Anomaly detection method, device and equipment of photovoltaic panel and medium

Abstract

The invention provides an anomaly detection method, device and equipment for a photovoltaic panel and a medium, and effectively solves the problem that a traditional anomaly detection method for the photovoltaic panel has defects. The method comprises the following steps: processing an obtained original infrared image to obtain position information of a string image in the corresponding original infrared image; filtering incomplete group string images in the original infrared image, retaining the complete group string image, judging the position state of the complete group string image based on the position information, and changing the position state of the complete group string image; segmenting the complete string image after the position state is changed to obtain a plurality of photovoltaic panel images, and classifying the photovoltaic panel images through a plurality of classification dimensions to obtain a classification result; and determining that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and positioning an abnormal photovoltaic panel corresponding to the original infrared image based on an identifier.

Description

Technical Field

[0001] This application relates to the technical field of photovoltaic energy monitoring and image processing. Specifically, it relates to a method, device, equipment and medium for detecting anomalies in photovoltaic panels. Background Art

[0002] As an important part of green energy, the stable operation of photovoltaic energy is of great significance for ensuring energy supply. However, during the actual operation of photovoltaic panels, their performance may decline or they may be damaged due to various factors such as hot spot effects and diode failures, resulting in anomalies in the photovoltaic panels. In response to this situation, various methods for detecting anomalies in photovoltaic panels have gradually emerged.

[0003] Traditional anomaly detection methods rely on manual inspections, which are not only inefficient but also difficult to detect and locate anomalies in a timely manner. Moreover, since the specific anomaly manifestations of hot spot effects and diode failures are different, different anomaly detection methods need to be used to detect them. Therefore, it is particularly important to develop an intelligent detection method that can detect multiple anomalies and locate the anomaly status of photovoltaic panels. Summary of the Invention

[0004] In view of this, the purpose of this application is to provide a method, device, equipment and medium for detecting anomalies in photovoltaic panels, which effectively solves the problem of defects in traditional methods for detecting anomalies in photovoltaic panels.

[0005] In a first aspect, an embodiment of this application provides a method for detecting anomalies in photovoltaic panels. The method includes: Processing multiple original infrared images of multiple photovoltaic panel strings obtained through a pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangular; Filter out the incomplete string images in the original infrared images, retain the complete string images, determine the position status of the complete string images based on the position information, and change the position status of the complete string images; Segment the complete string images with the changed position status to obtain multiple photovoltaic panel images, and classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the classification result includes the identifier of the photovoltaic panel image; Determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and locate the abnormal photovoltaic panel corresponding to the original infrared image based on the identifier of the abnormal photovoltaic panel image.

[0006] In combination with the first aspect, the embodiments of the present application provide a first possible implementation manner of the first aspect, wherein determining the position state of the complete string image based on the position information and changing the position state of the complete string image includes: Determining whether the complete string image is in a horizontal position state based on the position information; If not, determining that the complete string image is in a rotated position state and changing the rotated position state.

[0007] In combination with the first aspect, the embodiments of the present application provide a second possible implementation manner of the first aspect, wherein changing the rotated position state includes: Determining the included angle between the long side of the complete string image and the horizontal line based on the position information; Moving the complete string image in the reverse direction of the included angle to the horizontal position state to complete the change of the position state.

[0008] In combination with the first aspect, the embodiments of the present application provide a third possible implementation manner of the first aspect, wherein filtering out the incomplete string images in the original infrared image and retaining the complete string images includes: Judging whether the string image meets the filtering condition based on the position information of the string image; If so, determining that the string image is an incomplete string image and filtering out the incomplete string image.

[0009] In combination with the first aspect, the embodiments of the present application provide a fourth possible implementation manner of the first aspect, wherein classifying the photovoltaic panel image through multiple classification dimensions to obtain a classification result includes: Extracting the topological structure features and color features in the photovoltaic panel image and mapping the topological structure features and color features into multiple high-dimensional features; Classifying the multiple high-dimensional features to obtain multiple classification dimensions for classifying the photovoltaic panel image.

[0010] In combination with the first aspect, the embodiments of the present application provide a fifth possible implementation manner of the first aspect, wherein the method further includes: If it is determined that the photovoltaic panel image does not have a topological structure and the corresponding color feature does not match the color anomaly condition; Then determining that the classification result of the photovoltaic panel image is normal.

[0011] In combination with the first aspect, the embodiments of the present application provide a sixth possible implementation manner of the first aspect, wherein processing multiple original infrared images of multiple groups of photovoltaic panel strings obtained by a pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images includes: Establish a coordinate system and determine the initial coordinates of each vertex of the string image based on the coordinate system; Normalize the initial coordinates to obtain the position information of each vertex of the string image.

[0012] In a second aspect, an embodiment of the present application provides an abnormal detection device for a photovoltaic panel, and the device includes: A processing module, configured to process multiple original infrared images of multiple photovoltaic panel strings obtained through a pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangles; A determination module, configured to filter out the incomplete string images in the original infrared images, retain the complete string images, determine the position state of the complete string images based on the position information, and change the position state of the complete string images; A segmentation module, configured to segment the complete string images with the changed position state to obtain multiple photovoltaic panel images, and classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the classification result includes the identifier of the photovoltaic panel image; A positioning module, configured to determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and position the abnormal photovoltaic panel corresponding to the original infrared image based on the identifier of the abnormal photovoltaic panel image.

[0013] In a third aspect, an embodiment of the present application provides an electronic device, including: a processor, a memory, and a bus, where the memory stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the memory through the bus, and when the machine-readable instructions are executed by the processor, the steps of any one of the abnormal detection methods for a photovoltaic panel are executed.

[0014] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is run by a processor, the steps of any one of the abnormal detection methods for a photovoltaic panel are executed.

[0015] An abnormal detection method for a photovoltaic panel provided by an embodiment of the present application. First, the method processes multiple original infrared images of multiple photovoltaic panel strings obtained through a pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangular; secondly, filter out the incomplete string images in the original infrared images, retain the complete string images, determine the position state of the complete string images based on the position information, and change the position state of the complete string images; then segment the complete string images with the changed position state to obtain multiple photovoltaic panel images, and classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the classification result includes the identifier of the photovoltaic panel image; finally, determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and locate the abnormal photovoltaic panel corresponding to the original infrared image based on the identifier of the abnormal photovoltaic panel image, achieving the effect of accurately detecting and classifying the abnormal state of the photovoltaic panel, reducing the false alarm rate and the missed alarm rate, and thus realizing the detection and positioning of various abnormalities of the photovoltaic panel, effectively solving the problem that the traditional abnormal detection method for the photovoltaic panel has defects, increasing the accuracy and effectiveness of the abnormal detection method for the photovoltaic panel, improving the convenience of the abnormal detection method for the photovoltaic panel, reducing the detection workload, realizing the automatic positioning of the abnormal state of the photovoltaic panel, reducing the number and cost of manual inspections, and reducing the operation and maintenance cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other relevant drawings can also be obtained based on these drawings.

[0017] Figure 1 The flowchart of the first abnormal detection method for a photovoltaic panel provided by an embodiment of the present application is shown; Figure 2 The schematic diagram of the complete string image with the rotated position state provided by an embodiment of the present application is shown; Figure 3 The schematic diagram of the incomplete string image provided by an embodiment of the present application is shown; Figure 4 The schematic diagram of the structure of the first abnormal detection device for a photovoltaic panel provided by an embodiment of the present application is shown; Figure 5 The schematic diagram of the structure of the first electronic device provided by an embodiment of the present application is shown. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. It should be understood that the accompanying drawings in this application are only for the purposes of illustration and description, and are not used to limit the protection scope of this application. Additionally, it should be understood that the schematic drawings are not drawn to the actual scale. The flowcharts used in this application illustrate the operations implemented according to some embodiments of this application. It should be understood that the operations in the flowchart may not be implemented in sequence, and steps without a logical context relationship may be reversed or implemented simultaneously. Moreover, those skilled in the art can add one or more other operations to the flowchart or remove one or more operations from the flowchart under the guidance of the content of this application.

[0019] In addition, the described embodiments are only some of the embodiments of this application, rather than all of the embodiments. The components of the embodiments of this application usually described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of this application claimed, but only represents the selected embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of this application.

[0020] It should be noted that the term "including" will be used in the embodiments of this application to indicate the existence of the features stated thereafter, but does not exclude the addition of other features.

[0021] Traditional anomaly detection methods rely on manual inspections, which are not only inefficient but also difficult to detect and locate anomalies in a timely manner. Moreover, since the specific anomaly manifestations of the hot spot effect and diode failures are different, different anomaly detection methods need to be used to detect the hot spot effect and diode failures. Therefore, it is necessary to design an intelligent detection method that can detect multiple anomalies and locate the abnormal state of photovoltaic panels.

[0022] Based on this, the embodiments of this application provide an anomaly detection method, device, equipment, and medium for photovoltaic panels, which will be described below through embodiments.

[0023] Embodiment 1 To facilitate the understanding of this embodiment, first, a detailed introduction will be given to an anomaly detection method for photovoltaic panels disclosed in the embodiments of this application. As Figure 1 shown in the flowchart of an anomaly detection method for photovoltaic panels, an anomaly detection method for photovoltaic panels provided by this application includes: S101. Process multiple original infrared images of multiple photovoltaic panel strings obtained, using a pre-trained string detection model, to obtain the position information of the string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangular; S102. Filter out the incomplete string images in the original infrared images, retain the complete string images, determine the position status of the complete string images based on the position information, and change the position status of the complete string images; S103. Segment the complete string images with the changed position status to obtain multiple photovoltaic panel images, and classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the classification result includes the identifier of the photovoltaic panel image; S104. Determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and locate the abnormal photovoltaic panel corresponding to the original infrared image based on the identifier of the abnormal photovoltaic panel image.

[0024] The method provided in this application is applied to an anomaly detection system, mainly carried out on intelligent devices such as hosts, computers, laptops, servers, etc. that have data calculation and data analysis capabilities.

[0025] In step S101, first, multiple original infrared images of multiple photovoltaic panel strings are obtained. The specific acquisition method is as follows: using an infrared thermal imaging camera carried by a drone, and controlling the drone to move at a fixed distance, so that the infrared thermal imaging camera takes multiple original infrared images of the photovoltaic panel strings in the air. Generally, the photovoltaic panel strings are arranged in 2 rows and 11 columns. Since the photovoltaic panel strings are composed of photovoltaic panels arranged in 2 rows and 11 columns, the string images are rectangular. And generally, there are multiple photovoltaic panel strings in the same photovoltaic power generation scene. The original infrared images include infrared images of multiple photovoltaic panel strings. Due to the shooting angle, it is easy for the obtained original infrared images to contain incomplete string images and complete string images, and it is also easy for the captured string images to be rotated. Among them, the rotated string images are likely to reduce the accuracy of anomaly detection. And a string image that is incomplete in this original infrared image may be complete in the next original infrared image. The incomplete string images are incomplete string images, which increases the calculation amount. Therefore, the original infrared images need to be processed.

[0026] Then, this application outputs multiple obtained original infrared images to a pre-trained string detection model, and processes the complete string images of multiple groups of photovoltaic panel strings obtained in advance through the pre-trained string detection model. The string detection model is a trained YOLOv8 OBB model, or can also be other models that support rotated bounding box identification and data calculation. The training method of the string detection model is existing and will not be elaborated here. The string detection model performs detection on each received original infrared image, detects the existence of each string image in the original infrared image, and obtains the position information of each string image in the original infrared image, that is, obtains the position information of the string images in the original infrared image.

[0027] In the specific implementation process of step S101, there is an embodiment: Processing multiple original infrared images of multiple groups of photovoltaic panel strings obtained through the pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images, including: S1011. Establish a coordinate system, and determine the initial coordinates of each vertex of the string image based on the coordinate system; S1012. Normalize the initial coordinates to obtain the position information of each vertex of the string image.

[0028] In steps S1011 - S1012, the string detection model first establishes a coordinate system for the original infrared image, that is, uses a certain vertex of the image as the origin. Here, the upper left vertex is used as the origin, and the two bounding boxes connected to the upper left vertex are the X-axis and Y-axis, that is, the long side is the X-axis and the short side is the Y-axis. If there is a string image in the original infrared image, and based on the coordinate system, determine the initial coordinates of each vertex of the string image. To facilitate processing and avoid too large a numerical difference between the initial coordinates, normalize the initial coordinates so that the coordinates of each vertex are between [0, 1]. Then, after normalization, obtain the position information of each vertex of the string image. The output form of the string detection model is: (cls, x 1 , y 1 , x 2 , y 2 ,x 3 , y 3 , x 4 , y 4 ) detection result, where cls represents the detected string image, x 1 ,y 1 , x 2 , y 2 ,x 3 , y 3 , x 4 , y4 They respectively correspond to the four vertices of the oriented bounding box of the string image. The actual coordinate positions of the four vertices of the determined bounding box can be calculated based on their lengths and widths of the oriented bounding box, and the angle Ɵ between the length of the oriented bounding box and the horizontal line is calculated. Then the position information includes the coordinates of the four vertices and the angle Ɵ between the length of the oriented bounding box and the horizontal line. The angle Ɵ represents the inclination degree of the complete string image relative to the horizontal position, such as Figure 2 shown, the oriented bounding box can accurately locate the string corresponding to the string image according to the string direction.

[0029] In step S102, after obtaining the position information of the string image in each original infrared image, it is determined whether there is a defective string image in the original infrared image based on the position information, and the defective string images in the original infrared image are filtered out. There are various ways of filtering, which are not limited here. Thus, only the existence of the complete string image is retained. Then, the position state of the complete string image is determined based on the position information of the complete string image, that is, it is determined whether the complete string image is in a rotated position state or a horizontal position state based on the angle in the position information of the complete string image. If it is in a rotated position state, the position state of the complete string image is changed to a horizontal position state. By filtering out the existence of defective string images, the increase in the detection workload is avoided, and the accuracy of the abnormal detection of the photovoltaic panel is ensured by filtering out defective string images and changing the position state.

[0030] In a specific implementation process of step S102, there is an embodiment: filtering out the defective string images in the original infrared image and retaining the complete string image includes: S10211. Determine whether the string image meets the filtering conditions based on the position information of the string image; S10212. If so, determine that the string image is a defective string image and filter out the defective string image.

[0031] In steps S10211 - S10212, it is determined whether the string image meets the filtering conditions based on the position information of the string image. The filtering condition is that at least one vertex of the string image has an intersection with the X-axis and Y-axis of the established coordinate system or the distance from the X-axis and Y-axis is lower than a preset filtering distance. Then, the position information of the string image is judged. If the position information of the string image indicates that its bounding box has an intersection with the X-axis and Y-axis of the coordinate system or the distance from the X-axis and Y-axis is lower than the preset filtering distance, it is determined that the string image is the defective string image and meets the filtering conditions, such as Figure 3The incomplete string image shown is filtered out, and if the filtering condition is not met, the string image is a complete string image, and only the existence of the complete string image is retained. The preset filtering distance can be specifically set according to the actual situation, so as to avoid repeated detection of the same string and reduce the workload of detecting anomalies.

[0032] In the specific implementation process of step S102, there is another embodiment: determining the position state of the complete string image based on the position information and changing the position state of the complete string image, including: S10221: Determine whether the complete string image is in a horizontal position state based on the position information; S10222: If not, determine that the complete string image is in a rotated position state and change the rotated position state.

[0033] In steps S10221 - S10222, determine whether the long side of the complete string image is in a horizontal position state based on the angle Ɵ between the length of the oriented bounding box in the position information and the horizontal line, that is, determine whether the long side of the complete string image is in a horizontal position state by whether the angle Ɵ between the long side of the complete string image and the horizontal line is 0. If the angle Ɵ is 0, the long side of the complete string image and the complete string image are in a rotated position state, and then the long side of the complete string image and the position state of the complete string image need to be changed to a horizontal position state to improve the accuracy of detecting photovoltaic panel anomalies. If so, determine that the complete string image is in a horizontal position state, and there is no need to change the position state of the complete string image, and the complete string image can be processed in the next step.

[0034] In the specific implementation process of step S10222, there is an embodiment: changing the rotated position state includes: S102221: Determine the angle between the long side of the complete string image and the horizontal line based on the position information; S102222: Move the complete string image in the reverse direction of the angle to the horizontal position state to complete the change of the position state.

[0035] In steps S102221 - S102222, based on the magnitude of the angle Ɵ between the length of the directed bounding box in the position information and the horizontal line, determine the magnitude of the angle by which the complete string image needs to be moved, that is, based on the angle Ɵ, determine the distance by which other vertices that need to be moved need to be moved, and then move in the reverse direction of the angle, so that the long side of the complete string image is parallel to the horizontal line, that is, move the complete string image to the horizontal position state to complete the change of the position state. After completing the change of the position state, to ensure the accuracy of the movement, collect the position information of the complete string image after the change of the position state again based on the string detection model. Based on whether there are the same X-axis or Y-axis in the angle Ɵ or any two position coordinates of the directed bounding box in the position information of the complete string image after the change of the state, if it exists or the angle Ɵ is 0, then confirm that the position of the complete string image after the change of the state is the horizontal position state, otherwise, steps S102221 - S102222 need to be repeated until the position of the complete string image is the horizontal position state.

[0036] In step S103, after obtaining the complete string image in the horizontal position state, segment the complete string image after the change of the position state to obtain multiple photovoltaic panel images. The specific segmentation method is as follows: number the detected complete string images, such as: 1, 2, 3... Then, according to the arrangement rule of the photovoltaic panel string, that is, 2 rows and 11 columns, equally divide the obtained horizontal string image, where the distance is the length of the long side or the short side of a single photovoltaic panel, to obtain photovoltaic panel images with a single photovoltaic panel as the unit. At the same time, number each photovoltaic panel image, such as: 1-1, 1-2, 1-3...1-21, 1-22, so that each obtained photovoltaic panel image corresponds one by one to the photovoltaic panel in the original image. Classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the abnormal dimension is obtained by extracting features from a large number of abnormal photovoltaic panels. The classification result includes the identifier of the photovoltaic panel image; the identifier is the number of the corresponding photovoltaic panel image, that is, based on the identifier, the corresponding photovoltaic panel can be located, and it is also convenient to quickly locate the abnormal photovoltaic panel.

[0037] In the specific implementation process of step S103, there is an embodiment: the classifying the photovoltaic panel images through multiple classification dimensions to obtain a classification result includes: S1031. Extract the topological structure features and color features in the photovoltaic panel image, and map the topological structure features and color features into multiple high-dimensional features; S1032. Classify the multiple high-dimensional features to obtain multiple classification dimensions for classifying the photovoltaic panel images.

[0038] In steps S1031 - S1032, the photovoltaic panel image is classified based on a pre - trained classification model. The specific training method of the classification model is an existing and mature technology, which will not be elaborated here. The classification model consists of 2 convolutional layers and 2 fully - connected layers, specifically a 4 - classification neural network model. The classification model takes a single infrared image of a photovoltaic panel as input and has 4 categories as the output layer. The convolutional layers are the main feature extraction units. They can extract shallow features such as the topological structure and color features including the fault area in the image, and map the topological structure features and the color features into multiple classification dimensions. The fully - connected layers are the main classification units, which classify the multiple high - dimensional features to obtain multiple classification dimensions for classifying the photovoltaic panel image. The two fully - connected layers can effectively classify the photovoltaic panel image using the topological structure features and the color features obtained by the convolutional layers. The multiple classification dimensions obtained include normal, single hot spot, multiple hot spots, and diode failure. The corresponding classification results are normal, single hot spot, multiple hot spots, and diode failure, and single hot spot, multiple hot spots, and diode failure are determined as abnormal. Among them, the topological structure feature of a single hot spot is that there is one point in the photovoltaic panel image; the topological structure feature of multiple hot spots is that there are multiple points in the photovoltaic panel image; the topological structure feature of diode failure is that there is a long - strip graph in the photovoltaic panel image. And single hot spot, multiple hot spots, and diode failure are also accompanied by color abnormalities, usually manifested as darker colors or uneven color distributions caused by temperature increases. For example, if the color of a certain area is significantly different from other areas, it may indicate color abnormality in that area. If there are no any graphs in the photovoltaic panel image and the color is normal or the color distribution is uniform, the corresponding photovoltaic panel is normal. Thus, the influence of factors such as background and angle rotation in the original image on the classification result can be excluded, and a high - precision classification result can be obtained through a shallow convolutional neural network.

[0039] In step S104, after obtaining the classification result, the classification result is judged. If the classification result includes at least one of the following: single hot spot, multiple hot spots, and diode failure, then based on the classification result, the photovoltaic panel image is determined as an abnormal photovoltaic panel image. After determining the abnormal photovoltaic panel image, the abnormal photovoltaic panel corresponding to the original infrared image can be quickly located based on the identifier (i.e., the number) of the abnormal photovoltaic panel image, thus achieving the effect of quick positioning. It is also possible to determine the treatment measures for the photovoltaic panel based on the classification result, which can be repair or scrapping determination, depending on the actual situation. And the relevant information of the abnormal photovoltaic panel is recorded in the database for subsequent analysis and improvement of the photovoltaic panel maintenance strategy.

[0040] In a specific implementation process of step S104, there is an embodiment: The method further includes: S1041. If it is determined that the photovoltaic panel image does not have a topological structure and the corresponding color feature does not match the color anomaly condition; S1042. Then determine that the classification result of the photovoltaic panel image is normal.

[0041] In steps S10321 - S10322, if the classification module cannot extract the topological structure from the photovoltaic panel image, that is, the photovoltaic panel image does not have the topological structures corresponding to single hot spot, multi - hot spot, and diode failure respectively, and the color feature of the photovoltaic panel image cannot match the color anomaly condition, where the color anomaly condition is that the color is darker or the color distribution is uneven, and the color anomaly condition is extracted from the color features of a large number of normal photovoltaic panel images, and the color feature of the photovoltaic panel image is normal or the color distribution is uniform, then determine that the classification result corresponding to the photovoltaic panel image is normal. Other methods can also be used to detect the photovoltaic panel image again to ensure the accuracy of the classification result. After obtaining the normal classification result, record the relevant information of the photovoltaic panel corresponding to the normal classification result. Embodiment 2 This application also provides an abnormal detection device for a photovoltaic panel, as Figure 4 shown in the block diagram of an abnormal detection device for a photovoltaic panel. The functions implemented by this abnormal detection device for a photovoltaic panel correspond to the steps of performing an abnormal detection method for a photovoltaic panel on a terminal device. This device can be understood as a component of a server including a processor. The device includes: A processing module 401, configured to process multiple original infrared images of multiple groups of photovoltaic panel strings obtained through a pre - trained string detection model to obtain the position information of the string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangular; A determination module 402, configured to filter out the incomplete string images in the original infrared images, retain the complete string images, determine the position status of the complete string images based on the position information, and change the position status of the complete string images; A segmentation module 403, configured to segment the complete string images with the changed position status to obtain multiple photovoltaic panel images, and classify the photovoltaic panel images through multiple classification dimensions to obtain a classification result; the classification result includes the identifier of the photovoltaic panel image; A positioning module 404, configured to determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and locate the abnormal photovoltaic panel corresponding to the original infrared image based on the identifier of the abnormal photovoltaic panel image.

[0042] In a feasible implementation manner, the determination module includes: A first determination module, configured to determine whether the complete string image is in a horizontal position state based on the position information; The first changing module is used to determine that the complete string image is in a rotation position state and change the rotation position state if no.

[0043] In a feasible implementation manner, the determination module further includes: A first determination module, configured to determine an angle between a long side of the complete string image and a horizontal line based on the position information; The second changing module is used to move the complete string image in the opposite direction according to the angle to a horizontal position state to complete the position state change.

[0044] In a feasible implementation manner, the determination module also includes: A judging module, used for judging whether the group string image meets the filtering condition based on the position information of the group string image; A filtering module is used to determine that the group string image is an incomplete group string image and filter out the incomplete group string image.

[0045] In a feasible implementation manner, the segmentation module includes: An extraction module, used to extract topological structure features and color features in the photovoltaic panel image, and map the topological structure features and color features into multiple high-dimensional features; The classification module is used to classify the multiple high-dimensional features to obtain multiple classification dimensions to classify the photovoltaic panel image.

[0046] In a feasible implementation manner, the positioning module includes: A second determination module is used to determine if the photovoltaic panel image does not have a topological structure and the corresponding color feature does not match the color abnormality condition; The second determination module is used to determine whether the classification result of the photovoltaic panel image is normal.

[0047] In a feasible implementation manner, the processing module includes: An establishing module, used for establishing a coordinate system, and determining the initial coordinates of each vertex of the group string image based on the coordinate system; The first processing module is used to normalize the initial coordinates to obtain the position information of each vertex of the string image.

[0048] Example 3 The present application also provides an electronic device, such as Figure 5As shown in the figure, it includes: a processor 501, a memory 502, and a bus 503. The memory 502 stores machine-readable instructions executable by the processor 501. When the electronic device is running, the processor 501 communicates with the memory 502 through the bus 503. When the machine-readable instructions are executed by the processor 501, the steps of any one of the abnormal detection methods of a photovoltaic panel are executed.

[0049] Embodiment 4 The present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the steps of any one of the abnormal detection methods of a photovoltaic panel are executed.

[0050] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems and devices described above can refer to the corresponding processes in the method embodiments, which will not be elaborated in the present application. In the several embodiments provided by the present application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For another example, 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 couplings or direct couplings or communication connections shown or discussed with each other can be through some communication interfaces. The indirect couplings or communication connections of the devices or modules can be in electrical, mechanical, or other forms.

[0051] The modules described as separate components may or may not be physically separated. The components shown as modules may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0052] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0053] When the above-mentioned function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a 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 causing a computer device (which may be a personal computer, a platform server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0054] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for detecting abnormality of a photovoltaic panel, characterized in that: The method comprises: A plurality of original infrared images of a plurality of photovoltaic panel strings are obtained by processing the pre-trained string detection model, and position information of the string images in the corresponding original infrared images is obtained; the string images include incomplete string images and complete string images; the string images are rectangular; filtering out the incomplete string image in the original infrared image, retaining the complete string image, determining the position state of the complete string image based on the position information, and changing the position state of the complete string image; Segmenting the complete string image after the position state change to obtain a plurality of photovoltaic panel images, and classifying the photovoltaic panel images through a plurality of classification dimensions to obtain classification results; the classification results include identifications of the photovoltaic panel images; The photovoltaic panel image is determined to be an abnormal photovoltaic panel image based on the classification result, and the abnormal photovoltaic panel corresponding to the original infrared image is located based on the identification of the abnormal photovoltaic panel image.

2. The method according to claim 1, characterized in that The determining the position state of the complete string image based on the position information and changing the position state of the complete string image includes: Determining whether the complete string image is in a horizontal position state based on the position information; If not, it is determined that the complete string image is in a rotation position state, and the rotation position state is changed.

3. The method according to claim 2, characterized in that The changing of the rotation position state comprises: Determining an angle between a long side of the complete string image and a horizontal line based on the position information; The complete group string image is moved in the reverse direction according to the angle to a horizontal position state to complete the position state change.

4. The method according to claim 1, characterized in that: The filtering out the incomplete string image in the original infrared image and retaining the complete string image includes: Determining whether the group string image meets the filtering condition based on the position information of the group string image; If so, it is determined that the group string image is an incomplete group string image, and the incomplete group string image is filtered out.

5. The method according to claim 1, characterized in that The step of classifying the photovoltaic panel image by multiple classification dimensions to obtain a classification result includes: Extracting topological structure features and color features from the photovoltaic panel image, and mapping the topological structure features and color features into multiple high-dimensional features; The multiple high-dimensional features are classified into multiple classification dimensions to classify the photovoltaic panel image.

6. The method according to claim 1, characterized in that The method further comprises: If it is determined that the photovoltaic panel image does not have a topological structure and the corresponding color feature does not match the color anomaly condition; The classification result of the photovoltaic panel image is determined to be normal.

7. The method according to claim 1, characterized in that The multiple original infrared images of multiple groups of photovoltaic panel strings acquired by processing the pre-trained string detection model to obtain the position information of the string images in the corresponding original infrared images include: Establishing a coordinate system, and determining the initial coordinates of each vertex of the string image based on the coordinate system; The initial coordinates are normalized to obtain position information of each vertex of the string image.

8. A photovoltaic panel abnormality detection device, characterized in that: The device comprises: A processing module, used for processing a plurality of original infrared images of a plurality of photovoltaic panel strings acquired by a pre-trained string detection model to obtain position information of string images in the corresponding original infrared images; the string images include incomplete string images and complete string images; the string images are rectangular; A determination module, used for filtering out the incomplete string image in the original infrared image, retaining the complete string image, determining the position state of the complete string image based on the position information, and changing the position state of the complete string image; A segmentation module, used for segmenting the complete string image after the position state is changed to obtain multiple photovoltaic panel images, and classifying the photovoltaic panel images through multiple classification dimensions to obtain classification results; the classification results include identifications of the photovoltaic panel images; A positioning module is used to determine that the photovoltaic panel image is an abnormal photovoltaic panel image based on the classification result, and to locate the abnormal photovoltaic panel corresponding to the original infrared image based on the identification of the abnormal photovoltaic panel image.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of a photovoltaic panel abnormality detection method as described in any one of claims 1 to 7 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the abnormality detection method for a photovoltaic panel as claimed in any one of claims 1 to 7 are executed.

Images