A photovoltaic module fault detection method and device, electronic equipment and storage medium

Abstract

The application discloses a photovoltaic module fault detection method and device, electronic equipment and storage medium. Including: acquiring an infrared image of a photovoltaic power station; using a pre-trained fault detection model to detect the infrared image to obtain a fault area and a first fault category, the fault detection model uses a dense link structure to enhance feature reuse; using a pre-trained fault category identification model to identify the fault area to obtain a second fault category, the fault category identification model is constructed according to the association function between the pre-acquired image HSV value, average temperature and maximum temperature corresponding to the fault area and the fault category; according to a preset selection condition, the final fault category of the photovoltaic module is selected from the first fault category and the second fault category. The method fuses the fault categories obtained by using two different models to obtain the final fault category, effectively improving the photovoltaic module fault detection accuracy.

Description

Technical Field

[0001] The present invention relates to the field of target detection technology, and in particular to a method, apparatus, electronic device and storage medium for detecting photovoltaic module faults. Background Technology

[0002] In recent years, energy and environmental issues have constrained my country's economic and social development, while photovoltaic (PV) power generation has provided an opportunity for the development of new energy sources. However, with the increasing installed capacity of PV systems, the problem of PV module failures has become increasingly prominent. PV module failures not only affect power generation and increase the operation and maintenance costs of PV power plants, but also pose numerous risks and safety hazards.

[0003] The specific scheme of the existing photovoltaic module fault detection method includes: acquiring images of the photovoltaic module through an image acquisition device, the images including infrared thermal imaging images and visible light images; stitching the images using an image stitching algorithm based on the enhanced KAZE algorithm; performing color processing on the images using image processing algorithms based on the HSV model and the YCbCr model; processing the images using median filtering, morphological image processing, edge detection, contour extraction, and region separation methods; extracting feature vectors of the photovoltaic module fault area from the infrared thermal imaging image and the visible light image using Local Binary Pattern (LBP) respectively; classifying and recognizing the obtained feature vectors using a convolutional neural network algorithm, and fusing the recognition results of the same location in the infrared thermal imaging image and the visible light image to determine the fault type.

[0004] The above solution only uses neural network algorithms to classify and identify faults in photovoltaic modules, which cannot effectively guarantee the accuracy of identification. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for detecting photovoltaic module faults, in order to solve the problem that existing technologies using a single neural network algorithm to classify and identify photovoltaic module faults cannot effectively guarantee the accuracy of identification.

[0006] According to one aspect of the present invention, a method for detecting faults in photovoltaic modules is provided, comprising:

[0007] Acquire infrared images of a photovoltaic power station, the infrared images showing faulty photovoltaic modules;

[0008] The infrared image is detected using a pre-trained fault detection model to obtain the fault region and the first fault category. The fault detection model uses a dense link structure to enhance feature reuse.

[0009] The fault area is identified using a pre-trained fault category recognition model to obtain a second fault category. The fault category recognition model is constructed based on the correlation function between the HSV value, average temperature, and maximum temperature of the image corresponding to the fault area and the fault category.

[0010] The final fault category of the photovoltaic module is selected from the first fault category and the second fault category based on preset selection criteria.

[0011] According to another aspect of the present invention, a photovoltaic module fault detection device is provided, comprising:

[0012] An acquisition module is used to acquire infrared images of a photovoltaic power station, wherein the infrared images show photovoltaic modules that have malfunctions;

[0013] The detection module is used to detect the infrared image using an optimized fault detection model to obtain the fault area and the first fault category;

[0014] The discrimination module is used to discriminate the fault area using a pre-trained fault category recognition model to obtain a second fault category. The fault category recognition model is constructed based on the correlation function between the image HSV value, average temperature, and maximum temperature of the fault area and the fault category.

[0015] The selection module is used to select the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions.

[0016] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0017] At least one processor;

[0018] and a memory communicatively connected to the at least one processor;

[0019] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the photovoltaic module fault detection method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions, the computer instructions being configured to cause a processor to execute and implement the photovoltaic module fault detection method according to any embodiment of the present invention.

[0021] The technical solution of this invention involves acquiring infrared images of a photovoltaic power station, where the infrared images show faulty photovoltaic modules; using a pre-trained fault detection model to detect the infrared images and obtain a fault area and a first fault category, wherein the fault detection model uses a dense link structure to enhance feature reuse; using a pre-trained fault category recognition model to discriminate the fault area and obtain a second fault category, wherein the fault category recognition model is constructed based on the correlation function between the image HSV value, average temperature, and maximum temperature corresponding to the fault area and the fault category; and selecting the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions. This technical solution uses the fault detection model to identify the infrared image to obtain the first fault category, and then uses the fault category recognition model constructed in advance based on the correlation function between the image HSV value, average temperature, and maximum temperature and the fault category to perform a second identification of the fault area to obtain the second fault category. The first fault category and the second fault category are then merged to obtain the final fault category of the photovoltaic module. This solution effectively improves the accuracy of photovoltaic module fault detection and identification by using two models to identify the fault category.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a network structure diagram of the fault detection model provided in Embodiment 1 of the present invention;

[0025] Figure 2 This is a schematic diagram of the SPPF module used in the YOLOv5 algorithm provided in an embodiment of the present invention;

[0026] Figure 3 This is a schematic diagram of the SPPF module in a fault detection model provided in an embodiment of the present invention;

[0027] Figure 4 This is a flowchart illustrating a photovoltaic module fault detection method provided in Embodiment 2 of the present invention;

[0028] Figure 5This is a schematic diagram of the structure of a photovoltaic module fault detection device provided in Embodiment 3 of the present invention;

[0029] Figure 6 This is a schematic diagram of the electronic device used in a photovoltaic module fault detection method according to an embodiment of the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention. It should be understood that the various steps described in the method embodiments of the present invention can be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0031] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0034] The names of the messages or information exchanged between the multiple devices in the embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0035] Example 1

[0036] Embodiment 1 of the present invention provides a training process for a fault detection model and a training process for a fault category identification model used in a photovoltaic module fault detection method.

[0037] The fault detection model is derived from the YOLOv5 algorithm through optimization. The main optimizations are to the SPPF module in the YOLOv5 network structure and to enhance feature reuse in dense link structures, such as... Figure 1 As shown, Figure 1 This is a network structure diagram of the fault detection model provided in Embodiment 1 of the present invention.

[0038] In this embodiment, the network structure of the fault detection model includes: an input layer, a backbone network, a neck network, and a detection module;

[0039] The input layer is used to preprocess the input image and enhance its features, outputting the first feature map.

[0040] The backbone network consists of multiple CBS modules and multiple CSP modules. The backbone network is used to extract features from the first feature map to obtain a second feature map.

[0041] The neck network consists of an SPPF module, multiple CBS modules, multiple CSP modules, and multiple sampling layers. The neck network is used to extract features from the second feature map at multiple scales to obtain feature information at different scales.

[0042] The detection module is used to process the feature information to obtain the detection result.

[0043] The input layer performs Mosaic data augmentation, adaptive anchor box calculation, and adaptive image scaling on the input images to enhance their richness. The backbone network includes a Focus structure and a CSP structure. The Focus structure uses 32 convolutional kernels. The original 608*608*3 image is input into the Focus structure, sliced ​​into a 304*304*12 feature map, and then convolved again with 32 kernels to finally become a 304*304*32 feature map. Specifically, the backbone network consists of the following modules arranged sequentially: the first CBS module, the second CBS module, the first CSP module, the third CBS module, the second CSP module, the fourth CBS module, the third CSP module, the fifth CBS module, and the fourth CSP module. The neck network consists of the following modules arranged sequentially: SPPF module, first CBS module, first upsampling layer, first Concat layer, first CSP module, second CBS module, second upsampling layer, second Concat layer, second CSP module, third CBS module, third Concat layer, third CSP module, fourth CBS module, fourth Concat layer, and fourth CSP module. The detection module includes three head detection heads and one NMS submodule. The three head detection heads are used to process the feature maps output by the second, third, and fourth CSP modules in the neck network, respectively, to obtain the coordinates, category information, and confidence scores of the predicted bounding boxes of insulator defects. The NMS module is used to remove redundant predicted bounding boxes to obtain the final detection results.

[0044] Figure 2 This is a schematic diagram of the SPPF module used in the YOLOv5 algorithm according to an embodiment of the present invention, as shown below. Figure 2 As shown, SPPF uses three 5*5 max pooling operations instead of the original 5*5, 9*9, and 13*13 max pooling operations. Multiple small-scale pooling kernels are cascaded to replace the single large-size pooling kernel in the SPP module. This further improves the running speed while retaining the original functionality—fusing feature maps from different receptive fields and enriching the expressive power of the feature maps. The SPPF module mainly utilizes the fact that images can learn features at multiple scales with the help of max pooling and skip connections, and then combines global and local features to increase the representativeness of the feature maps. Max pooling is a method that uses a rectangular mask to extract the maximum value from a set of image regions. Although max pooling can help reduce irrelevant data, it often leads to the discarding of less useful feature data. To address this issue, the SPPF module needs to be improved by leveraging a dense link structure similar to that in DenseNet to enhance feature reuse in the original SPPF structure.

[0045] Furthermore, the SPPF module of the fault detection model uses a dense link structure to fuse the feature information extracted by the CBS module with the feature information extracted by each subsequent max pooling layer, and then fuses the fused information with multiple fusion modules of the CBS module to reuse the feature information of each module.

[0046] Specifically, by employing the concept of dense connections, the feature information extracted by the CBS layer is fused with the features extracted by each subsequent MaxPool layer. This allows for the continuous reuse of features from the CBS layer, preventing feature loss due to downsampling in the MaxPool layer. Simultaneously, the feature information fused from the CBS layer and each MaxPool layer is fused back with the CBS layer. In other words, the CBS layer in the diagram is fused with the three yellow modules in the middle, further reusing the feature information from each layer. This ensures that feature information is not lost as much as possible, and also prevents the loss of small target features due to downsampling in the MaxPool layer, thus preserving the features of small targets and improving the detection performance of small targets.

[0047] SPPF module in fault detection model, such as Figure 3 As shown, Figure 3 This is a schematic diagram of the SPPF module in a fault detection model provided in an embodiment of the present invention. The dashed arrows indicate newly added link structures.

[0048] Furthermore, the network structure of the SPPF module of the fault detection model includes four layers: the first layer includes a CBS module and three max pooling layers; the second layer includes three fusion modules; the third layer includes a fusion module; and the fourth layer includes a CBS module.

[0049] The first fusion module in the second layer is used to superimpose the first output data of the CBS module in the first layer with the second output data of the first pooling layer in the first layer in the channel dimension to obtain the first data;

[0050] The second fusion module in the second layer is used to superimpose the first output data and the third output data of the second pooling layer in the first layer in the channel dimension to obtain the second data;

[0051] The third fusion module in the second layer is used to superimpose the first output data, the fourth output data of the third pooling layer in the first layer, and the first data in the channel dimension to obtain the third data;

[0052] The fusion module in the third layer is used to superimpose the first output data, the first data, the second data, and the third data in the channel dimension to obtain the fourth data.

[0053] For example, the SPPF module of the fault detection model takes X as input, and the output after passing through the CBS module is Y1. The output of the first pooling layer is Y2, the output of the second pooling layer is Y3, and the output of the third pooling layer is Y4. The input of the first Concat layer is Y1 and Y2. The Concat layer mainly stacks Y1 and Y2 in the channel dimension without involving convolution operations, and its output is C1 = [Y1, Y2]. The input of the second Concat layer is Y1 and Y3, and its output is C2 = [Y1, Y3]. The input of the third Concat layer is Y1, Y4, and C1, and its output is C3 = [Y1, Y4, C1]. The input of the fourth Concat layer is Y1, C1, C2, and C3, and its output is C4 = [Y1, C1, C2, C3].

[0054] In this embodiment, the training process of the fault detection model includes: inputting images of faulty components from the photovoltaic module database into the fault detection model for training; stopping training when the number of training iterations reaches a preset value to obtain a pre-trained fault detection model.

[0055] The fault component image may include a fault component. The training process includes: inputting the fault component into a pre-built fault detection model; preprocessing and enhancing the image features through the input layer of the fault detection model; extracting features from the enhanced image using the backbone network to obtain a feature map; extracting multi-scale features from the feature map using the neck network to obtain feature information at different scales; and detecting the detection boxes of the feature information to obtain the fault category corresponding to the fault component. This training process is repeated until the preset number of training iterations is reached, at which point model training stops, resulting in a pre-trained fault detection model.

[0056] It should be noted that, based on the inventors' previous research on photovoltaic panel fault data, they discovered a certain relationship between the fault category of the module and the HSV value of the fault area image, as well as the average and maximum temperatures of the image location where the fault area is located. Therefore, a large number of fault areas were collected in advance, and the HSV values ​​of a large number of fault areas, as well as the average and maximum temperatures of the image locations where the fault areas are located, were extracted. Based on the correspondence between the values, a correlation function was established, and a mathematical model of image HSV value-average temperature-maximum temperature-fault category was constructed. By strengthening and training this mathematical model, a photovoltaic module fault category recognition model was obtained.

[0057] In this embodiment, the construction and training process of the fault category identification model includes:

[0058] The fault area image is extracted from the infrared image, and the HSV value of the fault area is calculated from the fault area image using OpenCV software.

[0059] Infrared acquisition software is used to collect temperature information of all locations in the image of the fault area, and the average temperature and maximum temperature of the fault area are calculated based on the temperature information.

[0060] Construct a functional relationship between the HSV value, the average temperature, and the maximum temperature, and construct a mathematical model corresponding to the functional relationship as a fault category identification model;

[0061] The HSV value, the average temperature, and the highest temperature are used as inputs to the fault category identification model, and the actual fault category is used as the label to train the model and obtain the optimal parameters.

[0062] The infrared image can be an infrared image of the faulty component. Infrared images can be acquired in several ways, including: Method 1: capturing infrared images of the faulty component using a drone equipped with a gimbal camera during aerial inspections of the photovoltaic power station; Method 2: downloading infrared images of the faulty component from an image library. The image HSV value represents the HSV (Hue, Saturation, Value) color space, where H represents hue, S represents saturation, and V represents brightness. HSV can very intuitively express the hue, vividness, and brightness of a color, facilitating color comparison.

[0063] Infrared images contain temperature information, so the temperature information of the location of the fault area can be extracted from the infrared image, and the average temperature and maximum temperature of the area can be calculated.

[0064] Specifically, assuming the image HSV value, average temperature, maximum temperature, and true fault category are represented by A, B, C, and D respectively, we first construct a mathematical model D = P1*A + P2*B + P3*C. The three sets of parameters P1, P2, and P3 need to be optimized using reinforcement learning; this can be understood as training the constructed mathematical model to optimize its parameters. We acquire a large number of infrared images of faulty components collected previously. From these images, we extract the image HSV value, average temperature, and maximum temperature as input, and the true fault category as the label. We then train the model to obtain the optimal three sets of parameters P1, P2, and P3. Substituting these optimal parameters into the constructed mathematical model D = P1*A + P2*B + P3*C yields the pre-trained fault category recognition model.

[0065] Example 2

[0066] Figure 4This is a flowchart illustrating a photovoltaic module fault detection method provided in Embodiment 2 of the present invention. This method is applicable to the identification of module faults in photovoltaic power plants. The method can be executed by a photovoltaic module fault detection device, which can be implemented by software and / or hardware and is generally integrated into an electronic device. In this embodiment, the electronic device includes, but is not limited to, a computer device.

[0067] like Figure 4 As shown in Embodiment 2 of the present invention, a photovoltaic module fault detection method includes the following steps:

[0068] S110. Obtain an infrared image of the photovoltaic power station, wherein the infrared image shows a faulty photovoltaic module.

[0069] A photovoltaic (PV) power station refers to a power generation system that utilizes solar energy, employs special materials such as crystalline silicon panels, inverters, and other electronic components, and is connected to and transmits electricity to the power grid. An infrared image, short for thermal infrared image, is an image formed by a thermal infrared scanner receiving and recording the thermal radiation energy emitted by a target object.

[0070] In this embodiment, infrared images of a photovoltaic power station can be obtained through various methods. The infrared image of a photovoltaic power station refers to the infrared image obtained after photographing the components within the photovoltaic power station. For example, infrared images of a photovoltaic power station can be obtained through the following two methods:

[0071] Method 1: Real-time data collection and acquisition. Specifically, a DJI M300 drone equipped with an H20T gimbal camera is used to conduct drone inspections of the photovoltaic power station along a pre-planned route, continuously capturing infrared images during the inspection process;

[0072] Method 2: Direct Acquisition. Specifically, infrared images of photovoltaic power plants are obtained from an infrared image library, which includes infrared images of various photovoltaic power plants. Examples can be provided to illustrate how the infrared images in the library were acquired.

[0073] In both methods described above, an infrared image can display only one component, or it can display two or more components. Some infrared images may contain fault-free components, some may contain faulty components, and some may contain both. It is understandable that most acquired infrared images will display faulty components, while a small number of infrared images, or none at all, will display fault-free components.

[0074] It should be noted that the acquired infrared image contains temperature information, and the temperature information of the corresponding area can be extracted from the infrared image.

[0075] S120. The infrared image is detected using a pre-trained fault detection model to obtain the fault area and the first fault category. The fault detection model uses a dense link structure to enhance feature reuse.

[0076] The purpose of using a dense link structure in the fault detection model is to enhance feature reuse, reduce lost feature information, retain feature information of small targets in component failures, and improve the detection effect of small targets.

[0077] In this embodiment, after obtaining the pre-trained fault detection model, the acquired infrared image can be input into the pre-trained fault detection model. The input layer, backbone network, neck network, and detection module of the pre-trained fault detection model perform feature extraction and feature enhancement on the infrared image to detect faulty components, outputting the fault region and fault category. The fault category is designated as the first fault category. The first fault category can be understood as the fault category output by the pre-trained fault detection model, and the fault region can be understood as the location of the faulty component on the infrared image.

[0078] It is understandable that infrared images can be input into a pre-trained fault detection model in batches. A batch may include one infrared image or multiple infrared images. If multiple infrared images are input, the model outputs the fault areas and fault categories of the multiple infrared images. At least one faulty component's fault area and fault category can be identified on a single infrared image.

[0079] The first fault category can include several categories such as string open circuit, string short circuit, component missing, junction box failure, breakage, obstruction of hot spots and hot spots.

[0080] S130. Use a pre-trained fault category recognition model to identify the fault area and obtain a second fault category.

[0081] The fault category identification model is constructed based on the pre-acquired HSV values, average temperature, and maximum temperature of the image corresponding to the fault area, and the correlation function between these values ​​and the fault category. The construction and training process of the fault category identification model has been described in detail in Example 1 and will not be repeated here.

[0082] Specifically, the photovoltaic module fault category identification model is used to identify the fault area and obtain a second fault category. This includes: inputting the image HSV value, average temperature and maximum temperature corresponding to the fault area into the photovoltaic module fault category identification model, and outputting the second fault category based on the correlation function between the image HSV value, average temperature and maximum temperature and the fault category.

[0083] In this embodiment, the fault area obtained in step S120 is first extracted from the infrared image to obtain a fault area image. OpenCV software is used to calculate the HSV value corresponding to the fault area from the fault area image. Infrared acquisition software is used to collect temperature information at all locations in the fault area image. Based on the temperature information at all locations, the average temperature and maximum temperature corresponding to the fault area are calculated. The average temperature, maximum temperature, and HSV value of the fault area are input into a pre-trained fault category recognition model. A numerical value is calculated using the mathematical model D = P1*A + P2*B + P3*C. The fault category corresponding to this value is output as the second fault category. For example, if the image HSV value is 200, the average temperature is 40 degrees Celsius, and the maximum temperature is 60 degrees Celsius, then the corresponding second fault category is component breakage.

[0084] S140. Select the final fault category of the photovoltaic module from the first fault category and the second fault category according to the preset selection conditions.

[0085] In this embodiment, after the fault detection model, which has been pre-trained, obtains a first fault category, and the fault category recognition model, which has been pre-trained, obtains a second fault category, it is necessary to select one of the first and second fault categories as the final fault category using pre-set selection criteria to output the final fault category. Specifically, the selection of the first or second fault category as the final fault category can be determined based on the category information of the first and second fault categories.

[0086] Furthermore, selecting the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions includes: obtaining first fault category information of the first fault category and second fault category information of the second fault category; if the first fault category information and the second fault category information are the same, then setting the first fault category or the second fault category as the final fault category; if the first fault category information and the second fault category information are different, then obtaining the fault level values ​​corresponding to the first fault category and the second fault category, and taking the fault category corresponding to the fault level value with the larger value as the final fault category.

[0087] Among them, fault category information can be used to characterize the specific category of the fault. For example, if both the first fault category information and the second fault category information are component missing, then the final fault category of the fault area can be determined to be component missing.

[0088] The fault level value characterizes the severity of a fault; a higher fault level value indicates a more severe fault. Fault categories with higher severity levels are designated as the final fault categories. The order of fault level values ​​for each fault category is as follows: Component missing fault level value > String short circuit fault level value > String open circuit fault level value > Component broken fault level value > Junction box fault level value > Blocked hot spot fault level value > Hot spot fault level value. For example, if the first fault category is string short circuit and the second fault category is junction box fault, since the fault level value for string short circuit is higher than that for junction box fault, string short circuit is designated as the final fault category.

[0089] In this embodiment, after obtaining the final fault category, the final fault category can also be displayed on the corresponding infrared image. For example, if the final fault category is hot spot, then "hot spot" can be displayed on the infrared image.

[0090] This invention provides a photovoltaic module fault detection method in Embodiment 2. First, an infrared image of a photovoltaic power station is acquired, showing faulty photovoltaic modules. Then, a pre-trained fault detection model is used to detect the infrared image, obtaining a fault region and a first fault category. The fault detection model uses a dense link structure to enhance feature reuse. Next, a pre-trained fault category recognition model is used to discriminate the fault region, obtaining a second fault category. Finally, based on preset selection criteria, the final fault category of the photovoltaic module is selected from the first and second fault categories. This method, using a fault detection model, can accurately detect and identify module faults, effectively improving the detection and identification of small targets within module faults. By constructing a photovoltaic module fault category recognition model based on the correlation function between the HSV value, average temperature, and maximum temperature of the image corresponding to the fault region and the fault category, and using this model to perform secondary detection on the fault region, the accuracy of fault type identification can be effectively improved.

[0091] Example 3

[0092] Figure 5 This is a schematic diagram of a photovoltaic module fault detection device provided in Embodiment 3 of the present invention. The device is applicable to the identification of module faults in photovoltaic power plants. The device can be implemented by software and / or hardware and is generally integrated into electronic equipment.

[0093] like Figure 5 As shown, the device includes: an acquisition module 110, a detection module 120, a discrimination module 130, and a selection module 140.

[0094] The acquisition module 110 is used to acquire infrared images of a photovoltaic power station, wherein the infrared images show photovoltaic modules with faults;

[0095] The detection module 120 is used to detect the infrared image using a pre-trained fault detection model to obtain the fault area and a first fault category. The fault detection model uses a dense link structure to enhance feature reuse.

[0096] The discrimination module 130 is used to discriminate the fault area using a pre-trained fault category recognition model to obtain a second fault category. The fault category recognition model is constructed based on the pre-acquired correlation function between the image HSV value, average temperature, and maximum temperature of the fault area and the fault category.

[0097] Selection module 140 is used to select the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions.

[0098] In this embodiment, the device first acquires an infrared image of the photovoltaic power station through the acquisition module 110, the infrared image showing a faulty photovoltaic module; then, the detection module 120 uses a pre-trained fault detection model to detect the infrared image, obtaining a fault area and a first fault category. The fault detection model uses a dense link structure to enhance feature reuse. Next, the discrimination module 130 uses a pre-trained fault category recognition model to discriminate the fault area, obtaining a second fault category. The fault category recognition model is constructed based on the correlation function between the pre-acquired image HSV value, average temperature, and maximum temperature of the fault area and the fault category. Finally, the selection module 140 determines the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions.

[0099] This embodiment provides a photovoltaic module fault detection device that can fuse fault categories obtained using two different models to obtain a final fault category, effectively improving the accuracy of photovoltaic module fault detection.

[0100] Furthermore, the training process of the fault detection model includes: inputting images of faulty components from the photovoltaic module database into the fault detection model for training; stopping training when the number of training iterations reaches a preset value to obtain a pre-trained fault detection model.

[0101] Based on the above optimizations, the network structure of the fault detection model includes: an input layer, a backbone network, a neck network, and a detection module. The input layer is used to preprocess the input image and enhance its features to output a first feature map. The backbone network consists of multiple CBS modules and multiple CSP modules, and is used to extract features from the first feature map to obtain a second feature map. The neck network consists of an SPPF module, multiple CBS modules, multiple CSP modules, and multiple sampling layers, and is used to extract multi-scale features from the second feature map to obtain feature information at different scales. The detection module is used to process the feature information to obtain the detection result.

[0102] Based on the above scheme, the SPPF module of the fault detection model uses a dense link structure to fuse the feature information extracted by the CBS module with the feature information extracted by each subsequent max pooling layer, and then fuses the fused information with multiple fusion modules of the CBS module to reuse the feature information of each module.

[0103] Furthermore, the network structure of the SPPF module of the fault detection model includes four layers: the first layer includes a CBS module and three max pooling layers; the second layer includes three fusion modules; the third layer includes a fusion module; and the fourth layer includes a CBS module.

[0104] The first fusion module in the second layer is used to superimpose the first output data of the CBS module in the first layer with the second output data of the first pooling layer in the first layer in the channel dimension to obtain the first data;

[0105] The second fusion module in the second layer is used to superimpose the first output data and the third output data of the second pooling layer in the first layer in the channel dimension to obtain the second data;

[0106] The third fusion module in the second layer is used to superimpose the first output data, the fourth output data of the third pooling layer in the first layer, and the first data in the channel dimension to obtain the third data;

[0107] The fusion module in the third layer is used to superimpose the first output data, the first data, the second data, and the third data in the channel dimension to obtain the fourth data.

[0108] Furthermore, the discrimination module 140 is specifically used to: input the image HSV value, average temperature and maximum temperature corresponding to the fault area into the photovoltaic module fault category recognition model, and output a second fault category based on the correlation function between the image HSV value, average temperature and maximum temperature and the fault category.

[0109] Furthermore, the construction and training process of the fault category identification model includes:

[0110] The fault area image is extracted from the infrared image, and the HSV value of the fault area is calculated from the fault area image using OpenCV software.

[0111] Infrared acquisition software is used to collect temperature information of all locations in the image of the fault area, and the average temperature and maximum temperature of the fault area are calculated based on the temperature information.

[0112] Construct a functional relationship between the HSV value, the average temperature, and the maximum temperature, and construct a mathematical model corresponding to the functional relationship as a fault category identification model;

[0113] The HSV value, the average temperature, and the highest temperature are used as inputs to the fault category identification model, and the actual fault category is used as the label to train the model and obtain the optimal parameters.

[0114] Furthermore, the selection module 140 is specifically used for:

[0115] Obtain first fault category information for the first fault category and second fault category information for the second fault category;

[0116] If the first fault category information and the second fault category information are the same, then the first fault category or the second fault category is set as the final fault category;

[0117] If the first fault category information and the second fault category information are different, then obtain the fault level values ​​corresponding to the first fault category and the second fault category, and take the fault category corresponding to the fault level value with the larger value as the final fault category.

[0118] The photovoltaic module fault detection device described above can execute the photovoltaic module fault detection method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0119] Example 4

[0120] Figure 6A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0121] like Figure 6 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0122] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0123] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as photovoltaic module fault detection methods.

[0124] In some embodiments, the photovoltaic module fault detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the photovoltaic module fault detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the photovoltaic module fault detection method by any other suitable means (e.g., by means of firmware).

[0125] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0126] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0127] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0128] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0129] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0130] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0131] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0132] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting faults in photovoltaic modules, characterized in that, The method includes: Acquire infrared images of a photovoltaic power station, the infrared images showing faulty photovoltaic modules; The infrared image is detected using a pre-trained fault detection model to obtain the fault area and a first fault category. The fault detection model uses a dense link structure to enhance feature reuse. The first fault category includes at least one of the following: string open circuit, string short circuit, component missing, junction box failure, breakage, obstruction of hot spot and hot spot. The fault area is identified using a pre-trained fault category recognition model to obtain a second fault category. The fault category recognition model is constructed based on the correlation function between the HSV value, average temperature, and maximum temperature of the image corresponding to the fault area and the fault category. The final fault category of the photovoltaic module is selected from the first fault category and the second fault category according to the preset selection criteria; The construction and training process of the fault category identification model includes: The fault area image is extracted from the infrared image, and the HSV value of the fault area is calculated from the fault area image using OpenCV software. Infrared acquisition software is used to collect temperature information at all locations in the image of the fault area, and the average temperature and maximum temperature of the fault area are calculated based on the temperature information. Construct a functional relationship between the HSV value, the average temperature, and the maximum temperature, and construct a mathematical model corresponding to the functional relationship as a fault category identification model; The HSV value, the average temperature, and the highest temperature are used as inputs to the fault category identification model, and the actual fault category is used as the label to train the model and obtain the optimal parameters. The correlation function is constructed as follows: a mathematical model is constructed to characterize the correspondence between image HSV values, average temperature, maximum temperature and fault category. The mathematical model is trained based on the image HSV values ​​of the fault area and the average and maximum temperatures of the image location of the fault area within a historical time period to obtain the correlation function.

2. The method according to claim 1, characterized in that, The training process of the fault detection model includes: Images of faulty photovoltaic modules from the photovoltaic module database are input into the fault detection model for training. Training stops once the preset number of training iterations is reached, resulting in a pre-trained fault detection model.

3. The method according to claim 2, characterized in that, The network structure of the fault detection model includes: an input layer, a backbone network, a neck network, and a detection module; The input layer is used to preprocess the input image and enhance its features, outputting a first feature map. The backbone network consists of multiple CBS modules and multiple CSP modules. The backbone network is used to extract features from the first feature map to obtain a second feature map. The neck network consists of an SPPF module, multiple CBS modules, multiple CSP modules, and multiple sampling layers. The neck network is used to extract features from the second feature map at multiple scales to obtain feature information at different scales. The detection module is used to process the feature information to obtain the detection result.

4. The method according to claim 3, characterized in that, The SPPF module of the fault detection model uses a dense link structure to fuse the feature information extracted by the CBS module with the feature information extracted by each subsequent max pooling layer. The fused information is then fused with multiple fusion modules of the CBS module to reuse the feature information of each module.

5. The method according to claim 4, characterized in that, The network structure of the SPPF module of the fault detection model includes four layers: the first layer includes a CBS module and three max pooling layers; the second layer includes three fusion modules; the third layer includes a fusion module; and the fourth layer includes a CBS module. The first fusion module in the second layer is used to superimpose the first output data of the CBS module in the first layer with the second output data of the first pooling layer in the first layer in the channel dimension to obtain the first data; The second fusion module in the second layer is used to superimpose the first output data and the third output data of the second pooling layer in the first layer in the channel dimension to obtain the second data; The third fusion module in the second layer is used to superimpose the first output data, the fourth output data of the third pooling layer in the first layer, and the first data in the channel dimension to obtain the third data; The fusion module in the third layer is used to superimpose the first output data, the first data, the second data, and the third data in the channel dimension to obtain the fourth data.

6. The method according to claim 1, characterized in that, The step of using the photovoltaic module fault category identification model to determine the fault area and obtain a second fault category includes: The HSV value, average temperature, and maximum temperature of the image corresponding to the fault area are input into the photovoltaic module fault category identification model, and a second fault category is output based on the correlation function between the image HSV value, average temperature, maximum temperature and fault category.

7. The method according to claim 1, characterized in that, The step of selecting the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions includes: Obtain first fault category information for the first fault category and second fault category information for the second fault category; If the first fault category information and the second fault category information are the same, then the first fault category or the second fault category is set as the final fault category; If the first fault category information and the second fault category information are different, then obtain the fault level values ​​corresponding to the first fault category and the second fault category, and take the fault category corresponding to the fault level value with the larger value as the final fault category.

8. A photovoltaic module fault detection device, characterized in that, The device includes: An acquisition module is used to acquire infrared images of a photovoltaic power station, wherein the infrared images show photovoltaic modules that have malfunctions; The detection module is used to detect the infrared image using a pre-trained fault detection model to obtain the fault area and a first fault category. The fault detection model uses a dense link structure to enhance feature reuse. The first fault category includes at least one of the following: string open circuit, string short circuit, component missing, junction box failure, breakage, obstruction of hot spot and hot spot. The discrimination module is used to discriminate the fault area using a pre-trained fault category recognition model to obtain a second fault category. The fault category recognition model is constructed based on the pre-acquired correlation function between the image HSV value, average temperature, and maximum temperature of the fault area and the fault category. The selection module is used to select the final fault category of the photovoltaic module from the first fault category and the second fault category according to preset selection conditions; The construction and training process of the fault category recognition model includes: extracting fault area images from infrared images; calculating the HSV value of the fault area image from the fault area image using OpenCV software; collecting temperature information of all locations in the fault area image using infrared acquisition software; calculating the average temperature and maximum temperature of the fault area based on the temperature information; constructing a functional relationship between the HSV value, the average temperature, and the maximum temperature; constructing a mathematical model corresponding to the functional relationship as the fault category recognition model; using the HSV value, the average temperature, and the maximum temperature as inputs to the fault category recognition model, and using the real fault category as the label, training the model to obtain the optimal parameters. The correlation function is constructed as follows: a mathematical model is constructed to characterize the correspondence between image HSV values, average temperature, maximum temperature and fault category. The mathematical model is trained based on the image HSV values ​​of the fault area and the average and maximum temperatures of the image location of the fault area within a historical time period to obtain the correlation function.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the photovoltaic module fault detection method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the photovoltaic module fault detection method according to any one of claims 1-7.

Images