Fault diagnosis method and device for photovoltaic equipment and electronic equipment

By preprocessing and feature extraction of photovoltaic equipment images, combined with VGGnet and XGBoost technologies, the rapid and accurate diagnosis of photovoltaic equipment failures is achieved, the problem of low fault detection efficiency of photovoltaic system is solved, and the accuracy and efficiency of fault detection are improved.

CN118968132APending Publication Date: 2024-11-15STATE NUCLEAR ELECTRIC POWER PLANNING DESIGN & RES INST CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202410975882.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

During operation, photovoltaic systems often fail due to equipment aging and environmental factors, which affects operating efficiency and safety, and lacks fast and accurate fault diagnosis methods.

Method used

By acquiring the photovoltaic equipment images, inputting the feature extraction network and fault classification network of the fault diagnosis model after preprocessing, using technologies such as VGGnet and XGBoost to perform feature extraction and fault classification, and quickly and accurately determine the fault type of photovoltaic equipment.

Benefits of technology

It realizes rapid and accurate diagnosis of photovoltaic equipment faults, improves fault detection efficiency and accuracy, and reduces the cost of troubleshooting.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968132A_ABST
    Figure CN118968132A_ABST
Patent Text Reader

Abstract

The invention provides a fault diagnosis method and device for photovoltaic equipment and electronic equipment, and the method comprises the steps: obtaining a first photovoltaic equipment image which is obtained through the shooting of the photovoltaic equipment to be subjected to fault diagnosis; preprocessing the first photovoltaic equipment image to obtain a second photovoltaic equipment image; inputting the second photovoltaic equipment image into a feature extraction network of a fault diagnosis model, and performing feature extraction on the second photovoltaic equipment image through the feature extraction network to obtain feature information of the second photovoltaic equipment image; and inputting the feature information into a fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain a target fault type of the photovoltaic equipment. Therefore, the fault type of the photovoltaic equipment can be quickly and accurately determined.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of photovoltaic technology, and in particular to a fault diagnosis method, device and electronic equipment for photovoltaic equipment. Background Art

[0002] With the rapid development of renewable energy, photovoltaic technology has been widely used as an efficient and environmentally friendly way of power generation. However, during the operation of photovoltaic systems, various faults often occur due to equipment aging, environmental factors and other reasons, which seriously affect the operating efficiency and safety of photovoltaic systems. Therefore, how to achieve fast and accurate photovoltaic fault diagnosis is an urgent need in the current photovoltaic technology field. Summary of the invention

[0003] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0004] To this end, the first objective of the present application is to provide a fault diagnosis method for a photovoltaic device.

[0005] A second objective of the present application is to provide a fault diagnosis device for a photovoltaic device.

[0006] The third objective of the present application is to provide an electronic device.

[0007] A fourth objective of the present application is to provide a computer-readable storage medium.

[0008] A fifth object of the present application is to provide a computer program product.

[0009] To achieve the above-mentioned purpose, the first aspect of the present application proposes a fault diagnosis method for photovoltaic equipment, including: acquiring a first photovoltaic equipment image, wherein the first photovoltaic equipment image is obtained by photographing the photovoltaic equipment to be fault diagnosed; preprocessing the first photovoltaic equipment image to obtain a second photovoltaic equipment image; inputting the second photovoltaic equipment image into a feature extraction network of a fault diagnosis model to perform feature extraction on the second photovoltaic equipment image through the feature extraction network to obtain feature information of the second photovoltaic equipment image; inputting the feature information into a fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain a target fault type of the photovoltaic equipment.

[0010] To achieve the above-mentioned purpose, the second aspect of the present application proposes a fault diagnosis device for photovoltaic equipment, including: an acquisition module, used to acquire a first photovoltaic equipment image, wherein the first photovoltaic equipment image is obtained by photographing the photovoltaic equipment to be fault diagnosed; a processing module, used to preprocess the first photovoltaic equipment image to obtain a second photovoltaic equipment image; a feature extraction module, used to input the second photovoltaic equipment image into a feature extraction network of a fault diagnosis model, so as to extract features from the second photovoltaic equipment image through the feature extraction network, and obtain feature information of the second photovoltaic equipment image; a classification module, used to input the feature information into a fault classification network of the fault diagnosis model, so as to classify the fault through the fault classification network, and obtain the target fault type of the photovoltaic equipment.

[0011] To achieve the above-mentioned purpose, the third aspect of the embodiment of the present application proposes an electronic device, comprising: a processor, and a memory connected to the processor for communication; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the fault diagnosis method of the photovoltaic device as described in the first aspect of the embodiment of the present application.

[0012] To achieve the above-mentioned purpose, the fourth aspect of the embodiment of the present application proposes a computer-readable storage medium, in which computer execution instructions are stored. When the computer execution instructions are executed by the processor, they are used to implement the fault diagnosis method of the photovoltaic equipment as described in the first aspect of the embodiment of the present application.

[0013] To achieve the above-mentioned purpose, the fifth aspect of the present application proposes a computer program product, including a computer program, which, when executed by a processor, implements the fault diagnosis method of the photovoltaic device as described in the first aspect of the embodiment of the present application.

[0014] The technical solution provided by this application brings at least the following beneficial effects:

[0015] The present application obtains a first photovoltaic device image, wherein the first photovoltaic device image is obtained by photographing the photovoltaic device to be diagnosed for a fault; the first photovoltaic device image is preprocessed to obtain a second photovoltaic device image; the second photovoltaic device image is input into a feature extraction network of a fault diagnosis model to extract features of the second photovoltaic device image through the feature extraction network to obtain feature information of the second photovoltaic device image; the feature information is input into a fault classification network of a fault diagnosis model to classify faults through the fault classification network to obtain a target fault type of the photovoltaic device. Thus, the fault type of the photovoltaic device can be determined quickly and accurately.

[0016] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0018] Figure 1 A schematic diagram of a flow chart of a fault diagnosis method for photovoltaic equipment provided in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of the structure of the fault diagnosis model provided in the embodiment of the present application;

[0020] Figure 3 It is a structural diagram of the VGG19 model in the related art;

[0021] Figure 4 A schematic diagram of fault classification based on a fault classification network provided in an embodiment of the present application;

[0022] Figure 5 A schematic diagram of the creation process of the XGboost classifier;

[0023] Figure 6 A schematic diagram of the structure of a fault diagnosis device for photovoltaic equipment provided in an embodiment of the present application;

[0024] Figure 7 It is a block diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0025] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0026] The following describes a fault diagnosis method, apparatus, electronic device, and readable storage medium for photovoltaic equipment according to embodiments of the present application with reference to the accompanying drawings.

[0027] Figure 1A flowchart of a fault diagnosis method for a photovoltaic device provided in an embodiment of the present application. It should be noted that the fault diagnosis method for a photovoltaic device provided in an embodiment of the present application is performed by a fault diagnosis device for the photovoltaic device. The fault diagnosis device for the photovoltaic device is referred to as the fault diagnosis device hereinafter. The fault diagnosis device may be an electronic device or may be configured in an electronic device, such as a fault diagnosis software configured in the electronic device, and the present application does not limit this.

[0028] like Figure 1 As shown, the fault diagnosis method of the photovoltaic device includes the following steps:

[0029] Step 101 : acquiring a first photovoltaic device image, wherein the first photovoltaic device image is obtained by photographing a photovoltaic device to be diagnosed for a fault.

[0030] The photovoltaic equipment may include any component or device in a solar photovoltaic system, and this application does not impose any limitation on this.

[0031] In some embodiments, a drone or other camera equipped with a camera can be used to photograph the photovoltaic device to be diagnosed for fault, so as to obtain a first photovoltaic device image. Thus, the fault diagnosis device can obtain the first photovoltaic device image from the drone or other camera, and based on the first photovoltaic device image, perform fault diagnosis on the photovoltaic device through the following steps 102-104.

[0032] Step 102: preprocess the first photovoltaic device image to obtain a second photovoltaic device image.

[0033] In some embodiments, step 102 may be implemented by:

[0034] Based on the threshold segmentation technology, the first photovoltaic device image is segmented to eliminate the background in the first photovoltaic device image to obtain a second photovoltaic device image.

[0035] The threshold segmentation technology may include Otsu threshold segmentation or other threshold segmentation technologies, which are not limited in the present application.

[0036] Taking Otsu threshold segmentation as an example, the following method can be used to segment the first photovoltaic device image: the number of pixels of each gray level in the first photovoltaic device image is counted to form a gray histogram, and based on the gray histogram, the average gray value of all pixels in the first photovoltaic device image is calculated, and each possible threshold is tried from the minimum gray level to the maximum gray level. For each threshold, the first photovoltaic device image is divided into two categories: foreground and background, and then the inter-class variance of the two categories is calculated. From all possible thresholds, the threshold that produces the largest inter-class variance is selected as the optimal threshold, and the first photovoltaic device image is converted into a binary image using the found optimal threshold, that is, pixels below the threshold are marked as background, and pixels above the threshold are marked as foreground. The binary image is the second photovoltaic device image obtained by preprocessing the first photovoltaic device image. Among them, the inter-class variance is the square of the difference between the means of the two categories multiplied by the respective class probabilities.

[0037] The first photovoltaic device image is preprocessed by using a threshold segmentation technology in a simple manner, thereby enabling the preprocessing of the first photovoltaic device image to be achieved at a relatively low computational cost.

[0038] Step 103: input the second photovoltaic device image into the feature extraction network of the fault diagnosis model, so as to extract features of the second photovoltaic device image through the feature extraction network to obtain feature information of the second photovoltaic device image.

[0039] Step 104 , inputting the characteristic information into the fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain the target fault type of the photovoltaic device.

[0040] The fault diagnosis model is a neural network model capable of realizing fault diagnosis or identification, and may include a feature extraction network for feature extraction and a fault classification network connected to the feature extraction network for fault classification.

[0041] The input of the feature extraction network is the image, and the output is the feature information of the image. The input of the fault classification network is the feature information, and the output is the fault type obtained by fault diagnosis. Thus, the second photovoltaic device image corresponding to the photovoltaic device is input into the feature extraction network of the fault diagnosis model, and the feature extraction network can be used to extract the features of the second photovoltaic device image to obtain the feature information of the second photovoltaic device image. The feature information is input into the fault classification network, and the fault classification network can be used to perform fault classification to obtain the target fault type of the photovoltaic device.

[0042] For example, the feature extraction network may be a VGGnet (Visual Geometry Group Network) or other types of neural networks with feature extraction capabilities, and the fault classification network may be an XGboost (Extreme Gradient Boosting) classifier or other types of neural networks with classification capabilities, and this application does not limit this. Among them, VGGnet may include VGG16, VGG19, etc.

[0043] Among them, VGGnet, as a classic convolutional neural network model in deep learning, has attracted widespread attention for its excellent feature extraction capabilities. VGGnet can obtain the same receptive field as a single large convolution kernel by stacking small convolution kernels, and also uses the nonlinear transformation of multiple activation functions to obtain more expressive features than the features obtained by a single nonlinear transformation. VGGnet can extract deep features in images by stacking multiple convolution layers and pooling layers, and has excellent performance for tasks such as image classification and target detection. In the field of photovoltaic fault diagnosis, VGGNet can be used to extract fault features in photovoltaic equipment images, providing strong support for subsequent fault diagnosis. XGBoost, as an efficient ensemble learning algorithm, performs well in solving classification and regression problems. XGBoost is an ensemble learning algorithm based on gradient boosting decision trees, which has the advantages of high efficiency, accuracy, and easy parameter adjustment. The algorithm can effectively improve classification accuracy and generalization ability by constructing and integrating multiple weak classifiers. In the field of photovoltaic fault diagnosis, XGBoost can be used to classify and identify the extracted fault features to achieve rapid diagnosis of photovoltaic faults.

[0044] That is, in the embodiment of the present application, the second photovoltaic device image obtained after preprocessing can be input into VGGnet to extract features of the second photovoltaic device image to obtain representative high-dimensional features, and the high-dimensional features are input into the XGBoost model for fault classification to obtain the target fault type of the photovoltaic device. By combining VGGNet and XGBoost for fault diagnosis, the advantages of both can be fully utilized to achieve rapid and accurate diagnosis of photovoltaic device faults.

[0045] In some embodiments, the fault diagnosis model after training can be obtained by jointly training the feature extraction network and the fault classification network in the fault diagnosis model, and then the trained fault diagnosis model is used to perform the process of step 103 and step 104. The process of jointly training the feature extraction network and the fault classification network in the fault diagnosis model can include: obtaining multiple sample photovoltaic device images, wherein each sample photovoltaic device image is obtained by photographing the photovoltaic device, and each sample photovoltaic device image is labeled with the fault type corresponding to the photovoltaic device when the photovoltaic device is photographed; preprocessing each sample photovoltaic device image to obtain the corresponding preprocessed sample photovoltaic device image; based on the preprocessed sample photovoltaic device image, the feature extraction network and the fault classification network of the fault diagnosis model are trained. The feature extraction network can be used to extract representative high-dimensional features in the preprocessed sample photovoltaic device image, and then each high-dimensional feature is used as the input of the fault classification network, and multiple rounds of iterative training are performed to obtain a trained fault diagnosis model.

[0046] In addition, the preprocessed sample photovoltaic equipment images can be divided into a test set and a training set according to a preset ratio, such as 2:3, as training data for training the fault diagnosis model. When training the fault diagnosis model, the learning rate r can be set to 0.01, the number of iterations variable is Num, and the training data is input. When the classification loss function loss of the model on the training data is less than 0.01 or equal to the number of iterations variable Num, the model parameters at this time are saved to obtain a trained fault diagnosis model with photovoltaic fault diagnosis capabilities.

[0047] Among them, when the fault classification network is an XGBoost model, the XGBoost model includes multiple decision trees. In each round of iteration, part of the data in the training set can be randomly selected to input into the XGboost classifier to train a decision tree. The construction process of each decision tree uses the prediction results of the previous decision tree.

[0048] In an embodiment of the present application, a first photovoltaic device image is acquired, wherein the first photovoltaic device image is acquired by photographing a photovoltaic device to be subjected to fault diagnosis, the first photovoltaic device image is preprocessed to acquire a second photovoltaic device image, the second photovoltaic device image is input into a feature extraction network of a fault diagnosis model, feature extraction is performed on the second photovoltaic device image through the feature extraction network to acquire feature information of the second photovoltaic device image, the feature information is input into a fault classification network of the fault diagnosis model, fault classification is performed through the fault classification network to acquire a target fault type of the photovoltaic device, and the fault type of the photovoltaic device can be determined quickly and accurately.

[0049] The fault diagnosis model provided in the embodiment of the present application is further described below in conjunction with the accompanying drawings.

[0050] Figure 2 A schematic diagram of the structure of the fault diagnosis model provided in the embodiment of the present application.

[0051] refer to Figure 2 The fault diagnosis model includes a feature extraction network 201 and a fault classification network 202. The feature extraction network 201 includes a feature extraction sub-network 2011 and a fully connected sub-network 2012 connected to the feature extraction sub-network 2011, and the fault classification network 202 is connected to the feature extraction sub-network 201 and the fully connected sub-network 202. Figure 2 The fault classification network 202 is taken as an example to illustrate.

[0052] Wherein, the feature extraction subnetwork 2011 is used to extract local features in the second photovoltaic device image;

[0053] A fully connected sub-network 2012, used for integrating local features input into the fully connected sub-network 2012 to obtain global features in the second photovoltaic device image;

[0054] The fault classification network 202 is used to perform fault classification based on the local features output by the feature extraction subnetwork 2011 and the global features output by the fully connected subnetwork 2012 to obtain the target fault type of the photovoltaic equipment.

[0055] Specifically, the second photovoltaic device image can be input into the feature extraction subnetwork 2011 to extract local features in the second photovoltaic device image through the feature extraction subnetwork 2011, and the local features output by the feature extraction subnetwork 2011 can be input into the fully connected subnetwork 2012 to integrate the local features through the fully connected subnetwork 2012 to obtain global features in the second photovoltaic device image. The fault classification network 202 is connected to the feature extraction subnetwork 2011 and the fully connected subnetwork 2012, and can perform fault classification based on the local features output by the feature extraction subnetwork 2011 and the global features output by the fully connected subnetwork 2012 to obtain the target fault type of the photovoltaic device.

[0056] In some embodiments, the feature extraction subnetwork includes M sequentially connected feature extraction structures, each feature extraction structure is used to extract local features in the second photovoltaic device image, and the i-th feature extraction structure is connected to the i+2-th feature extraction structure, i is an integer greater than or equal to 1 and less than or equal to M-2, and the M-1-th feature extraction structure is connected to the fully connected subnetwork; wherein,

[0057] The (i+2)th feature extraction structure is used to extract features based on the local features output by the (i)th feature extraction structure and the (i+1)th feature extraction structure;

[0058] The fully connected sub-network 2012 is used to integrate the local features output by the M-1th feature extraction structure and the Mth feature extraction structure to obtain the global features in the second photovoltaic device image.

[0059] In some embodiments, each feature extraction structure includes a plurality of convolution kernels and a maximum pooling layer connected in sequence; wherein,

[0060] Each convolution kernel is used to extract local features in the second photovoltaic device image;

[0061] The maximum pooling layer is used to compress the local features of the input maximum pooling layer to obtain compressed features.

[0062] Among them, the number of convolution kernels included in different feature extraction structures can be the same or different, and this application does not impose any limitation on this.

[0063] In some embodiments, the fully connected sub-network 2012 may include a plurality of sequentially connected fully connected layers.

[0064] For example, refer to Figure 2 , taking the fault classification network as an XGBoost classifier, M = 5 as an example, the feature extraction subnetwork may include 5 sequentially connected feature extraction structures, each of which is used to extract local features in the second photovoltaic device image. Among them, the first feature extraction structure includes two sequentially connected convolutional layers ( Figure 2 The first and second Conv in ) and a max pooling layer ( Figure 2 The first Pool / 2 in the convolution layer, each convolution layer includes 64 3*3 convolution kernels; the second feature extraction structure includes two convolution layers connected in sequence ( Figure 2 The 3rd and 4th Conv in ) and a max pooling layer ( Figure 2 The second Pool / 2 in the convolution layer includes 128 3*3 convolution kernels; the third feature extraction structure includes three convolution layers connected in sequence ( Figure 2 The 5th to 7th Conv in ) and a max pooling layer ( Figure 2 The third Pool / 2 in the convolution layer includes 256 3*3 convolution kernels; the fourth feature extraction structure includes three convolution layers connected in sequence ( Figure 2 The 8th to 10th Conv in ) and a max pooling layer ( Figure 2 The fourth Pool / 2 in the convolution layer includes 512 3*3 convolution kernels; the fifth feature extraction structure includes three convolution layers connected in sequence ( Figure 211th to 13th Conv in ) and a max pooling layer ( Figure 2 The 5th Pool / 2 in , each convolutional layer includes 512 3*3 convolution kernels. The fully connected subnetwork 2012 includes 3 maximum pooling layers.

[0065] In addition, the output of the first feature extraction structure (i.e., the output of the first Pool / 2) is connected to the input of the third feature extraction structure (i.e., the input of the fifth Conv), the output of the second feature extraction structure (i.e., the output of the second Pool / 2) is connected to the input of the fourth feature extraction structure (i.e., the input of the eighth Conv), the output of the third feature extraction structure (i.e., the output of the third Pool / 2) is connected to the input of the fifth feature extraction structure (i.e., the input of the eleventh Conv), and the output of the fourth feature extraction structure (i.e., the output of the fourth Pool / 2) is connected to the input of the fully connected sub-network 2012.

[0066] Correspondingly, the first feature extraction structure is used to extract features from the second photovoltaic device image and output local features in the second photovoltaic device image; the second feature extraction structure is used to extract features based on the first feature extraction structure and output another local feature in the second photovoltaic device image; the third feature extraction structure is used to extract features based on the local features output by the first feature extraction structure and the second feature extraction structure, and output another local feature in the second photovoltaic device image; the fourth feature extraction structure is used to extract features based on the local features output by the second feature extraction structure and the third feature extraction structure, and output another local feature in the second photovoltaic device image; the fifth feature extraction structure is used to extract features based on the local features output by the third feature extraction structure and the fourth feature extraction structure, and output another local feature in the second photovoltaic device image; the fully connected subnetwork 2012 is used to integrate the local features output by the four feature extraction structures and the fifth feature extraction structure to obtain global features in the second photovoltaic device image.

[0067] Among them, the dimensions of local features output by different feature extraction structures are different.

[0068] It should be noted that after each convolutional layer, the operation of the ReLU activation function can also be included.

[0069] For example, assuming that the size of the second photovoltaic device image input by the first feature extraction structure is 224*224*3 (width*height*number of channels), after two convolutions + ReLu with 64 3*3 convolution kernels, the size of the feature map after convolution is 224*224*64, and then after the maximum pooling layer (the size of the maximum pooling layer is 2*2, the effect is to halve the image size), the size of the feature map after pooling becomes 112*112*64. Furthermore, the feature map output by the first feature extraction structure is convolved twice + ReLu with 128 3*3 convolution kernels, the size of the feature map after convolution becomes 112*112*128, and then after the 2*2 maximum pooling layer, the size of the feature map after pooling becomes 56*56*128. Furthermore, the feature maps output by the first and second feature extraction structures are convolved three times with 256 3*3 convolution kernels + ReLu, and the size of the feature maps after convolution becomes 56*56*25, and then pooled by a 2*2 maximum pooling layer, and the size of the feature maps after pooling becomes 28*28*256. Furthermore, the feature maps output by the second and third feature extraction structures are convolved three times with 512 3*3 convolution kernels + ReLu, and the size of the feature maps after convolution becomes 28*28*512, and then pooled by a 2*2 maximum pooling layer, and the size of the feature maps after pooling becomes 14*14*512. Furthermore, the feature maps output by the third and fourth feature extraction structures are convolved three times with 512 3*3 convolution kernels + ReLu, and the size of the feature map after convolution becomes 14*14*512, and then pooled by a 2*2 maximum pooling layer, and the size of the feature map after pooling becomes 7*7*512. Furthermore, the feature maps output by the fourth and fifth feature extraction structures can be integrated through three layers of 1*1*4096 fully connected layers to obtain the global features in the second photovoltaic device image. Furthermore, the output of the fifth feature extraction structure and the fully connected subnetwork 202 can be input into the XGBoost classifier, so that the XGBoost classifier can perform fault classification based on the input information to obtain the target fault type of the photovoltaic device.

[0070] It is understandable that in some embodiments, the VGG16 model in the related art can be improved to form a residual structure at the bottleneck of each stage transmission, and the following can be obtained: Figure 2 The feature extraction network in the fault diagnosis model shown is used to reduce the information loss in the forward pass and the impact of unpredictable gradient changes in the back propagation process, improve the convergence efficiency of the model, and thus improve the training efficiency.

[0071] In some embodiments, similar concepts can also be used to improve the VGG19 model in the related art to obtain the feature extraction network in the fault diagnosis model of the embodiment of the present application. The structure of the VGG19 model is Figure 3 The form shown.

[0072] Among them, the VGG19 model includes two convolutional layers connected in sequence ( Figure 3 Conv1_1 and Conv1_2 in ) and a max pooling layer ( Figure 3 maxpool1 in ), 2 convolutional layers ( Figure 3 Conv2_1 and Conv2_2 in ) and a max pooling layer ( Figure 3 maxpool2 in ), 4 convolutional layers ( Figure 3 Conv3_1, Conv3_2, Conv3_3, Conv3_4) and a maximum pooling layer ( Figure 3 maxpool3 in ), 4 convolutional layers ( Figure 3 Conv4_1, Conv4_2, Conv4_3, Conv4_4) and a maximum pooling layer ( Figure 3 maxpool4 in), 4 convolutional layers ( Figure 3 Conv5_1, Conv5_2, Conv5_3, Conv5_4) and a maximum pooling layer ( Figure 3 maxpool5 in ), two fully connected layers ( Figure 3 FC1 and FC2 in ) and the normalization layer ( Figure 3 in .

[0073] In an embodiment of the present application, the normalization layer of the VGG19 model can be replaced with the fault classification network in the present application, and the output of Maxpool1 is connected to the input of Conv3_1, the output of Maxpool2 is connected to the input of Conv4_1, the output of Maxpool3 is connected to the input of Conv5_1, the output of Maxpool4 is connected to the input of FC1, and the output of Maxpool5 is connected to the input of the fault classification network, thereby forming a residual structure at the bottleneck of each stage of transmission to reduce the information loss in the forward transmission process and the impact of unpredictable gradient changes in the back propagation process, improve the convergence efficiency of the model, and thus improve the training efficiency.

[0074] Reference below Figure 4 , the process of fault classification using the fault classification network in the fault diagnosis model provided in the embodiment of the present application is described.

[0075] refer to Figure 4, the fault classification network can be an XGBoost classifier, including multiple decision trees ( Figure 4 Take a fault classification network including k decision trees, namely decision tree-1, decision tree-2, ..., decision tree-k, as an example for illustration). The number of nodes included in different decision trees and the structure of the decision trees may be the same or different, and the present application does not limit this.

[0076] refer to Figure 4 , the feature information can be input into each decision tree to make a fault category decision through each decision tree, and the decision results output by each decision tree are obtained, wherein the decision results output by each decision tree include multiple scores, each score corresponds to a fault type, indicating the probability of the photovoltaic equipment having the corresponding fault type, and then based on the weights corresponding to each decision tree (the weights can be determined according to the importance of each decision tree), the multiple scores output by each decision tree are weighted and summed according to the corresponding fault type to obtain the total score corresponding to each fault type, so that the fault type with the highest total score is used as the target fault type of the photovoltaic equipment.

[0077] The following is an explanation of the creation process of the XGboost classifier.

[0078] (1) XGboost is an additive operation formula composed of K base models. In the embodiment of the present application, a classification regression tree is selected to predict the target value:

[0079]

[0080] where f t is the tth base model, is the predicted value of the i-th sample, f t (x i ) represents the predicted value of the t-th base model for sample i. The i-th sample can be the i-th preprocessed sample photovoltaic device image.

[0081] (2) The loss function can be obtained by and the true value y i To express:

[0082]

[0083] Where n is the number of samples. The prediction accuracy of the model is determined by the deviation and variance of the model. The loss function represents the deviation of the model. If you want a small variance, you need a simple model. The objective function consists of the loss function L of the model and the regularization term Ω that suppresses the complexity of the model. The objective function is:

[0084]

[0085] Where: l is a differentiable convex function used to measure the predicted classification value and the true value y i The difference between is called the loss function; the Ω function represents the regularization term to prevent the model from overfitting.

[0086] (3) The boosting model is a forward addition. Taking the model of step t as an example, the model performs forward addition on the i-th sample x. i The prediction is:

[0087]

[0088] in The predicted value given by the model at step t-1 is a known constant, f t (x i ) is the predicted value of the new model that needs to be added in this iteration. At this time, the objective function can be written as:

[0089]

[0090] Finding the optimal objective function at this time is equivalent to solving f t (x i ).

[0091] (4) According to Taylor's formula, the objective function is Carrying out Taylor's second-order expansion, we can obtain:

[0092]

[0093] where g i is the first-order derivative of the loss function, h i is the second-order derivative of the loss function, which is a known number that can be found, namely:

[0094]

[0095] Since at step t is a known value, so is a constant and has no effect on the optimization of the function, so it can be removed and the objective function becomes:

[0096]

[0097] (5) Define the decision tree as f t (x) = w q(x) , x is a sample, where q(x) represents the leaf node where the sample is located, and w q It represents the leaf node value w, so w q(x)represents the value w (i.e., predicted value) of each sample. The complexity of the decision tree can be composed of the number of leaves T. The fewer leaf nodes, the simpler the model. In addition, the leaf nodes should not have too high weights w. The regularization term of the objective function (i.e., the complexity of the model) can be defined as:

[0098]

[0099] Among them, γ and λ are coefficients that control the proportion of the two parts.

[0100] Definition I j ={iq(x i )=j} is the sample set of the jth leaf node, so the objective function can be written as:

[0101]

[0102] definition Then the objective function is:

[0103]

[0104] G j and H j is the result obtained in the previous t-1 steps. Its value is known and can be regarded as a constant. Only the leaf node w of the last tree j Uncertain, the objective function is w j Find the first-order derivative and set it equal to 0, then you can get the weight corresponding to the leaf node j:

[0105]

[0106] Objective function:

[0107]

[0108] In the process of growing a decision tree, X represents the set of samples. Figure 5 , a greedy algorithm can be used to find the optimal split point of the leaf node:

[0109] 1. Starting from a decision tree with a depth of 0, enumerate all available features for each leaf node (where, for the root node, the available features include all sample features, i.e. Figure 5 Neutron feature-1, sub-feature-2, sub-feature-3, ..., sub-feature-q. For the second layer, the available features include all sample features except the split features of the nodes in the previous layer, and the same applies to other layers);

[0110] Second, for each feature, the training samples belonging to the node are arranged in ascending order of importance according to the feature value, the best split point of the feature is determined by linear scanning, and the splitting benefit of the feature is recorded;

[0111] 3. Select the feature with the largest benefit as the split feature, use the best split point (also called the best branch point) of the feature as the split position, split two new leaf nodes on the left and right at the node, and associate the corresponding sample set with each new node;

[0112] 4. Go back to step 1 and execute recursively until specific conditions are met to obtain a decision tree.

[0113] Among them, for the decision tree, the process of calculating the splitting benefit to determine whether to split and determine the candidate splitting points is:

[0114] Assuming that the feature splitting is completed at a certain node, the objective function before splitting is:

[0115]

[0116] Among them G L , G R are the first-order derivatives on the left and right sides of the split node, respectively, H L , H R are the sum of the second-order derivatives on the left and right sides of the split node, respectively.

[0117] The objective function after splitting is:

[0118]

[0119] For the objective function, the profit after splitting is:

[0120]

[0121] Among them, the first item on the right represents the structural score of the left subtree of the leaf node after the split, the second item on the right represents the structural score of the right subtree of the leaf node after the split, the third item represents the structural score of the leaf node itself if it is not split, and the last item represents the complexity (the regularization coefficient of the newly added leaf).

[0122] Following the process from step 1 to step 4 above, after obtaining the first decision tree, decision tree-1, we can determine the residual between the predicted value and the true value of sample X by decision tree-1, and use this residual as part of the loss function L1 to train the second decision tree, decision tree-2, by optimizing L1. This process can be expressed as f1(X,L1). Following the process from step 1 to step 4 above again, after obtaining the second decision tree, decision tree-2, we can determine the residual between the predicted value and the true value of sample X by decision tree-2, and use this residual as part of the loss function L2 to train the third decision tree, decision tree-3, by optimizing L2. This process can be expressed as f2(X,L2). By iterating the above process for k steps, we can obtain the kth decision tree, decision tree-k.

[0123] Thus, the creation of XGBoost classifier is completed.

[0124] In order to implement the above embodiment, the embodiment of the present application also proposes a fault diagnosis device for photovoltaic equipment.

[0125] Figure 6 A schematic structural diagram of a fault diagnosis device for photovoltaic equipment provided in an embodiment of the present application.

[0126] like Figure 6 As shown, the fault diagnosis device 600 of the photovoltaic device includes:

[0127] An acquisition module 610 is used to acquire a first photovoltaic device image, wherein the first photovoltaic device image is obtained by photographing the photovoltaic device to be diagnosed for a fault;

[0128] A processing module 620 is used to pre-process the first photovoltaic device image to obtain a second photovoltaic device image;

[0129] A feature extraction module 630 is used to input the second photovoltaic device image into a feature extraction network of the fault diagnosis model to extract features of the second photovoltaic device image through the feature extraction network to obtain feature information of the second photovoltaic device image;

[0130] The classification module 640 is used to input the characteristic information into the fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain the target fault type of the photovoltaic device.

[0131] Optionally, the feature extraction network includes a feature extraction subnetwork and a fully connected subnetwork connected to the feature extraction subnetwork, and the fault classification network is connected to the feature extraction subnetwork and the fully connected subnetwork; wherein,

[0132] a feature extraction subnetwork, for extracting local features in the second photovoltaic device image;

[0133] A fully connected sub-network, used for integrating local features input into the fully connected sub-network to obtain global features in the second photovoltaic device image;

[0134] The fault classification network is used to classify faults based on the local features output by the feature extraction subnetwork and the global features output by the fully connected subnetwork to obtain the target fault type of the photovoltaic equipment.

[0135] Optionally, the feature extraction subnetwork includes M sequentially connected feature extraction structures, each feature extraction structure is used to extract local features in the second photovoltaic device image, and the i-th feature extraction structure is connected to the i+2-th feature extraction structure, i is an integer greater than or equal to 1 and less than or equal to M-2, and the M-1-th feature extraction structure is connected to the fully connected subnetwork; wherein,

[0136] The (i+2)th feature extraction structure is used to extract features based on the local features output by the (i)th feature extraction structure and the (i+1)th feature extraction structure;

[0137] The fully connected sub-network is used to integrate the local features output by the M-1th feature extraction structure and the Mth feature extraction structure to obtain the global features in the second photovoltaic device image.

[0138] Optionally, each feature extraction structure includes a plurality of convolution kernels and a maximum pooling layer connected in sequence; wherein,

[0139] Each convolution kernel is used to extract local features in the second photovoltaic device image;

[0140] The maximum pooling layer is used to compress the local features of the input maximum pooling layer to obtain compressed features.

[0141] Optionally, the fault classification network includes a plurality of decision trees; a classification module, for:

[0142] Input the characteristic information into each decision tree to make a fault category decision through each decision tree, and obtain a decision result output by each decision tree, wherein the decision result output by each decision tree includes multiple scores, each score corresponds to a fault type, indicating the probability of the photovoltaic equipment having the corresponding fault type;

[0143] The multiple scores output by each decision tree are weighted and summed according to the corresponding fault type to obtain the total score corresponding to each fault type;

[0144] The fault type with the highest total score is used as the target fault type of the photovoltaic equipment.

[0145] Optionally, the acquisition module 610 is further used to: acquire a plurality of sample photovoltaic device images, wherein each sample photovoltaic device image is obtained by photographing the photovoltaic device, and each sample photovoltaic device image is labeled with a fault type corresponding to the photovoltaic device when the photovoltaic device is photographed;

[0146] The processing module 620 is further used to: pre-process each sample photovoltaic device image to obtain a corresponding pre-processed sample photovoltaic device image;

[0147] The device 600 further includes a training module, which is used to:

[0148] Based on the preprocessed sample photovoltaic equipment images, the feature extraction network and fault classification network of the fault diagnosis model are trained.

[0149] Optionally, the processing module 620 is configured to:

[0150] Based on the threshold segmentation technology, the first photovoltaic device image is segmented to eliminate the background in the first photovoltaic device image to obtain a second photovoltaic device image.

[0151] It should be noted that the above explanation of the embodiment of the fault diagnosis method for photovoltaic equipment is also applicable to the fault diagnosis device for photovoltaic equipment of this embodiment, which will not be described in detail here.

[0152] Figure 7 A schematic diagram of the structure of an electronic device provided for an embodiment of the present application. Among them, the electronic device 700 in this embodiment is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, 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 processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0153] like Figure 7 As shown, the electronic device 700 includes:

[0154] The memory 701 and the processor 702, a bus 703 connecting different components (including the memory 701 and the processor 702), the memory 701 stores a computer program, and when the processor 702 executes the program, the fault diagnosis method of the photovoltaic device of the embodiment of the present application is implemented.

[0155] Bus 703 represents one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor or a local bus using any of a variety of bus structures. For example, these architectures include but are not limited to Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MAC) bus, Enhanced ISA bus, Video Electronics Standards Association (VESA) local bus and Peripheral Component Interconnect (PCI) bus.

[0156] The electronic device 700 typically includes a variety of electronic device readable media. These media can be any available media that can be accessed by the electronic device 700, including volatile and non-volatile media, removable and non-removable media.

[0157] The memory 701 may also include computer system readable media in the form of volatile memory, such as random access memory (RAM) 704 and / or cache memory 705. The electronic device 700 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the storage system 706 may be used to read and write non-removable, non-volatile magnetic media ( Figure 7 not shown, usually called a "hard drive"). Although Figure 7 Not shown, a disk drive for reading and writing to a removable non-volatile disk (e.g., a "floppy disk"), and an optical disk drive for reading and writing to a removable non-volatile optical disk (e.g., a CD-ROM, a DVD-ROM, or other optical medium) may be provided. In these cases, each drive may be connected to bus 703 via one or more data medium interfaces. Memory 701 may include at least one program product having a set (e.g., at least one) of program modules that are configured to perform the functions of each embodiment of the present application.

[0158] A program / utility 708 having a set (at least one) of program modules 707 may be stored, for example, in the memory 701, such program modules 707 including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The program modules 707 generally perform the functions and / or methods of the embodiments described herein.

[0159] The electronic device 700 may also communicate with one or more external devices 709 (e.g., keyboard, pointing device, display 711, etc.), one or more devices that enable a user to interact with the electronic device 700, and / or any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., network card, modem, etc.). Such communication may be performed through an input / output (I / O) interface 712. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 713. Figure 7 As shown, the network adapter 713 communicates with other modules of the electronic device 700 via the bus 703. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0160] The processor 702 executes various functional applications and data processing by running the programs stored in the memory 701 .

[0161] It should be noted that the implementation process and technical principles of the electronic device of this embodiment refer to the aforementioned explanation of the fault diagnosis method of the photovoltaic device in the embodiment of the present application, and will not be repeated here.

[0162] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0163] In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0164] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0165] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0166] The present application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, the present disclosure is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0167] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0168] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0169] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0170] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0171] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0172] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0173] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0174] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A fault diagnosis method for photovoltaic equipment, characterized in that: The following steps are involved: Acquire a first photovoltaic device image, wherein the first photovoltaic device image is obtained by photographing the photovoltaic device to be diagnosed for a fault; Preprocessing the first photovoltaic device image to obtain a second photovoltaic device image; Inputting the second photovoltaic device image into a feature extraction network of a fault diagnosis model, so as to extract features of the second photovoltaic device image through the feature extraction network to obtain feature information of the second photovoltaic device image; The characteristic information is input into the fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain the target fault type of the photovoltaic device.

2. The method according to claim 1, characterized in that The feature extraction network includes a feature extraction subnetwork and a fully connected subnetwork connected to the feature extraction subnetwork, and the fault classification network is connected to the feature extraction subnetwork and the fully connected subnetwork; wherein, The feature extraction subnetwork is used to extract local features in the second photovoltaic device image; The fully connected sub-network is used to integrate local features input into the fully connected sub-network to obtain global features in the second photovoltaic device image; The fault classification network is used to perform fault classification based on the local features output by the feature extraction subnetwork and the global features output by the fully connected subnetwork to obtain the target fault type of the photovoltaic device.

3. The method according to claim 2, characterized in that The feature extraction subnetwork includes M sequentially connected feature extraction structures, each of which is used to extract local features in the second photovoltaic device image, and the i-th feature extraction structure is connected to the i+2-th feature extraction structure, i is an integer greater than or equal to 1 and less than or equal to M-2, and the M-1-th feature extraction structure is connected to the fully connected subnetwork; wherein, The (i+2)th feature extraction structure is used to perform feature extraction based on the local features output by the (i)th feature extraction structure and the (i+1)th feature extraction structure; The fully connected sub-network is used to integrate the local features output by the M-1th feature extraction structure and the Mth feature extraction structure to obtain the global features in the second photovoltaic device image.

4. The method according to claim 3, characterized in that Each feature extraction structure includes multiple convolution kernels and maximum pooling layers connected in sequence; Each of the convolution kernels is used to extract local features in the second photovoltaic device image; The maximum pooling layer is used to compress the local features input into the maximum pooling layer to obtain compressed features.

5. The method according to claim 1, characterized in that The fault classification network includes a plurality of decision trees; the fault classification network of the fault diagnosis model inputs the characteristic information to perform fault classification through the fault classification network to obtain the target fault type of the photovoltaic device, including: Inputting the characteristic information into each decision tree to make a fault category decision through each decision tree to obtain a decision result output by each decision tree, wherein the decision result output by each decision tree includes a plurality of scores, each score corresponding to a fault type, indicating the probability of the photovoltaic device having the corresponding fault type; Performing weighted summation on the multiple scores outputted by each decision tree according to the corresponding fault type to obtain a total score corresponding to each fault type; The fault type with the highest corresponding total score is used as the target fault type of the photovoltaic device.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Acquire a plurality of sample photovoltaic device images, wherein each of the sample photovoltaic device images is obtained by photographing the photovoltaic device, and each of the sample photovoltaic device images is labeled with a fault type corresponding to the photovoltaic device when the photovoltaic device is photographed; Preprocessing each of the sample photovoltaic device images to obtain a corresponding preprocessed sample photovoltaic device image; Based on the preprocessed sample photovoltaic device images, the feature extraction network and the fault classification network of the fault diagnosis model are trained.

7. The method according to any one of claims 1 to 5, characterized in that The preprocessing of the first photovoltaic device image to obtain a second photovoltaic device image includes: Based on the threshold segmentation technology, the first photovoltaic device image is segmented to eliminate the background in the first photovoltaic device image to obtain the second photovoltaic device image.

8. A fault diagnosis device for photovoltaic equipment, characterized in that: include: An acquisition module, configured to acquire a first photovoltaic device image, wherein the first photovoltaic device image is obtained by photographing the photovoltaic device to be subjected to fault diagnosis; A processing module, used for preprocessing the first photovoltaic device image to obtain a second photovoltaic device image; a feature extraction module, used for inputting the second photovoltaic device image into a feature extraction network of a fault diagnosis model, so as to extract features of the second photovoltaic device image through the feature extraction network and obtain feature information of the second photovoltaic device image; The classification module is used to input the characteristic information into the fault classification network of the fault diagnosis model to perform fault classification through the fault classification network to obtain the target fault type of the photovoltaic device.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

Citation Information

Patent Citations

  • Malicious traffic detection method, device, equipment and medium based on VGG neural network,

    CN111835769A

  • Fault type identification method and device for weak fault current, equipment and medium

    CN115545085A

  • Photovoltaic panel fault identification method based on cascaded region convolutional neural network

    CN116503308A

  • Photovoltaic panel high-order texture positioning and fault detection method based on multilayer features

    CN117934364A

  • Intelligent classification and identification method and system for furnace conditions of blast furnace

    CN118015348A