Transmission line reliability prediction methods, systems, and equipment based on SENet and EffNet

Abstract

A method, system, and device for predicting the reliability of transmission lines based on SENet and EffNet are disclosed. The method first acquires image data of the target area and constructs a neural network model of the transmission line, and then performs reliability calculation through the neural network model. This invention integrates multi-source data such as images, electrical data, and weather data through comprehensive data analysis methods. It extracts and fuses multi-dimensional features through convolutional layers, fusion modules, and pooling layers of convolutional neural networks to form a comprehensive feature vector. This vector is integrated through fully connected layers and combined with nonlinear transformation and normalization processing to enhance the training efficiency and stability of the model. The feature scale is adjusted through a composite scaling layer. Finally, the reliability prediction of the transmission line is output by the decision layer, which can effectively reflect the real-time status of the transmission line, improve the accuracy of reliability assessment, and solve the problems of neglecting real-time environment and external factors and relying on historical experience in traditional methods.

Description

Technical Field

[0001] This invention relates to a method for predicting the reliability of transmission lines, belonging to the field of transmission line safety technology, and particularly to a method, system, and equipment for predicting the reliability of transmission lines based on SENet and EffNet. Background Technology

[0002] Transmission lines are an important component of the power system, and their reliability directly affects the stable supply of power. Therefore, it is essential to predict the reliability of transmission lines.

[0003] Transmission line reliability prediction is the process of assessing and calculating the reliability of transmission lines in a power system. Its purpose is to determine the probability that a transmission line can operate normally without failure within a certain period. Currently, existing reliability prediction methods typically use the failure rate of transmission lines as a predictive indicator. This relies on historical failure data and statistical analysis, neglecting real-time environmental and external factors, such as the impact of weather conditions on transmission line reliability. Furthermore, once a transmission line fails and is repaired, the underlying cause is usually resolved, making the recurrence of the same failure type relatively unlikely. Therefore, using only the failure rate as the sole indicator for transmission line reliability prediction cannot comprehensively reflect the overall health and reliability of the transmission lines, leading to low accuracy in reliability calculations. Therefore, a more accurate prediction method is urgently needed to more comprehensively reflect the reliability of transmission lines. Summary of the Invention

[0004] The purpose of this invention is to overcome the above-mentioned defects and problems in the prior art and to provide a highly accurate method, system and equipment for predicting the reliability of transmission lines based on SENet and EffNet.

[0005] To achieve the above objectives, the technical solution of this invention is: a transmission line reliability prediction method based on SENet and EffNet, comprising:

[0006] S1. Acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices;

[0007] S2. Construct a CNN model for power transmission lines based on the Keras framework; the CNN model includes convolutional layers, fusion modules, pooling layers, fully connected layers, activation functions, batch normalization layers, composite scaling layers, and decision layers; the fusion module is obtained by fusing the CNN-SENet module and the EfficientNet module, the CNN-SENet module includes the SE module, and the EfficientNet module includes the MBConv module; the Keras framework includes a stitching layer;

[0008] S3. Input the image data into the convolutional layer to obtain the image feature map, and input the image feature map into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature map. Then, input the enhanced features into the pooling layer to obtain the global features.

[0009] The steps for fusing the SE module and the MBConv module are as follows:

[0010] S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution;

[0011] S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution.

[0012] S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module.

[0013] S4. Real-time acquisition of electrical parameters, environmental parameters of transmission lines and weather data of target area, and data splicing of electrical parameters and weather data through splicing layer, and then inputting spliced ​​data and global features into fully connected layer for fusion to obtain comprehensive feature vector;

[0014] S5. The comprehensive feature vector is transformed nonlinearly by the activation function to obtain the activated feature vector. Then, it is input into the batch normalization layer to obtain the normalized feature vector. Next, the normalized feature vector is input into the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector.

[0015] S6. Input the final feature vector into the decision layer for prediction to obtain the transmission line reliability prediction result;

[0016] S7. Map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line using visual elements.

[0017] Step S32 specifically includes:

[0018] S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel;

[0019] S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights;

[0020] The expression for the first fully connected layer is as follows:

[0021] X se1 =ReLU(Dense(X) gap W se1 ));

[0022] Where: X se1 X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap W is the first global feature. se1 These are the weights of the first connection layer;

[0023] The expression for the second fully connected layer is as follows:

[0024] X se2 =σ(Dense(X) se1 W se2 ));

[0025] Where: X se2 W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer;

[0026] S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows:

[0027] X′=X⊙X se2 ;

[0028] Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation;

[0029] S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows:

[0030] Y = W·X′;

[0031] Where: Y is the output feature map after projection convolution, and W is the weight matrix.

[0032] Step S6 specifically includes:

[0033] S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function.

[0034] The expression for the objective function is as follows:

[0035]

[0036] Where: w is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables.

[0037] The expression for the constraint condition is as follows:

[0038] y i (w·x i +b)≥1-ξ i ξ i ≥0, i=1,2,...,n;

[0039] Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term;

[0040] S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability.

[0041] The expression for the decision function is as follows:

[0042] y = f(w·x + b);

[0043] Where: y is the reliability prediction value, f is the activation function, w is the weight vector, x is the input of the decision layer, and b is the bias term;

[0044] S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

[0045] Step S7 specifically includes:

[0046] S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map;

[0047] S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

[0048] The activation function is the Leaky ReLU function.

[0049] A transmission line reliability prediction system based on SENet and EffNet, the system comprising:

[0050] The image data acquisition module is used to acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices.

[0051] A CNN model building module for power transmission lines is provided, used to construct a CNN model for power transmission lines based on the Keras framework. The CNN model includes convolutional layers, a fusion module, pooling layers, fully connected layers, activation functions, batch normalization layers, composite scaling layers, and decision layers. The fusion module is obtained by fusing a CNN-SENet module and an EfficientNet module. The CNN-SENet module includes an SE module, and the EfficientNet module includes an MBConv module. The Keras framework includes a stitching layer.

[0052] The feature enhancement module is used to input image data into the convolutional layer to obtain image feature maps, and then input the image feature maps into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature maps. Finally, the enhanced features are input into the pooling layer to obtain global features.

[0053] The steps for fusing the SE module and the MBConv module are as follows:

[0054] S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution;

[0055] S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution.

[0056] S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module.

[0057] The feature fusion module is used to acquire electrical parameters, environmental parameters and weather data of the transmission line and the target area in real time. The electrical parameters and weather data are stitched together through the stitching layer. Then, the stitched data and global features are input into the fully connected layer for fusion to obtain a comprehensive feature vector.

[0058] The activation normalization module is used to perform a nonlinear transformation on the comprehensive feature vector by the activation function to obtain the activated feature vector. Then, it is input into the batch normalization layer to obtain the normalized feature vector. Next, the normalized feature vector is input into the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector.

[0059] The decision prediction module is used to input the final feature vector into the decision layer for prediction and obtain the reliability prediction results of the transmission line.

[0060] The results display module is used to map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line using visual elements.

[0061] The feature enhancement module performs recalibration and projection convolution according to the following steps:

[0062] S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel;

[0063] S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights;

[0064] The expression for the first fully connected layer is as follows:

[0065] X se1 =ReLU(Dense(X) gap W se1 ));

[0066] Where: X se1 X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap W is the first global feature. se1 These are the weights of the first connection layer;

[0067] The expression for the second fully connected layer is as follows:

[0068] X se2 =σ(Dense(X) se1 W se2 ));

[0069] Where: X se2 W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer;

[0070] S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows:

[0071] X′=X⊙X se2 ;

[0072] Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation;

[0073] S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows:

[0074] Y = W·X′;

[0075] Where: Y is the output feature map after projection convolution, and W is the weight matrix.

[0076] The decision prediction module performs reliability prediction of transmission lines according to the following steps:

[0077] S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function.

[0078] The expression for the objective function is as follows:

[0079]

[0080] Where: w is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables.

[0081] The expression for the constraint condition is as follows:

[0082] y i (w·x i +b)≥1-ξ i ξ i ≥0, i=1,2,...,n;

[0083] Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term;

[0084] S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability.

[0085] The expression for the decision function is as follows:

[0086] y = f(w·x + b);

[0087] Where: y is the reliability prediction value, f is the activation function, w is the weight vector, x is the input of the decision layer, and b is the bias term;

[0088] S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

[0089] The results display module demonstrates reliability according to the following steps:

[0090] S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map;

[0091] S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

[0092] A transmission line reliability prediction device based on SENet and EffNet, the device including a processor and a memory;

[0093] The memory is used to store computer program code and to transmit the computer program code to the processor;

[0094] The processor is used to execute the above-described transmission line reliability prediction method based on SENet and EffNet according to the instructions in the computer program code.

[0095] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0096] This invention discloses a transmission line reliability prediction method based on SENet and EffNet. The method first acquires image data of the target area and constructs a neural network model of the transmission line, then performs reliability prediction using the neural network model. In application, this scheme comprehensively considers multiple data sources such as image data, electrical data, and weather data to fully reflect the actual operating status of the transmission line and the impact of the external environment. Furthermore, through structures such as convolutional layers, fusion modules, and pooling layers, it extracts and fuses multi-level features to obtain richer and more meaningful feature representations. By fusing image features, electrical data, and weather data through a fully connected layer, a comprehensive feature vector is constructed, effectively integrating features from multiple data sources to form a comprehensive feature vector for the transmission line. A comprehensive description of the power line status is achieved through nonlinear transformation, which captures complex feature relationships. Furthermore, feature normalization reduces internal covariate bias during model training, improving training efficiency and stability. Additionally, a composite scaling layer adjusts the feature scale, and a decision layer outputs a predicted reliability value for the transmission line. In summary, this invention, by introducing multi-source data, multi-level feature extraction and fusion, and nonlinear transformation and normalization, can more comprehensively and accurately assess the reliability of transmission lines. Compared to methods relying solely on historical fault data and statistical analysis, it better reflects the overall health and real-time operating status of transmission lines, thereby improving the accuracy of transmission line reliability prediction. Attached Figure Description

[0097] Figure 1 This is a flowchart of the method steps of the present invention.

[0098] Figure 2 This is an architecture diagram of the CNN model of this invention.

[0099] Figure 3 This is a schematic diagram of the system structure of the present invention.

[0100] Figure 4 This is a schematic diagram of the device structure of the present invention.

[0101] In the diagram: 1. Image data acquisition module; 2. Transmission line CNN model construction module; 3. Feature enhancement module; 4. Feature fusion module; 5. Activation normalization module; 6. Decision prediction module; 7. Result display module; 8. Processor; 9. Memory; 10. Computer program code. Detailed Implementation

[0102] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0103] Example 1:

[0104] See Figure 1A transmission line reliability prediction method based on SENet and EffNet includes:

[0105] S1. Acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices;

[0106] In this embodiment, high-resolution camera equipment or drones are used to capture image data of the power transmission lines in the target area. The acquired image data contains various details of the power transmission lines, including conductors, insulators, towers, line hardware, guy wires, grounding devices, etc.

[0107] S2. Constructing a CNN model for power transmission lines based on the Keras framework; see [link / reference] Figure 2 The CNN model includes convolutional layers (preferably at least three convolutional layers), a fusion module, pooling layers, fully connected layers, activation functions, batch normalization layers, compound scaling layers, and decision layers; the fusion module is obtained by fusing the CNN-SENet module and the EfficientNet module, wherein the CNN-SENet module includes the SE module and the EfficientNet module includes the MBConv module; the Keras framework includes a splicing layer;

[0108] S3. Input the image data into the convolutional layer to obtain the image feature map, and input the image feature map into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature map. Then, input the enhanced features into the pooling layer to obtain the global features.

[0109] Furthermore, in this scheme, image data is input into the convolutional layers of a Convolutional Neural Network (CNN). These convolutional layers extract features from the image through a series of convolutional operations. The principle of convolution is to slide a convolutional kernel (filter) across the image and calculate the dot product between the kernel and a local region of the image, thereby extracting the features of that region. Multiple convolutional kernels can extract different features, such as edges, textures, and shapes. Through multiple convolutional operations, the obtained image features gradually transform from low-level features to high-level features, such as from edges and textures to objects and scenes. These high-level features are then further processed and combined in subsequent layers to form a comprehensive understanding of the image.

[0110] Furthermore, the fusion module in this scheme is based on the fusion of the CNN-SENet and EfficientNet modules. The design purpose of the fusion module is to further enhance the extracted image features, making them more recognizable and representative. Specifically, the CNN-SENet module introduces an attention mechanism, which enhances the focus on important features by weighting the channels of the feature map. The EfficientNet module, on the other hand, improves model performance while maintaining efficient computation through a compound scaling strategy. After the image features are input into the fusion module, the enhanced features output by the fusion module include both the key features brought by the attention mechanism of the CNN-SENet module and the efficient features optimized by the EfficientNet module.

[0111] Furthermore, the main function of the pooling layer in this scheme is to reduce the dimensionality and compress the feature map while retaining important feature information. Pooling reduces the size of the feature map, thereby lowering computational complexity and enhancing the robustness of the model. In this embodiment, the pooling operation can be max pooling or average pooling. In max pooling, the maximum value within the pooling window is used as the output, highlighting salient features in the feature map; in average pooling, the average value within the pooling window is used as the output, smoothing noise in the feature map. After inputting the enhanced features into the pooling layer, the global features obtained through pooling not only retain important information from the original feature map but also reduce its size. Global features provide a holistic description of the input image, containing important information from various parts of the image; through the processing of the pooling layer, global features provide a concise and effective input for subsequent feature fusion and decision-making.

[0112] In this embodiment, the steps for fusing the SE module and the MBConv module are as follows:

[0113] S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution;

[0114] S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution.

[0115] Furthermore, specifically including:

[0116] S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel;

[0117] Furthermore, global average pooling is applied to the feature map after depthwise separable convolution to obtain the first global feature for each channel. Global average pooling is a method that compresses the spatial dimensions of a feature map into a single value by averaging all pixels in each channel to generate a global feature vector. Specifically, assuming the feature map has dimensions H×W×C, where H represents the height, W represents the width, and C represents the number of channels, global average pooling averages the H×W values ​​of each channel to obtain a vector of size 1×1×C. This compresses the spatial information of each channel into a global feature, preserving the overall information of each channel while reducing the amount of data, providing a concise input for subsequent fully connected layer processing.

[0118] S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights;

[0119] Furthermore, the first global feature is processed through a first fully connected layer, a second fully connected layer, and an activation function to generate channel attention weights. First, the first fully connected layer compresses the dimension of the global feature vector from C to a smaller dimension r, where r is a hyperparameter, typically set to a small value to reduce computation. Next, a ReLU activation function is applied, adding non-linear features and enabling the model to learn more complex features. Then, the second fully connected layer restores the dimension from r to C, and a σ(Sigmoid) activation function is used to constrain the output value between 0 and 1, generating channel attention weights. These channel attention weights reflect the importance of each channel; a larger value indicates higher channel importance.

[0120] The expression for the first fully connected layer is as follows:

[0121] X se1 =ReLU(Dense(X) gap W se1 ));

[0122] Where: X se1 X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap W is the first global feature. se1 These are the weights of the first connection layer;

[0123] The expression for the second fully connected layer is as follows:

[0124] X se2 =σ(Dense(X) se1 W se2 ));

[0125] Where: X se2W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer;

[0126] S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows:

[0127] X′=X⊙X se2 ;

[0128] Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation;

[0129] Furthermore, the generated channel attention weights are multiplied by the feature map after depthwise separable convolution to recalibrate the number of channels. Specifically, the channel attention weights are 1×1×C vectors, which are multiplied by each channel of the feature map to adjust the feature value of each channel. Assuming the feature map is H×W×C and the channel attention weights are 1×1×C, then the recalibrated feature map is: X′=X⊙X se2 Through the above methods, important channels will be amplified and unimportant channels will be suppressed, thereby realizing the recalibration of the importance of feature map channels. The recalibrated feature map retains important feature information, enhances the feature expression ability of the model, and helps to improve the recognition accuracy of the model.

[0130] S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows:

[0131] Y = W·X′;

[0132] Where: Y is the output feature map after projection convolution, and W is the weight matrix.

[0133] Furthermore, the number of channels in the recalibrated feature map is projected back to the initial number of channels. Projective convolution is implemented using a 1×1 convolution, projecting the number of channels in the recalibrated feature map from C back to the initial number of channels C0. 1×1 convolution is an efficient convolution operation that integrates and compresses information from multiple channels by performing a linear transformation at each pixel. Assuming the recalibrated feature map is H×W×C and the initial number of channels is C0, then the output of the projective convolution is H×W×C0, and the formula for calculating the 1×1 convolution is Y=W·X′. Through this operation, the number of channels in the feature map is compressed to the initial number of channels while retaining important recalibrated feature information. The effect of projective convolution is to reduce the number of channels in the feature map, lower computational complexity, and maintain the size and important features of the feature map, ensuring the computational efficiency and feature representation ability of subsequent network layers.

[0134] In this embodiment, global average pooling is used to extract global features for each channel. Fully connected layers and activation functions are used to generate channel attention weights, and the channels of the feature map are recalibrated and optimized. Finally, the number of channels in the feature map is adjusted by projection convolution. This not only effectively enhances the expressive power of the feature map, enabling the model to better focus on important features, but also optimizes the structure and information content of the feature map while maintaining computational efficiency, providing more valuable input for subsequent network layers.

[0135] S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module.

[0136] In this embodiment, the feature map is obtained through the first few convolutional layers and contains various feature information of the image. The initial number of channels in the feature map represents its depth, i.e., how many independent feature channels the feature map contains. The input of the feature map is the foundation of the entire fusion process. Subsequent operations will be processed and optimized based on these input feature maps. The process of inputting the feature map can be understood as using the output of the previous layer as the input of the current layer. These feature maps contain various feature information extracted by the previous few layers, such as edges and textures.

[0137] The MBConv module expands the initial number of channels in the feature map and performs depthwise separable convolution on it. MBConv is a core component of EfficientNet; it increases the expressive power of the feature map by expanding its channel count. Depthwise separable convolution is an efficient convolution operation that decomposes standard convolution into depthwise convolution and pointwise convolution, significantly reducing computation. Depthwise convolution performs convolution operations independently on each channel, while pointwise convolution integrates the features from each channel through 1x1 convolution. After depthwise separable convolution, the size and number of channels of the feature map are optimized, extracting more useful features. Specifically, depthwise convolution processes each channel independently, preserving the spatial features of each channel, while pointwise convolution fuses the information from each channel through 1x1 convolution operations, further extracting higher-level features. The advantage of depthwise separable convolution is that it can extract richer feature information while maintaining computational efficiency.

[0138] After depthwise separable convolution, the SE module is embedded into the MBConv module, and the number of channels of the feature map after depthwise separable convolution is recalibrated and projective convolution is performed. The SE module recalibrates the number of channels of the feature map by calculating the importance weight of each channel, that is, adjusting the weight of each channel so that important feature channels are enhanced and unimportant feature channels are suppressed. Projective convolution is used to project the number of channels of the recalibrated feature map back to the initial number of channels, thereby maintaining the consistency of the size and number of channels of the feature map. Projective convolution is implemented by 1x1 convolution, which compresses and integrates the recalibrated feature map so that the number of channels of the feature map is consistent with the initial number of channels. Through the recalibration of the SE module, important features are enhanced, while the number of channels of the feature map is kept unchanged through projective convolution.

[0139] The feature map after projection convolution is subjected to pointwise convolution. With the number of channels in the pointwise convolutional feature map being the same as the initial number of channels and a stride of 1, residual connections are applied to the pointwise convolutional feature map to fuse the MBConv and SE modules. Pointwise convolution integrates the channels of the feature map through 1x1 convolution, ensuring the number of channels in the feature map matches the initial number. Residual connections are an effective feature fusion method. By adding the input feature map to the processed feature map, information from the input feature map is preserved while enhancing the expressive power of the processed feature map. Through residual connections, the features of the MBConv and SE modules are effectively fused. The output feature map contains both the rich features extracted by depthwise separable convolution and the important features recalibrated by the SE module, thereby improving the expressive power and discriminative power of the feature map. The advantage of residual connections is that they can alleviate the gradient vanishing problem, promote network training and convergence, and ensure effective transfer and fusion of feature maps between different modules.

[0140] The fusion of the MBConv and SE modules in this implementation results in an output feature map that contains both the rich features extracted by depthwise separable convolution and the important features recalibrated by the SE module, thereby effectively improving the expressive power and discriminative power of the feature map.

[0141] S4. Real-time acquisition of electrical parameters, environmental parameters of transmission lines and weather data of target area, and data splicing of electrical parameters and weather data through splicing layer, and then inputting spliced ​​data and global features into fully connected layer for fusion to obtain comprehensive feature vector;

[0142] Furthermore, electrical and environmental parameters can be acquired in real time by sensors installed on the transmission lines. Electrical parameters include current and voltage, while environmental parameters include temperature and humidity. Weather data can be obtained from meteorological stations or online weather services, including temperature, humidity, wind speed, and rainfall.

[0143] Sensor data and weather data are important external factors affecting the reliability of transmission lines. In this scheme, the above data are input into the fully connected layer along with global features. Through the linear transformation of the fully connected layer, the data from different sources are fused to obtain a comprehensive feature vector. The comprehensive feature vector contains integrated information from image features, sensor data, and weather data, which can comprehensively reflect the status of the transmission line and the influence of the external environment. Through feature fusion, the predictive ability of transmission line reliability is enhanced.

[0144] Furthermore, the CNN model in this scheme uses Keras for deep learning model construction and training. Keras includes a concatenation layer that inputs sensor data, weather data, and global features into a fully connected layer to obtain a comprehensive feature vector. During operation, sensor data and weather data are concatenated through the concatenation layer to integrate data from different sources into a single input. Specifically, the concatenation layer in Keras is used to concatenate multiple tensors along a specified axis. Assuming the shape of the sensor data is (N, D1) and the shape of the weather data is (N, D2), where N is the number of samples, and D1 and D2 are the feature dimensions of the sensor and weather data, respectively, the concatenation layer concatenates these two data points along their feature dimensions to obtain concatenated data of shape (N, D1 + D2). This can be implemented using Keras's concatenation layer. In this way, the concatenated data contains all the features of both sensor and weather data, providing a complete input for subsequent feature fusion. This method integrates data from different sources into a whole, preserving their respective feature information, facilitating subsequent processing and analysis.

[0145] In this scheme, the concatenated data is fused with global features through a fully connected layer to obtain a comprehensive feature vector. A fully connected layer (Dense layer) is a common neural network layer used to map input data to a new feature space.

[0146] In this embodiment, the shape of the global feature is (N, G), where G is the dimension of the global feature. During operation, the concatenated data is first processed through one or more fully connected layers to extract high-level features. The output shape of the fully connected layer is (N, H), where H is the output dimension of the fully connected layer. Then, the processed data is concatenated with the global feature and further processed through one or more fully connected layers to obtain the final comprehensive feature vector. This can be implemented using Keras's Dense and Concatenate layers. In this way, the comprehensive feature vector contains fused information from sensor data, weather data, and global features.

[0147] In this embodiment, features from different sources are fused to generate a comprehensive feature vector containing all important information, which facilitates subsequent model training and prediction. Through this feature fusion, the model can better capture the relationships between data, improving the accuracy and robustness of predictions.

[0148] S5. The comprehensive feature vector is transformed nonlinearly by the activation function to obtain the activated feature vector. Then, it is input into the batch normalization layer to obtain the normalized feature vector. Next, the normalized feature vector is input into the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector.

[0149] Furthermore, in this scheme, the comprehensive feature vector is nonlinearly transformed through an activation function to obtain an activated feature vector. The role of the activation function is to introduce nonlinear factors, enabling the neural network to fit complex nonlinear relationships. In this embodiment, the activation function is preferably the Leaky ReLU function, which has the advantages of simple calculation and fast convergence. When the input of the Leaky ReLU function is non-negative, the output of the Leaky ReLU function is the same as the input; when the input of the Leaky ReLU function is negative, the output of the Leaky ReLU function is positive. By inputting the comprehensive feature vector into the activation function and performing a nonlinear transformation, the linear relationship in the feature vector is broken, thereby enabling a better fit to complex feature relationships. The activated feature vector contains feature information after nonlinear transformation, enhancing the model's expressive and fitting capabilities. By introducing nonlinear factors, the model can capture complex relationships between features, improving the prediction accuracy of transmission line reliability.

[0150] Furthermore, the main function of the batch normalization layer is to standardize the feature vectors, making their distribution more stable, thereby accelerating the model training process and improving model stability. Specifically, the batch normalization layer normalizes the feature vectors by calculating their mean and standard deviation, ensuring the mean is zero and the standard deviation is one. The normalized feature vectors have a more stable distribution, avoiding the adverse effects of excessively large or small feature values ​​on model training. Inputting the activated feature vectors into the batch normalization layer, through normalization, results in a more stable distribution, which is beneficial for subsequent feature processing and decision-making, making the feature vector distribution more stable and improving the model's training efficiency and stability.

[0151] Furthermore, the normalized feature vectors are input into a composite scaling layer to obtain the final feature vectors. The role of the composite scaling layer is to further adjust the feature vectors, making their scale more reasonable, thereby improving the model's predictive ability. The composite scaling layer adjusts the scale of the feature vectors by performing a linear transformation, ensuring that each dimension of the feature vector has similar importance at the same scale. Specifically, the composite scaling layer uses a learned scaling factor to scale and adjust each dimension of the feature vector, ensuring that each dimension has similar importance at the same scale. Inputting the normalized feature vectors into the composite scaling layer, through scaling adjustment, results in a final feature vector with a more reasonable scale, which is helpful for subsequent decision-making. By scaling adjustment, the various dimensions of the feature vector have similar importance at the same scale, thus improving the model's predictive ability.

[0152] S6. Input the final feature vector into the decision layer for prediction to obtain the transmission line reliability prediction result;

[0153] Furthermore, in this scheme, the final feature vector is input into the decision layer to obtain the transmission line reliability prediction value. The role of the decision layer is to process the final feature vector to obtain the transmission line reliability prediction value. The decision layer uses a classifier to obtain the prediction result by performing linear and nonlinear transformations on the final feature vector. Specifically, the decision layer calculates the score of each category by performing a linear transformation on the final feature vector, and then performs linear and nonlinear transformations through an activation function to obtain the final prediction value. The prediction value reflects the probability that the transmission line can operate normally and without failure within a certain period of time.

[0154] Furthermore, specifically including:

[0155] S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function.

[0156] The expression for the objective function is as follows:

[0157]

[0158] Where: w is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables.

[0159] The expression for the constraint condition is as follows:

[0160] y i (w·x i +b)≥1-ξi ξ i ≥0, i=1,2,...,n;

[0161] Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term;

[0162] Furthermore, in this scheme, the final feature vector is input into the decision layer to minimize the objective function within the decision layer. The decision layer is typically the last layer of the neural network, responsible for generating the final prediction result. The objective function (loss function) is the function that needs to be optimized during model training. After the final feature vector is input into the decision layer, training is performed using the backpropagation algorithm. The purpose is to adjust the model's parameters (including weight vectors and bias terms) to minimize the value of the objective function. The backpropagation algorithm updates the parameters step by step by calculating the gradient of the objective function with respect to each parameter, causing the objective function to gradually converge to its minimum value. In this way, the model's weight vector and bias terms are continuously adjusted during training to minimize the value of the objective function. By optimizing the model's parameters, the model can more accurately fit the training data, improving the accuracy of predictions.

[0163] S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability.

[0164] The expression for the decision function is as follows:

[0165] y = f(w·x + b);

[0166] Where: y is the reliability prediction value, f is the activation function, W is the weight vector, x is the input of the decision layer, and b is the bias term;

[0167] Furthermore, in this scheme, the decision function is the function by which the model generates the prediction results, typically a linear or nonlinear transformation. Assuming the input to the decision layer is x, the weight vector is W, and the bias term is b, the decision function can be expressed as y = f(w·x + b). By calculating the decision function, the model's predicted value can be obtained. In the transmission line reliability prediction task, the predicted value represents the reliability score of the transmission line. In this way, the model's output is the predicted reliability value of the transmission line.

[0168] S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

[0169] Furthermore, the final output of the decision function, the model's predicted value, is a probability distribution calculated using the Softmax function. Through the Softmax function, the model's output is transformed into a probability distribution, with each category corresponding to a probability value. For the transmission line reliability prediction task, these probability values ​​can be represented as probabilities of different reliability levels. For example, assuming the model outputs three categories: high reliability, medium reliability, and low reliability, the model's output might be [0.7, 0.2, 0.1], indicating that the transmission line has a 70% probability of being highly reliable, a 20% probability of being moderately reliable, and a 10% probability of being low reliable.

[0170] This implementation generates the final prediction results, providing a quantitative assessment of the reliability of transmission lines. The reliability predictions can be used for preventative maintenance and decision support, helping to improve the reliability and safety of transmission lines.

[0171] S7. Map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line using visual elements.

[0172] Furthermore, specifically including:

[0173] S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map;

[0174] S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

[0175] Furthermore, in this scheme, the output results of the model are statistically analyzed and organized to form a dataset containing the reliability prediction values ​​of all transmission lines in the target area. Then, through the GIS (Geographic Information System) interface, these reliability prediction values ​​are mapped to the corresponding locations on the map.

[0176] A GIS interface is a tool for processing and displaying geospatial data. It can combine geographic data with attribute data for visualization. Each transmission line corresponds to a geographic path on the map, and the reliability prediction values ​​of these paths are accurately mapped to the corresponding locations on the map. Complex prediction data can be geospatially visualized, allowing users to intuitively view and analyze the reliability of transmission lines in different regions.

[0177] To make the reliability information of power transmission lines on the map more intuitive and easier to understand, different visual elements such as colors, line thickness, and transparency can be used to represent different reliability levels. For example, high-reliability power transmission lines can be represented in green, medium-reliability lines in yellow, and low-reliability lines in red. Furthermore, the different reliability levels can be further distinguished by adjusting the line thickness and transparency.

[0178] In this embodiment, different visual elements are used to represent the reliability of the corresponding transmission lines, enabling users to easily identify which transmission lines require key attention and maintenance. By differentiating the visual elements, the reliability information of the transmission lines becomes more intuitive and easier to understand, helping users to make quick decisions and take appropriate measures.

[0179] Example 2:

[0180] See Figure 3 A transmission line reliability prediction method system based on SENet and EffNet, the system comprising:

[0181] Image data acquisition module 1 is used to acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices;

[0182] Module 2 for building a CNN model for power transmission lines is used to build a CNN model for power transmission lines based on the Keras framework. The CNN model includes convolutional layers, a fusion module, pooling layers, fully connected layers, activation functions, batch normalization layers, composite scaling layers, and decision layers. The fusion module is obtained by fusing the CNN-SENet module and the EfficientNet module. The CNN-SENet module includes the SE module, and the EfficientNet module includes the MBConv module. The Keras framework includes a stitching layer.

[0183] Feature enhancement module 3 is used to input image data into the convolutional layer to obtain image feature maps, and input the image feature maps into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature maps. Then, the enhanced features are input into the pooling layer to obtain global features.

[0184] The steps for fusing the SE module and the MBConv module are as follows:

[0185] S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution;

[0186] S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution.

[0187] Furthermore, the feature enhancement module 3 performs recalibration and projection convolution according to the following steps:

[0188] S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel;

[0189] S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights;

[0190] The expression for the first fully connected layer is as follows:

[0191] X se1 =ReLU(Dense(X) gap W se1 ));

[0192] Where: X se1 X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap As the first global feature, X se1 These are the weights of the first connection layer;

[0193] The expression for the second fully connected layer is as follows:

[0194] X se2 =σ(Dense(X) se1 W se2 ));

[0195] Where: X se2 W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer;

[0196] S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows:

[0197] X′=X⊙X se2 ;

[0198] Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation;

[0199] S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows:

[0200] Y = W·X′;

[0201] Where: Y is the output feature map after projection convolution, and W is the weight matrix.

[0202] S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module.

[0203] Feature fusion module 4 is used to acquire electrical parameters, environmental parameters and weather data of the transmission line and the target area in real time. The electrical parameters and weather data are spliced ​​together through the splicing layer. Then, the spliced ​​data and global features are input into the fully connected layer for fusion to obtain a comprehensive feature vector.

[0204] The activation normalization module 5 is used to perform a nonlinear transformation on the comprehensive feature vector by the activation function to obtain the activated feature vector, which is then input to the batch normalization layer to obtain the normalized feature vector. The normalized feature vector is then input to the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector.

[0205] Decision prediction module 6 is used to input the final feature vector into the decision layer for prediction and obtain the reliability prediction result of the transmission line.

[0206] Furthermore, the decision prediction module 6 performs reliability prediction of the transmission line according to the following steps:

[0207] S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function.

[0208] The expression for the objective function is as follows:

[0209]

[0210] Where: w is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables.

[0211] The expression for the constraint condition is as follows:

[0212] y i (w·x i +b)≥1-ξ i ξ i ≥0, i=1,2,...,n;

[0213] Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term;

[0214] S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability.

[0215] The expression for the decision function is as follows:

[0216] y = f(w·x + b);

[0217] Where: y is the reliability prediction value, f is the activation function, w is the weight vector, x is the input of the decision layer, and b is the bias term;

[0218] S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

[0219] The results display module 7 is used to map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line using visual elements.

[0220] Furthermore, the result display module 7 displays the reliability according to the following steps:

[0221] S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map;

[0222] S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

[0223] Example 3:

[0224] See Figure 4 A transmission line reliability prediction device based on SENet and EffNet, the device including a processor 8 and a memory 9;

[0225] The memory 9 is used to store computer program code 91 and to transmit the computer program code 91 to the processor 8;

[0226] The processor 8 is used to perform the transmission line reliability prediction method based on SENet and EffNet as described in Embodiment 1, according to the instructions in the computer program code 91.

[0227] Generally, the computer instructions for implementing the method of the present invention can be carried on any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media can include any computer-readable medium except for the signal itself, which is temporarily propagating.

[0228] Computer-readable storage media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EKROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0229] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smarttalk, and C++—as well as conventional procedural programming languages—such as the "C" language or similar programming languages. In particular, Python, suitable for neural network computation, and platform frameworks based on TensorFlow, PyTorch, etc., can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer or to an external computer (e.g., via the Internet using an Internet service provider) through any type of network, including a local area network (LAN) or a wide area network (WAN).

[0230] For details regarding the aforementioned equipment and non-transitory computer-readable storage media, please refer to the specific description of a transmission line reliability prediction method based on SENet and EffNet and its beneficial effects, which will not be repeated here.

[0231] Although embodiments of the present invention have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for predicting the reliability of transmission lines based on SENet and EffNet, characterized in that, include: S1. Acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices; S2. Construct a CNN model for power transmission lines based on the Keras framework; the CNN model includes convolutional layers, fusion modules, pooling layers, fully connected layers, activation functions, batch normalization layers, composite scaling layers, and decision layers; the fusion module is obtained by fusing the CNN-SENet module and the EfficientNet module, the CNN-SENet module includes the SE module, and the EfficientNet module includes the MBConv module; the Keras framework includes a stitching layer; S3. Input the image data into the convolutional layer to obtain the image feature map, and input the image feature map into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature map. Then, input the enhanced features into the pooling layer to obtain the global features. The steps for fusing the SE module and the MBConv module are as follows: S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution; S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution. S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module. S4. Real-time acquisition of electrical parameters, environmental parameters of transmission lines and weather data of target area, and data splicing of electrical parameters and weather data through splicing layer, and then inputting spliced ​​data and global features into fully connected layer for fusion to obtain comprehensive feature vector; S5. The comprehensive feature vector is transformed nonlinearly by the activation function to obtain the activated feature vector. Then, it is input into the batch normalization layer to obtain the normalized feature vector. Next, the normalized feature vector is input into the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector. S6. Input the final feature vector into the decision layer for prediction to obtain the transmission line reliability prediction result; S7. Map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line using visual elements.

2. The transmission line reliability prediction method based on SENet and EffNet according to claim 1, characterized in that: Step S32 specifically includes: S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel; S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights; The expression for the first fully connected layer is as follows: X se1 =ReLU(Dense(X gap ,W se1 )): Where: X se1 X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap W is the first global feature. se1 These are the weights of the first connection layer; The expression for the second fully connected layer is as follows: X se2 =σ(Dense(X se1 ,W se2 )); Where: X se2 W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer; S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows: X′=X⊙X se2 ; Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation; S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows: Y = W·X′; Where: Y is the output feature map after projection convolution, and W is the weight matrix.

3. The transmission line reliability prediction method based on SENet and EffNet according to claim 1, characterized in that: Step S6 specifically includes: S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function. The expression for the objective function is as follows: Where: W is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables. The expression for the constraint condition is as follows: y i (w x i +b)≥1-ξ i ,x i ≥0, i=1, 2,..., n; Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term; S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability. The expression for the decision function is as follows: y = f(w·x + b); Where: y is the reliability prediction value, f is the activation function, w is the weight vector, x is the input of the decision layer, and b is the bias term; S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

4. The transmission line reliability prediction method based on SENet and EffNet according to claim 3, characterized in that: Step S7 specifically includes: S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map; S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

5. The transmission line reliability prediction method based on SENet and EffNet according to claim 1, characterized in that: The activation function is the Leaky ReLU function.

6. A transmission line reliability prediction system based on SENet and EffNet, characterized in that, The system includes: Image data acquisition module (1) is used to acquire image data of the target area of ​​the transmission line; the image data includes conductors, insulators, towers, line hardware, guy wires, and grounding devices; A CNN model construction module (2) for power transmission lines is used to construct a CNN model for power transmission lines based on the Keras framework. The CNN model includes convolutional layers, fusion modules, pooling layers, fully connected layers, activation functions, batch normalization layers, composite scaling layers, and decision layers. The fusion module is obtained by fusing the CNN-SENet module and the EfficientNet module. The CNN-SENet module includes the SE module, and the EfficientNet module includes the MBConv module. The Keras framework includes a splicing layer. The feature enhancement module (3) is used to input image data into the convolutional layer to obtain image feature maps, and input the image feature maps into the fusion module to fuse the SE module and the MBConv module, and output the enhanced features of the image feature maps. Then, the enhanced features are input into the pooling layer to obtain global features. The steps for fusing the SE module and the MBConv module are as follows: S31. Input image feature map, expand the initial number of channels of image feature map through MBConv module, and perform depthwise separable convolution to obtain feature map after depthwise separable convolution; S32. Embed the SE module into the MBConv module, and recalibrate and project convolution the number of channels of the feature map after depthwise separable convolution. S33. Perform pointwise convolution on the output feature map after projection convolution. Then, with the number of channels in the pointwise convolution feature map being the same as the initial number of channels and the stride being 1, perform residual connection on the pointwise convolution feature map to achieve the fusion of the SE module and the MBConv module. The feature fusion module (4) is used to acquire the electrical parameters, environmental parameters and weather data of the transmission line and the target area in real time. The electrical parameters and weather data are spliced ​​together through the splicing layer. Then, the spliced ​​data and global features are input into the fully connected layer for fusion to obtain a comprehensive feature vector. The activation normalization module (5) is used to perform nonlinear transformation of the comprehensive feature vector by the activation function to obtain the activated feature vector, and then input it into the batch normalization layer to obtain the normalized feature vector. Then, the normalized feature vector is input into the composite scaling layer to adjust the scale of the feature vector to obtain the final feature vector. The decision prediction module (6) is used to input the final feature vector into the decision layer for prediction and obtain the reliability prediction result of the transmission line. The results display module (7) is used to map the reliability prediction results to the corresponding locations on the map and display the reliability of the transmission line in terms of visual elements.

7. The transmission line reliability prediction system based on SENet and EffNet according to claim 6, characterized in that: The feature enhancement module (3) performs recalibration and projection convolution according to the following steps: S321. Perform global average pooling on the feature map after depthwise separable convolution to obtain the first global feature of each channel; S322. The first global feature is processed through the first fully connected layer, the second fully connected layer and the activation function to generate channel attention weights; The expression for the first fully connected layer is as follows: X se1 =ReLU(Dense(X gap ,W se1 )): Where: X sel X represents the features processed by the first fully connected layer (Dense) and the ReLU activation function. gap W is the first global feature. sel These are the weights of the first connection layer; The expression for the second fully connected layer is as follows: X se2 =σ(Dense(X se1 ,W se2 )); Where: X se2 W represents the channel attention weights processed by the second fully connected layer Dense and the σ activation function. se2 These are the weights of the second connection layer; S323. Multiply the generated channel attention weights by the feature map after depthwise separable convolution to recalibrate the number of channels, obtaining the recalibrated feature map, the expression of which is as follows: X′=X⊙X se2 ; Where: X′ is the recalibrated feature map, X is the feature map after depthwise separable convolution, and ⊙ represents the element-wise multiplication operation; S324. Project the number of channels of the recalibrated feature map onto the initial number of channels to complete the projective convolution, the expression of which is as follows: Y = W·X′; Where: Y is the output feature map after projection convolution, and W is the weight matrix.

8. The transmission line reliability prediction system based on SENet and EffNet according to claim 7, characterized in that: The decision prediction module (6) performs reliability prediction of the transmission line according to the following steps: S61. Input the final feature vector into the decision layer to minimize the objective function in the decision layer. Under the condition that the objective function is minimized, determine the weight vector and bias term of the CNN model according to the constraints of the objective function. The expression for the objective function is as follows: Where: W is the weight vector, ||W|| 2 Let ξ be the coefficient of the weight vector, C be the penalty parameter, and ξ be the coefficient of the weight vector. i Let n be the number of samples, and n be the number of slack variables. The expression for the constraint condition is as follows: y i (w x i +b)≥1-ξ i ,x i ≥0, i=1, 2,..., n; Where: w is the weight vector, y i Let x be the label of sample i. i Let b be the i-th final feature vector, and b be the bias term; S62. Calculate the decision function in the decision layer based on the weight vector and bias term to obtain the predicted value of the transmission line reliability. The expression for the decision function is as follows: y = f(w·x + b); Where: y is the reliability prediction value, f is the activation function, w is the weight vector, x is the input of the decision layer, and b is the bias term; S63. The reliability prediction values ​​are statistically distributed using the Softmax function to represent the probabilities of different reliability levels.

9. A transmission line reliability prediction system based on SENet and EffNet according to claim 8, characterized in that: The result display module (7) displays the reliability according to the following steps: S71. Calculate the reliability prediction values ​​for all target areas, and map the reliability prediction values ​​of the transmission lines in all target areas to the corresponding locations on the map through the GIS interface; each transmission line in the map corresponds to a geographical path on the map; S72. Based on different reliability prediction values, the transmission lines corresponding to different reliability prediction values ​​are displayed using different visual elements.

10. A transmission line reliability prediction device based on SENet and EffNet, characterized in that: The device includes a processor (8) and a memory (9); The memory (9) is used to store computer program code (91) and transmit the computer program code (91) to the processor (8); The processor (8) is used to execute the transmission line reliability prediction method based on SENet and EffNet according to any one of the instructions in the computer program code (91).

Images