905, configured to resample the feature points, calculate a transformation matrix, and realize the alignment of the visible light image and the infrared thermal imaging image;

[0127] A mapping module 906, configured to map the position information of the photovoltaic module sub-region into the infrared thermal imaging image through the transformation matrix, and segment the infrared thermal imaging image sub-graph;

[0128] A training module 907, configured to train multiple classifiers according to the infrared thermal imaging image sub-graph, and obtain multiple trained classifiers;

[0129] A detection module 908, configured to input the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to realize hot spot detection of the photovoltaic module.

[0130] It should be noted that the system provided in this embodiment is only illustrated by the above division of each functional module. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure is divided into different functional modules to complete all or part of the functions described above.

[0131] It can be understood that the terms "first", "second", etc. used in the above system can be used to describe various modules, but these modules are not limited by these terms. These terms are only used to distinguish the first module from another module. For example, without departing from the scope of the present invention, the first extraction module can be called the second extraction module, and similarly, the second extraction module can be called the first extraction module. Both the first extraction module and the second extraction module are data processing modules, but not the same data processing module.

[0132] Embodiment 3:

[0133] This embodiment provides a computer device, such as Figure 10As shown in the figure, it includes a processor 1002, a memory, an input device 1003, a display device 1004, and a network interface 1005 connected through a device bus 1001. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium 1006 and an internal memory 1007. The non-volatile storage medium 1006 stores an operating system, a computer program, and a database. The internal memory 1007 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the processor 1002 executes the computer program stored in the memory, the photovoltaic module hot spot detection method of the above-mentioned Embodiment 1 is implemented as follows:

[0134] Obtain the visible light image and the infrared thermal imaging image of the photovoltaic module; perform cropping preprocessing on the visible light image and the infrared thermal imaging image so that the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image is the same, and obtain the cropped visible light image and the cropped infrared thermal imaging image; extract the position information of the photovoltaic module sub-regions in the cropped visible light image; based on edge detection, extract the edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and perform feature point matching; resample the feature points, calculate the transformation matrix, and realize the alignment of the visible light image and the infrared thermal imaging image; map the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix, and segment to obtain the infrared thermal imaging image sub-map; according to the infrared thermal imaging image sub-map, train multiple classifiers to obtain multiple trained classifiers; input the infrared thermal imaging image to be measured of the photovoltaic module into the multiple trained classifiers to realize the hot spot detection of the photovoltaic module.

[0135] Embodiment 4:

[0136] This embodiment provides a computer-readable storage medium that stores a computer program. When the computer program is executed by a processor, the photovoltaic module hot spot detection method of the above-mentioned Embodiment 1 is implemented as follows:

[0137] Obtain the visible light image and infrared thermal imaging image of the photovoltaic module; perform cropping preprocessing on the visible light image and infrared thermal imaging image to make the number of complete photovoltaic module sub-regions in the visible light image and infrared thermal imaging image the same, and obtain the cropped visible light image and cropped infrared thermal imaging image; extract the position information of the photovoltaic module sub-regions in the cropped visible light image; based on edge detection, extract the edge feature points of the cropped visible light image and cropped infrared thermal imaging image, and perform feature point matching; resample the feature points, calculate the transformation matrix, and realize the alignment of the visible light image and infrared thermal imaging image; map the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix, and segment to obtain the sub-images of the infrared thermal imaging image; according to the sub-images of the infrared thermal imaging image, train multiple classifiers to obtain multiple trained classifiers; input the infrared thermal imaging image to be measured of the photovoltaic module into the multiple trained classifiers to realize the hot spot detection of the photovoltaic module.

[0138] It should be noted that the computer-readable storage medium in this embodiment can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0139] In this embodiment, the computer-readable storage medium can be any tangible medium that includes or stores a program, and this program can be used by or in combination with an instruction execution system, apparatus, or device. And in this embodiment, the computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable storage medium other than the computer-readable storage medium, and this computer-readable signal medium can send, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. The computer program included on the computer-readable storage medium can be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0140] The above computer-readable storage medium can be written in one or more programming languages or combinations thereof for executing the computer program of this embodiment. The above programming languages include object-oriented programming languages such as Java, Python, C++, and also include conventional procedural programming languages such as C language or similar programming languages. The program can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0141] In summary, the present invention is based on machine vision detection technology and introduces a multi-source information matching strategy, aiming to make full use of the rich texture and color information in visible light images to make up for the deficiency of infrared thermal imaging images in detail resolution, thereby improving the robustness of the method. In the process of extracting image matching points, an adaptive edge threshold algorithm and a feature point sampling algorithm are combined to find the projective transformation matrix to achieve more accurate image information matching. At the same time, the present invention proposes a multi-classifier voting mechanism, which can reduce the error rate, overfitting risk of a single classifier, and dependence on specific data or features, effectively reduce the misdetection situation, and thus significantly improve the accuracy and reliability of hot spot detection.

[0142] The above is only a preferred embodiment of the present invention for patents, but the protection scope of the present invention for patents is not limited thereto. Any person skilled in the art within the scope disclosed by the present invention for patents, according to the technical solution of the present invention for patents and its inventive concept, makes equivalent substitutions or changes, all belong to the protection scope of the present invention for patents.

Claims

1. A method for detecting hot spots of a photovoltaic module, characterized in that, The method includes: Obtaining a visible light image and an infrared thermal imaging image of a photovoltaic module; Performing cropping preprocessing on the visible light image and the infrared thermal imaging image to make the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image the same, obtaining a cropped visible light image and a cropped infrared thermal imaging image; Extracting the position information of the photovoltaic module sub-regions in the cropped visible light image; Based on edge detection, extracting edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and performing feature point matching; Resampling the feature points, calculating a transformation matrix, and realizing the alignment of the visible light image and the infrared thermal imaging image; Mapping the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix, and segmenting to obtain sub-images of the infrared thermal imaging image; Training multiple classifiers according to the sub-images of the infrared thermal imaging image to obtain multiple trained classifiers; Inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to realize hot spot detection of the photovoltaic module.

2. The method for detecting hot spots of a photovoltaic module according to claim 1, characterized in that The extracting the position information of the photovoltaic module sub-regions in the cropped visible light image specifically includes: Converting the color space of the cropped visible light image from RGB to HSV, and using a threshold method in the S channel to separate the natural environment background and the region of the photovoltaic module; Processing the overall region of the photovoltaic module using Gaussian filtering and opening operation to remove noise points and smooth the boundaries; Using connected region labeling to label each sub-region of the photovoltaic module and calculating the rectangular similarity characteristics to screen out the sub-regions belonging to the photovoltaic module, thereby obtaining the position and region information of several complete sub-regions of the photovoltaic module in the cropped visible light image.

3. The method for detecting hot spots of a photovoltaic module according to claim 1, characterized in that, The based on edge detection, extracting edge feature points of the cropped visible light image and the cropped infrared thermal imaging image, and performing feature point matching specifically includes: Performing edge extraction on the cropped visible light image and the cropped infrared thermal imaging image to obtain visible light image edge information and infrared thermal imaging image edge information; Calculating the gray level histograms of the visible light image edge information and the infrared thermal imaging image edge information, statistically analyzing the maximum frequency gray level value μ and the gray level standard deviation σ in the gray level histograms, and using the 3σ principle to extract the points with gray level histogram gray level values between μ + 2σ and 255 as edge feature points; Graying the visible light image and the infrared thermal imaging image, and using the gray level matching method of normalized cross-correlation to calculate the similarity between the feature points of the visible light image and the infrared thermal imaging image, and determining the corresponding relationship of the feature points on the visible light image and the infrared thermal imaging image.

4. The method for detecting hot spots of a photovoltaic module according to claim 3, characterized in that The resampling the feature points, calculating a transformation matrix, and realizing the alignment of the visible light image and the infrared thermal imaging image specifically includes: Resampling the feature points, and removing abnormal feature points through multiple random sampling and verification processes; According to the resampled feature points, calculating a transformation matrix, and performing affine transformation on the grayed visible light image to realize the alignment of the visible light image and the infrared thermal imaging image.

5. The method for detecting hot spots of a photovoltaic module according to claim 1, wherein The mapping the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix, and segmenting to obtain sub-images of the infrared thermal imaging image specifically includes: Map the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through a transformation matrix, and calculate the intersection of each sub-region and the infrared thermal imaging image respectively to obtain several infrared thermal imaging image sub-graphs segmented according to the regional position information.

6. The method for detecting hot spots of a photovoltaic module according to claim 1, characterized in that, Training multiple classifiers based on the infrared thermal imaging image sub-graphs to obtain multiple trained classifiers specifically includes: Dividing a training set according to the infrared thermal imaging image sub-graphs, setting the label of the photovoltaic module sub-region with a hot spot defect to 0, and 1 otherwise; Calculating multiple features of the images in the training set and inputting them into multiple classifiers for training to obtain multiple trained classifiers.

7. The method for detecting hot spots of a photovoltaic module according to claim 1, characterized in that, Inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to implement hot spot detection of the photovoltaic module, specifically including: Inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers and judging whether there are spot defects in the photovoltaic module by a voting strategy; If the number of classifiers judging the existence of hot spot defects is more than the number of classifiers judging the non-existence of hot plate defects, it is determined that there are hot spot defects in the photovoltaic module sub-region; For the sub-region with a hot spot defect, calculate the actual position of the sub-region in the infrared thermal imaging image to be measured and make a bounding box mark to indicate the existence of a hot spot defect.

8. A photovoltaic module hot spot detection system, characterized in that, The system includes: An acquisition module for acquiring the visible light image and the infrared thermal imaging image of the photovoltaic module; A cropping module for preprocessing the visible light image and the infrared thermal imaging image by cropping to make the number of complete photovoltaic module sub-regions in the visible light image and the infrared thermal imaging image the same, obtaining a cropped visible light image and a cropped infrared thermal imaging image; A first extraction module for extracting the position information of the photovoltaic module sub-regions of the cropped visible light image; A second extraction module for extracting the edge feature points of the cropped visible light image and the cropped infrared thermal imaging image based on edge detection and performing feature point matching; A resampling module for resampling the feature points, calculating the transformation matrix, and realizing the alignment of the visible light image and the infrared thermal imaging image; A mapping module for mapping the position information of the photovoltaic module sub-regions into the infrared thermal imaging image through the transformation matrix and segmenting to obtain infrared thermal imaging image sub-graphs; A training module for training multiple classifiers according to the infrared thermal imaging image sub-graphs to obtain multiple trained classifiers; A detection module for inputting the infrared thermal imaging image to be measured of the photovoltaic module into multiple trained classifiers to implement hot spot detection of the photovoltaic module.

9. A computer device, comprising a processor and a memory for storing processor-executable programs, characterized in that, When the processor executes the program stored in the memory, it realizes the photovoltaic module hot spot detection method according to any one of claims 1-7.

10. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it realizes the photovoltaic module hot spot detection method according to any one of claims 1-7.

Images