Power transmission line defect identification method and device, computer equipment, storage medium and computer program product

Abstract

The invention relates to a power transmission line defect identification method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: preprocessing a visible light image of a to-be-analyzed power transmission line to obtain a preprocessed visible light image; inputting the pre-processed visible light image into a trained defect identification model to obtain a first confidence degree of the power transmission line under each preset defect type; inputting the X-ray image of the power transmission line into the trained defect identification model to obtain a second confidence degree of the power transmission line under each preset defect type; according to the first confidence coefficient and the second confidence coefficient, determining a target confidence coefficient of the power transmission line under each preset defect type; and screening the defect type with the target confidence greater than the preset confidence from the preset defect types as a target defect type of the power transmission line, and obtaining a defect identification result of the power transmission line based on the target defect type. By adopting the method, the defect identification accuracy of the power transmission line can be improved.

Description

Technical Field

[0001] This application relates to the field of power grid technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for identifying defects in transmission lines. Background Technology

[0002] Currently, in order to ensure the safe and stable operation of the power grid, accurate identification of transmission line defects is of paramount importance.

[0003] In traditional technology, manual experience is usually used to identify defects in transmission lines. However, this method is subjective and prone to errors, resulting in low accuracy in identifying defects in transmission lines. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for identifying transmission line defects that can improve the accuracy of defect identification in transmission lines, in order to address the aforementioned technical problems.

[0005] In a first aspect, this application provides a method for identifying defects in transmission lines, including:

[0006] A visible light image of the transmission line to be analyzed is acquired, and the visible light image is preprocessed to obtain a preprocessed visible light image.

[0007] Extract the image features corresponding to the preprocessed visible light image, and determine the similarity between the image features and preset image features;

[0008] If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0009] Obtain an X-ray image of the transmission line, input the X-ray image into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type;

[0010] Based on the first confidence level and the second confidence level, the target confidence level of the transmission line under each preset defect type is determined;

[0011] From the preset defect types, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained.

[0012] In one embodiment, determining the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level includes:

[0013] Obtain the first weight corresponding to the first confidence level and the second weight corresponding to the second confidence level, and determine the confidence level difference between the first confidence level and the second confidence level;

[0014] When the confidence difference is less than or equal to the confidence difference threshold, the first confidence and the second confidence are weighted and summed according to the first weight and the second weight to obtain the target confidence of the transmission line under each preset defect type.

[0015] If the confidence difference is greater than the confidence difference threshold, the first weight and the second weight are corrected to obtain the first target weight corresponding to the first confidence and the second target weight corresponding to the second confidence. The first confidence and the second confidence are then weighted and summed according to the first target weight and the second target weight to obtain the target confidence of the transmission line under each preset defect type.

[0016] In one embodiment, the step of correcting the first weight and the second weight to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level includes:

[0017] The visible light image and the X-ray image are respectively input into a cross-modal attention model to obtain the first feature contribution value corresponding to the visible light image and the second feature contribution value corresponding to the X-ray image;

[0018] Based on the first feature contribution and the second feature contribution respectively, query the correspondence between feature contribution and correction coefficient to obtain the first correction coefficient corresponding to the first feature contribution and the second correction coefficient corresponding to the second feature contribution;

[0019] The first weight and the second weight are corrected according to the first correction coefficient and the second correction coefficient to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level.

[0020] In one embodiment, extracting the image features corresponding to the preprocessed visible light image includes:

[0021] The preprocessed visible light image is input into the texture feature extraction model to obtain the texture features corresponding to the preprocessed visible light image.

[0022] The preprocessed visible light image is input into the shape feature extraction model to obtain the shape features corresponding to the preprocessed visible light image;

[0023] The preprocessed visible light image is input into the color feature extraction model to obtain the color features corresponding to the preprocessed visible light image.

[0024] The preprocessed visible light image is input into the component structure feature extraction model to obtain the component structure features corresponding to the preprocessed visible light image; the component structure features are used to represent the structural features of the transmission line components of the transmission line;

[0025] Based on the texture features, shape features, color features, and component structure features, the image features corresponding to the preprocessed visible light image are obtained.

[0026] In one embodiment, the preset image features include preset texture features, preset shape features, preset color features, and preset component structure features;

[0027] Determining the similarity between the image features and preset image features includes:

[0028] Determine the first similarity between the texture feature and the preset texture feature;

[0029] Determine a second similarity between the shape feature and the preset shape feature;

[0030] Determine the third similarity between the color feature and the preset color feature;

[0031] A fourth similarity is determined between the structural features of the component and the preset structural features of the component;

[0032] The first similarity, the second similarity, the third similarity, and the fourth similarity are fused to obtain the similarity between the image features and the preset image features.

[0033] In one embodiment, before inputting the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type, the method further includes:

[0034] The X-ray image is subjected to multi-scale decomposition processing to obtain the high-frequency image component and the low-frequency image component in the X-ray image;

[0035] The high-frequency image component is denoised to obtain the denoised high-frequency image component, and the low-frequency image component is smoothed to obtain the smoothed low-frequency image component.

[0036] The denoised high-frequency image components and the smoothed low-frequency image components are reconstructed to obtain the processed X-ray image.

[0037] The step of inputting the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type includes:

[0038] The processed X-ray image is input into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type.

[0039] Secondly, this application also provides a transmission line defect identification device, comprising:

[0040] The image acquisition module is used to acquire a visible light image of the transmission line to be analyzed, and to preprocess the visible light image to obtain a preprocessed visible light image.

[0041] The similarity determination module is used to extract the image features corresponding to the preprocessed visible light image and determine the similarity between the image features and preset image features;

[0042] The first identification module is used to input the preprocessed visible light image into the trained defect identification model when the similarity is less than or equal to the similarity threshold, so as to obtain the first confidence level of the transmission line under each preset defect type.

[0043] The second identification module is used to acquire X-ray images of the transmission line, input the X-ray images into the trained defect identification model, and obtain the second confidence level of the transmission line under each preset defect type.

[0044] A confidence level determination module is used to determine the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level;

[0045] The result determination module is used to select the defect type with a target confidence level greater than a preset confidence level from the preset defect types, and use it as the target defect type of the transmission line. Based on the target defect type, the defect identification result of the transmission line is obtained.

[0046] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0047] A visible light image of the transmission line to be analyzed is acquired, and the visible light image is preprocessed to obtain a preprocessed visible light image.

[0048] Extract the image features corresponding to the preprocessed visible light image, and determine the similarity between the image features and preset image features;

[0049] If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0050] Obtain an X-ray image of the transmission line, input the X-ray image into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type;

[0051] Based on the first confidence level and the second confidence level, the target confidence level of the transmission line under each preset defect type is determined;

[0052] From the preset defect types, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained.

[0053] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0054] A visible light image of the transmission line to be analyzed is acquired, and the visible light image is preprocessed to obtain a preprocessed visible light image.

[0055] Extract the image features corresponding to the preprocessed visible light image, and determine the similarity between the image features and preset image features;

[0056] If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0057] Obtain an X-ray image of the transmission line, input the X-ray image into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type;

[0058] Based on the first confidence level and the second confidence level, the target confidence level of the transmission line under each preset defect type is determined;

[0059] From the preset defect types, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained.

[0060] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0061] A visible light image of the transmission line to be analyzed is acquired, and the visible light image is preprocessed to obtain a preprocessed visible light image.

[0062] Extract the image features corresponding to the preprocessed visible light image, and determine the similarity between the image features and preset image features;

[0063] If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0064] Obtain an X-ray image of the transmission line, input the X-ray image into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type;

[0065] Based on the first confidence level and the second confidence level, the target confidence level of the transmission line under each preset defect type is determined;

[0066] From the preset defect types, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained.

[0067] The aforementioned method, apparatus, computer equipment, storage medium, and computer program product for identifying transmission line defects first acquire a visible light image of the transmission line to be analyzed, preprocesses the visible light image to obtain a preprocessed visible light image, extracts the image features corresponding to the preprocessed visible light image, determines the similarity between the image features and preset image features, and inputs the preprocessed visible light image into a trained defect identification model when the similarity is less than or equal to a similarity threshold to obtain a first confidence level of the transmission line under each preset defect type. Next, an X-ray image of the transmission line is acquired and input into the trained defect identification model to obtain a second confidence level of the transmission line under each preset defect type. Then, based on the first and second confidence levels, a target confidence level of the transmission line under each preset defect type is determined. Finally, from each preset defect type, defect types with a target confidence level greater than a preset confidence level are selected as target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained. In this way, when identifying defects in transmission lines, visible light and X-ray images of the transmission line are acquired simultaneously and input into the trained defect identification model to obtain the first and second confidence scores for the corresponding defect types. The complementary characteristics of the two images are utilized to avoid the limitations of single-image detection. Furthermore, before inputting the visible light image into the model, its image features are extracted and compared with preset features for similarity verification. Only when the similarity is below a threshold is the defect identification process initiated, effectively filtering invalid samples such as background interference and image blurring, reducing model misjudgments. The first and second confidence scores are then weighted and fused to obtain the target confidence score. Defect types with target confidence scores above the threshold are selected as the final result, further improving the reliability and accuracy of the identification results and enhancing the accuracy of transmission line defect identification. Moreover, this process requires no human intervention, avoiding the subjective factors and errors inherent in manual experience-based identification methods, which can lead to lower accuracy in transmission line defect identification, thus improving the overall accuracy of transmission line defect identification. Attached Figure Description

[0068] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0069] Figure 1 This is a flowchart illustrating a method for identifying defects in transmission lines in one embodiment;

[0070] Figure 2This is a flowchart illustrating the steps for determining the target confidence level of a transmission line under various preset defect types in one embodiment.

[0071] Figure 3 This is a flowchart illustrating a transmission line defect identification method in another embodiment;

[0072] Figure 4 This is a structural block diagram of a transmission line defect identification device in one embodiment;

[0073] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0074] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0075] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0076] In one exemplary embodiment, such as Figure 1 As shown, a method for identifying defects in transmission lines is provided. This embodiment illustrates the application of this method to a server. It is understood that this method can also be applied to a terminal, or to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, and tablets; the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. In this embodiment, the method includes the following steps:

[0077] Step S101: Obtain the visible light image of the transmission line to be analyzed, and preprocess the visible light image to obtain the preprocessed visible light image.

[0078] Transmission lines refer to power facility systems consisting of components such as conductors, insulators, fittings, towers, and grounding devices, used to transmit electrical energy.

[0079] Visible light images refer to two-dimensional images of the power transmission lines to be analyzed, obtained by using visible light imaging equipment (such as industrial cameras or UAV-borne visible light cameras) under natural light or conventional lighting conditions.

[0080] Among them, the preprocessed visible light image refers to the visible light image after preprocessing.

[0081] For example, the server uses a visible light imaging device deployed along the power transmission line to capture images of the power transmission line under natural light or conventional lighting conditions, obtaining a visible light image of the power transmission line to be analyzed. Then, a Gaussian filtering algorithm or a median filtering algorithm is used to eliminate Gaussian noise and salt-and-pepper noise in the visible light image, resulting in a denoised visible light image. Next, an adaptive contrast enhancement algorithm is used to improve the contrast between the power transmission line components and the background environment in the denoised visible light image, resulting in an enhanced visible light image. Finally, the enhanced visible light image is subjected to size normalization processing to obtain a preprocessed visible light image.

[0082] Step S102: Extract the image features corresponding to the preprocessed visible light image and determine the similarity between the image features and the preset image features.

[0083] Image features refer to the feature information extracted from the preprocessed visible light image.

[0084] Among them, the preset image features refer to the image features extracted from the visible light images corresponding to the preset transmission lines.

[0085] Here, similarity refers to the degree of similarity between the image features to be analyzed and the preset image features.

[0086] For example, the server inputs the preprocessed visible light image into the feature extraction model for feature extraction processing to obtain the image features corresponding to the preprocessed visible light image; then, the image features and preset image features are input into the similarity prediction model to obtain the similarity between the image features and the preset image features.

[0087] Step S103: If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0088] The similarity threshold refers to a pre-set similarity level.

[0089] Among them, the defect identification model refers to a neural network model that is built and trained based on a deep learning framework and used to identify the types of defects in power transmission lines, such as a convolutional neural network model.

[0090] Among them, the preset defect type refers to the common fault types of transmission lines.

[0091] Here, the first confidence level refers to the confidence level of the transmission line output by the trained defect recognition model based on the preprocessed visible light image under each preset defect type. It should be noted that the confidence level is used to represent the predicted probability corresponding to each preset defect type.

[0092] For example, the server obtains the current operating condition information of the transmission line, queries the correspondence between the current operating condition information and the candidate similarity threshold based on the current operating condition information, and obtains the candidate similarity threshold corresponding to the current operating condition information as the similarity threshold; when the similarity is less than or equal to the similarity threshold, the image feature vector corresponding to the preprocessed visible light image is extracted, and the image feature vector corresponding to the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence score of the transmission line under each preset defect type; when the similarity is greater than the similarity threshold, the target image feature with the highest similarity is selected from the preset image features, and the defect type corresponding to the target image feature is taken as the defect recognition result of the transmission line.

[0093] Step S104: Obtain X-ray images of the transmission line, input the X-ray images into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type.

[0094] Among them, X-ray images refer to grayscale images obtained by using X-ray imaging equipment adapted for power transmission line inspection to perform penetrating imaging of power transmission lines.

[0095] The second confidence level refers to the confidence level of the transmission line output by the trained defect recognition model based on X-ray images under each preset defect type.

[0096] For example, the server uses an X-ray imaging device connected to the power transmission line to perform a through-image to obtain an X-ray image of the power transmission line; then, it extracts the image feature vector corresponding to the X-ray image and inputs the image feature vector corresponding to the X-ray image into the trained defect recognition model to obtain the second confidence level of the power transmission line under each preset defect type.

[0097] Step S105: Determine the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level.

[0098] The target confidence level is used to represent the final confidence level of the transmission line under each preset defect type.

[0099] For example, the server performs validity verification on the first confidence level and the second confidence level to obtain the verification result; if the verification result indicates that both the first confidence level and the second confidence level have passed the verification, the first confidence level and the second confidence level are fused to obtain the target confidence level of the transmission line under each preset defect type.

[0100] Step S106: Select the defect types with a target confidence level greater than the preset confidence level from the preset defect types and use them as the target defect types of the transmission line. Based on the target defect types, obtain the defect identification results of the transmission line.

[0101] Among them, the pre-set confidence level refers to the pre-set confidence threshold.

[0102] Among them, the target defect type refers to the defect type among the preset defect types whose target confidence level is greater than the preset confidence level.

[0103] Among them, the defect identification results are used as the final output information to characterize the defect status of the transmission line.

[0104] For example, the server obtains the initial confidence threshold, operating years, and load status corresponding to the transmission line, and adjusts the initial confidence threshold according to the operating years and load status to obtain a preset confidence level. For instance, for older lines that have been in operation for more than 10 years and are operating under high load, the preset confidence threshold is appropriately lowered (e.g., from 0.7 to 0.6) to improve the sensitivity of defect identification and avoid missing hidden defects. For newly built lines or lines operating under low load, the preset confidence threshold is appropriately increased (e.g., from 0.7 to 0.8) to reduce the false detection rate caused by environmental interference. Then, from each preset defect type, the defect type with a target confidence level greater than the preset confidence level is selected and used as the target defect type of the transmission line. Finally, both the target defect type and the confidence level corresponding to the target defect type are used as the defect identification result of the transmission line.

[0105] In the aforementioned method for identifying transmission line defects, firstly, a visible light image of the transmission line to be analyzed is acquired. This image is then preprocessed to obtain a preprocessed visible light image. Next, image features corresponding to the preprocessed visible light image are extracted, and the similarity between these features and preset image features is determined. If the similarity is less than or equal to a similarity threshold, the preprocessed visible light image is input into a trained defect identification model to obtain the first confidence level of the transmission line under each preset defect type. Then, an X-ray image of the transmission line is acquired and input into the trained defect identification model to obtain the second confidence level of the transmission line under each preset defect type. Based on the first and second confidence levels, the target confidence level of the transmission line under each preset defect type is determined. Finally, from each preset defect type, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained. In this way, when identifying defects in transmission lines, visible light and X-ray images of the transmission line are acquired simultaneously and input into the trained defect identification model to obtain the first and second confidence scores for the corresponding defect types. The complementary characteristics of the two images are utilized to avoid the limitations of single-image detection. Furthermore, before inputting the visible light image into the model, its image features are extracted and compared with preset features for similarity verification. Only when the similarity is below a threshold is the defect identification process initiated, effectively filtering invalid samples such as background interference and image blurring, reducing model misjudgments. The first and second confidence scores are then weighted and fused to obtain the target confidence score. Defect types with target confidence scores above the threshold are selected as the final result, further improving the reliability and accuracy of the identification results and enhancing the accuracy of transmission line defect identification. Moreover, this process requires no human intervention, avoiding the subjective factors and errors inherent in manual experience-based identification methods, which can lead to lower accuracy in transmission line defect identification, thus improving the overall accuracy of transmission line defect identification.

[0106] In one exemplary embodiment, such as Figure 2 As shown, step S105 above, which determines the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level, specifically includes the following steps:

[0107] Step S201: Obtain the first weight corresponding to the first confidence level and the second weight corresponding to the second confidence level, and determine the confidence difference between the first confidence level and the second confidence level.

[0108] Step S202: When the confidence difference is less than or equal to the confidence difference threshold, the first confidence and the second confidence are weighted and summed according to the first weight and the second weight to obtain the target confidence of the transmission line under each preset defect type.

[0109] Step S203: If the confidence difference is greater than the confidence difference threshold, the first weight and the second weight are corrected to obtain the first target weight corresponding to the first confidence and the second target weight corresponding to the second confidence. The first confidence and the second confidence are weighted and summed according to the first target weight and the second target weight to obtain the target confidence of the transmission line under each preset defect type.

[0110] Here, the first weight refers to the weight coefficient pre-assigned to the first confidence level of the visible light image output.

[0111] The second weight refers to the weight coefficient pre-assigned to the second confidence level of the X-ray image output.

[0112] The confidence level difference refers to the absolute difference between the first confidence level and the second confidence level.

[0113] The confidence difference threshold refers to the pre-set confidence difference.

[0114] The first target weight refers to the weight coefficient obtained after correcting the first weight.

[0115] The second target weight refers to the weight coefficient obtained after correcting the second weight.

[0116] For example, the server determines a first weight corresponding to a first confidence level and a second weight corresponding to a second confidence level based on a preset weighting rule, and determines the absolute difference between the first and second confidence levels as the confidence difference. Then, based on a confidence difference threshold, the server judges the confidence difference. If the confidence difference is less than or equal to the confidence difference threshold, the first and second confidence levels are weighted and summed according to the first and second weights to obtain the target confidence level of the transmission line under each preset defect type. If the confidence difference is greater than the confidence difference threshold, the sharpness, signal-to-noise ratio, and defect feature recognition of the visible light image are input into the trained quality assessment result prediction model, and the sharpness, signal-to-noise ratio, and defect feature recognition of the X-ray image are input into the trained quality assessment result prediction model to obtain the visible light... The image quality assessment results are obtained from the image quality assessment results corresponding to the visible light image and the X-ray image. Then, based on the image quality assessment results corresponding to the visible light image and the X-ray image, the correction coefficients corresponding to the first weight and the second weight are determined. The first weight and the second weight are then corrected according to the correction coefficients corresponding to the first weight and the second weight to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level (the correction coefficients are positively correlated with the image quality assessment results; the higher the image quality assessment results, the larger the correction coefficient of the corresponding weight. The sum of the corrected first target weight and the second target weight is still 1). Next, the first confidence level and the second confidence level are weighted and summed according to the first target weight and the second target weight to obtain the target confidence level of the transmission line under each preset defect type.

[0117] It should be noted that the preset weight allocation rules are based on the detection characteristics of preset defect types. For surface defects such as insulator contamination and hardware corrosion, the first weight (corresponding to visible light images) is higher; for internal defects such as broken wires and broken core rods, the second weight (corresponding to X-ray images) is higher, and the sum of the first weight and the second weight is 1.

[0118] In this embodiment, the consistency of the two-dimensional detection results is determined by the difference between the first confidence level and the second confidence level, thereby achieving accurate fusion of the two-dimensional confidence levels. This further improves the fit between the target confidence level and the actual defect status of the transmission line, reduces the defect missed detection rate and false detection rate, and provides a more accurate basis for the identification of transmission line defects.

[0119] In an exemplary embodiment, step S203 above, which corrects the first weight and the second weight to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level, specifically includes the following: inputting the visible light image and the X-ray image into the cross-modal attention model to obtain the first feature contribution degree corresponding to the visible light image and the second feature contribution degree corresponding to the X-ray image; querying the correspondence between the feature contribution degree and the correction coefficient based on the first feature contribution degree and the second feature contribution degree to obtain the first correction coefficient corresponding to the first feature contribution degree and the second correction coefficient corresponding to the second feature contribution degree; and correcting the first weight and the second weight according to the first correction coefficient and the second correction coefficient to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level.

[0120] Among them, the cross-modal attention model refers to a multimodal feature fusion neural network model built based on the attention mechanism.

[0121] The first feature contribution refers to the feature contribution of the visible light image. Specifically, the feature contribution represents the quantified value of the contribution of the image features to the identification of a predetermined defect type in a power transmission line.

[0122] The second feature contribution refers to the feature contribution corresponding to the X-ray image.

[0123] The first correction coefficient refers to the correction coefficient that matches the contribution of the first feature.

[0124] The second correction coefficient refers to the correction coefficient that matches the contribution of the second feature.

[0125] For example, the server extracts the image feature vectors of the visible light image and the X-ray image, respectively, and inputs them into the cross-modal attention model to obtain the first feature contribution of the visible light image and the second feature contribution of the X-ray image. Then, based on the first and second feature contributions, the server queries the correspondence between the feature contribution and the correction coefficient to obtain the first correction coefficient corresponding to the first feature contribution and the second correction coefficient corresponding to the second feature contribution. Finally, the server corrects the first weight and the second weight according to the first and second correction coefficients to obtain the first target weight corresponding to the first confidence and the second target weight corresponding to the second confidence.

[0126] In this embodiment, by using a cross-modal attention mechanism to automatically focus on feature regions that are more discriminative for defect identification, the feature contribution of the two modal images is accurately quantified, avoiding the subjective bias of manually setting weights. Then, the initial weights are dynamically adjusted based on the corresponding correction coefficients matched with the feature contribution, which helps to improve the accuracy of dual-confidence fusion.

[0127] In an exemplary embodiment, step S102 above, extracting image features corresponding to the preprocessed visible light image, specifically includes the following: inputting the preprocessed visible light image into a texture feature extraction model to obtain texture features corresponding to the preprocessed visible light image; inputting the preprocessed visible light image into a shape feature extraction model to obtain shape features corresponding to the preprocessed visible light image; inputting the preprocessed visible light image into a color feature extraction model to obtain color features corresponding to the preprocessed visible light image; inputting the preprocessed visible light image into a component structure feature extraction model to obtain component structure features corresponding to the preprocessed visible light image; the component structure features are used to represent the structural features of the transmission line components of the transmission line; based on the texture features, shape features, color features, and component structure features, the image features corresponding to the preprocessed visible light image are obtained.

[0128] Among them, texture feature extraction model refers to network model used to extract texture features corresponding to preprocessed visible light images, such as fusion feature extraction model based on gray-level co-occurrence matrix combined with local binary mode.

[0129] Among them, texture features are a set of quantitative parameters used to characterize the surface texture attributes of transmission line components, specifically including parameters such as contrast, correlation, entropy, and energy.

[0130] Among them, the shape feature extraction model refers to the network model used to extract the shape features corresponding to the preprocessed visible light image, such as the geometric feature extraction model based on edge detection, Hough transform and contour moment analysis.

[0131] Among them, shape features are a set of quantitative parameters used to characterize the geometric shape of transmission line components, specifically including parameters such as the area, perimeter, aspect ratio, circularity, number of straight segments and angles of the component outline.

[0132] Among them, the color feature extraction model refers to the network model used to extract the color features corresponding to the preprocessed visible light image, such as the statistical feature extraction model based on color space.

[0133] Among them, color features are a set of quantitative parameters used to characterize the color attributes of the surface of transmission line components, specifically including parameters such as hue, saturation, and brightness.

[0134] Among them, the component structure feature extraction model refers to the network model used to extract the component structure features corresponding to the preprocessed visible light image, such as the fusion model based on target detection algorithm and topology analysis.

[0135] Among them, the component structural features are a set of quantitative features used to characterize the composition and connection relationship of transmission line components, specifically including parameters such as the number of transmission line components, their relative positions, and topological connection methods.

[0136] Among them, transmission line components refer to the core components that make up a transmission line, specifically including conductors, insulator strings, hardware, towers, crossarm covers and other components that directly participate in the transmission or support of electrical energy.

[0137] Among them, structural features are used to characterize the composition of an object, the relative positions of its parts, and the connection relationships.

[0138] For example, the server inputs the preprocessed visible light image into a texture feature extraction model to obtain the texture features corresponding to the preprocessed visible light image; then, it inputs the preprocessed visible light image into a shape feature extraction model to obtain the shape features corresponding to the preprocessed visible light image; then, it inputs the preprocessed visible light image into a color feature extraction model to obtain the color features corresponding to the preprocessed visible light image; next, it inputs the preprocessed visible light image into a component structure feature extraction model to obtain the structural features of the power transmission line components corresponding to the preprocessed visible light image, which are used as component structure features; then, the texture features, shape features, color features, and component structure features are dimensionally aligned and normalized to obtain the processed texture features, shape features, color features, and component structure features, and feature splicing is performed on the processed texture features, shape features, color features, and component structure features to obtain the image features corresponding to the preprocessed visible light image.

[0139] In this embodiment, by extracting four types of features from visible light images—texture, shape, color, and component structure—in modules and fusing them, the limitations of single features in defect representation are overcome, and a comprehensive characterization of the state of transmission line components is achieved. The fused image features have richer defect discrimination dimensions, effectively improving the sensitivity and recognition of various defects by the subsequent defect identification model.

[0140] In an exemplary embodiment, step S102 above includes preset image features such as preset texture features, preset shape features, preset color features, and preset component structure features.

[0141] Therefore, step S102 above determines the similarity between image features and preset image features, specifically including the following: determining the first similarity between texture features and preset texture features; determining the second similarity between shape features and preset shape features; determining the third similarity between color features and preset color features; determining the fourth similarity between component structure features and preset component structure features; and fusing the first, second, third, and fourth similarities to obtain the similarity between image features and preset image features.

[0142] Among them, the preset texture features refer to the texture features extracted from the visible light image corresponding to the preset transmission line.

[0143] Among them, the preset shape features refer to the shape features extracted from the visible light image corresponding to the preset transmission line.

[0144] Among them, the preset color features refer to the color features extracted from the visible light image corresponding to the preset transmission line.

[0145] Among them, the preset component structure features refer to the component structure features extracted from the visible light image corresponding to the preset transmission line.

[0146] The first similarity refers to the similarity between the texture features and the preset texture features.

[0147] The second similarity refers to the similarity between the shape features and the preset shape features.

[0148] The third similarity refers to the similarity between the color feature and the preset color feature.

[0149] The fourth similarity refers to the similarity between the structural features of the component and the preset structural features of the component.

[0150] For example, the server uses a cosine similarity algorithm to calculate the cosine value of the angle between the texture feature and the preset texture feature, as the first similarity between the texture feature and the preset texture feature; then, it uses a Hausdorff distance algorithm to calculate the mean minimum distance between the contour point set of the shape feature and the contour point set of the preset shape feature, as the second similarity between the shape feature and the preset shape feature; then, it uses a Bach distance algorithm to calculate the Bach distance between the color feature and the preset color feature, as the third similarity between the color feature and the preset color feature; next, it uses a Jaccard similarity coefficient algorithm to calculate the intersection-union ratio (IUU) between the component structure feature and the preset component structure feature, as the fourth similarity between the component structure feature and the preset component structure feature; then, according to the fusion coefficients of the first, second, third, and fourth similarities, it performs fusion processing on the first, second, third, and fourth similarities to obtain the similarity between the image feature and the preset image feature.

[0151] In this embodiment, the similarity between the real-time extracted texture, shape, color, and component structure features and the corresponding preset features is calculated and fused in a multi-dimensional manner. This avoids the one-sidedness of single-dimensional feature matching and comprehensively reflects the overall fit between real-time image features and preset features, significantly improving the accuracy and robustness of feature matching and providing a reliable similarity judgment basis for the accurate identification of subsequent transmission line defects.

[0152] In an exemplary embodiment, step S104, before inputting the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type, specifically includes the following: performing multi-scale decomposition processing on the X-ray image to obtain high-frequency image components and low-frequency image components in the X-ray image; performing denoising processing on the high-frequency image components to obtain denoised high-frequency image components, and performing smoothing processing on the low-frequency image components to obtain smoothed low-frequency image components; and performing reconstruction processing on the denoised high-frequency image components and the smoothed low-frequency image components to obtain the processed X-ray image.

[0153] Therefore, step S104 above, which inputs the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type, specifically includes the following: inputting the processed X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type.

[0154] Among them, the high-frequency image component is used to represent the image components of the edges, details, and abrupt grayscale regions of the X-ray image.

[0155] Among them, the denoised high-frequency image component refers to the high-frequency image component after denoising processing.

[0156] Among them, the low-frequency image component is used to represent the overall grayscale distribution, smooth areas, and background contours of the X-ray image.

[0157] Among them, the smoothed low-frequency image component refers to the low-frequency image component after smoothing.

[0158] For example, the server uses a wavelet decomposition algorithm to perform multi-scale decomposition processing on the X-ray image, obtaining high-frequency and low-frequency image components. Next, a wavelet threshold denoising algorithm is used to denoise the high-frequency image components, resulting in denoised high-frequency image components. A Gaussian weighted smoothing algorithm is used to smooth the low-frequency image components, resulting in smoothed low-frequency image components. Then, the denoised high-frequency and smoothed low-frequency image components are input into an inverse wavelet transform model. The inverse wavelet transform model reconstructs the denoised high-frequency and smoothed low-frequency image components, resulting in a processed X-ray image. Finally, the image feature vector corresponding to the processed X-ray image is extracted and input into a trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type.

[0159] In this embodiment, by implementing a multi-scale decomposition and reconstruction strategy for component differential processing of X-ray images, the X-ray images are split into high-frequency components that carry defect edges and details and low-frequency components that characterize the overall shape of the image. This solves the problem of low recognition accuracy caused by noise masking defect features in X-ray images, and provides high-quality image data support for the subsequent defect recognition model to output accurate second confidence scores.

[0160] In one exemplary embodiment, such as Figure 3 As shown, another method for identifying transmission line defects is provided. Taking the application of this method to a server as an example, the specific steps include:

[0161] Step S301: Obtain a visible light image of the transmission line to be analyzed, and preprocess the visible light image to obtain a preprocessed visible light image.

[0162] Step S302: Input the preprocessed visible light image into the texture feature extraction model to obtain the texture features corresponding to the preprocessed visible light image.

[0163] Step S303: Input the preprocessed visible light image into the shape feature extraction model to obtain the shape features corresponding to the preprocessed visible light image.

[0164] Step S304: Input the preprocessed visible light image into the color feature extraction model to obtain the color features corresponding to the preprocessed visible light image.

[0165] Step S305: Input the preprocessed visible light image into the component structure feature extraction model to obtain the component structure features corresponding to the preprocessed visible light image; the component structure features are used to represent the structural features of the transmission line components of the transmission line.

[0166] Step S306: Based on texture features, shape features, color features, and component structure features, obtain the image features corresponding to the preprocessed visible light image.

[0167] Step S307: Determine the similarity between the image features and the preset image features.

[0168] Step S308: If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type.

[0169] Step S309: Obtain X-ray images of the transmission line.

[0170] Step S310: Perform multi-scale decomposition processing on the X-ray image to obtain the high-frequency image component and the low-frequency image component in the X-ray image.

[0171] Step S311: Denoise the high-frequency image components to obtain the denoised high-frequency image components, and smooth the low-frequency image components to obtain the smoothed low-frequency image components.

[0172] Step S312: Reconstruct the denoised high-frequency image components and the smoothed low-frequency image components to obtain the processed X-ray image.

[0173] Step S313: Input the processed X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type.

[0174] Step S314: Determine the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level.

[0175] Step S315: Select the defect types with a target confidence level greater than the preset confidence level from the preset defect types and use them as the target defect types of the transmission line. Based on the target defect types, obtain the defect identification results of the transmission line.

[0176] In the aforementioned method for identifying transmission line defects, visible light and X-ray images of the transmission line are acquired simultaneously during defect identification. These images are then input into a trained defect identification model to obtain the first and second confidence scores for the corresponding defect types. The complementary nature of the two images avoids the limitations of single-image detection. Furthermore, before inputting the visible light image into the model, its image features are extracted and compared with preset features for similarity verification. Only when the similarity score is below a threshold is the defect identification process initiated, effectively filtering out invalid samples such as background interference and image blurring, reducing model misjudgments. The first and second confidence scores are then weighted and fused to obtain the target confidence score. Defect types with target confidence scores above the threshold are selected as the final result, further improving the reliability and accuracy of the identification results and enhancing the accuracy of transmission line defect identification. Moreover, this process requires no manual intervention, avoiding the subjective factors and errors inherent in manual experience-based identification methods, which can lead to lower accuracy in transmission line defect identification. This, in turn, improves the overall accuracy of transmission line defect identification.

[0177] In one exemplary embodiment, to more clearly illustrate the transmission line defect identification method provided by the embodiments of this application, the following specific embodiment will be used to describe the transmission line defect identification method in detail. In one embodiment, this application also provides another transmission line defect identification method. Specifically, it includes the following:

[0178] (1) Obtain visible light images of the transmission line and preprocess the visible light images, including operations such as erosion, dilation, and color change, to improve the effect and speed of subsequent processing.

[0179] (2) Extract the image features of the preprocessed visible light image and perform defect feature matching on the image features of the preprocessed visible light image. If the similarity between the image features and the preset image features is greater than the similarity threshold (the threshold can be adjusted according to the shooting environment (lighting, angle)), the defect identification result corresponding to the preset image features is directly used as the defect identification result of the transmission line; otherwise, proceed to the next step of analysis.

[0180] (3) Input the preprocessed visible light image into multiple defect recognition models (target detection algorithms) adapted to the visible light image, and obtain the confidence level of the transmission line output by each defect recognition model under each preset defect type. According to the model weight of each defect recognition model, the confidence level of the transmission line output by each defect recognition model under each preset defect type is fused to obtain the first confidence level of the transmission line under each preset defect type.

[0181] (4) Input the X-ray image of the transmission line into the defect recognition model (target detection algorithm) adapted to the X-ray image to obtain the second confidence level of the transmission line under each preset defect type.

[0182] (5) The first confidence level and the second confidence level of the transmission line under each preset defect type are fused to obtain the target confidence level of the transmission line under each preset defect type.

[0183] (6) Select the defect types with a target confidence level greater than the preset confidence level from each preset defect type, and use them as the target defect types of the transmission line. Then, use the target defect types and corresponding target confidence levels of the transmission line as the defect identification results of the transmission line.

[0184] In the above embodiments, when identifying transmission line defects, visible light and X-ray images of the transmission line are acquired simultaneously and input into the trained defect identification model to obtain the first and second confidence scores for the corresponding defect types. The complementary characteristics of the two images are utilized to avoid the limitations of single-image detection. Furthermore, before inputting the visible light image into the model, its image features are extracted and compared with preset features for similarity verification. Only when the similarity is below a threshold is the defect identification process initiated, effectively filtering invalid samples such as background interference and image blurring, reducing model misjudgments. The first and second confidence scores are then weighted and fused to obtain the target confidence score. Defect types with target confidence scores higher than the threshold are selected as the final result, further improving the reliability and accuracy of the identification results and enhancing the accuracy of transmission line defect identification. Moreover, this process requires no manual intervention, avoiding the subjective factors and errors inherent in manual experience-based identification methods, which can lead to lower accuracy in transmission line defect identification, thus improving the overall accuracy of transmission line defect identification. Meanwhile, by optimizing image quality through preprocessing, making rapid and accurate judgments through feature matching, weighted fusion of visible light confidence through multiple models, cross-modal fusion of X-ray deep defect information, and multi-layer verification and information complementarity design for screening effective defects with dual thresholds, the accuracy of power transmission line defect identification can be improved.

[0185] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0186] Based on the same inventive concept, this application also provides a transmission line defect identification device for implementing the above-described transmission line defect identification method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the transmission line defect identification device provided below can be found in the limitations of the transmission line defect identification method described above, and will not be repeated here.

[0187] In one exemplary embodiment, such as Figure 4 As shown, a power transmission line defect identification device is provided, comprising: an image acquisition module 401, a similarity determination module 402, a first identification module 403, a second identification module 404, a confidence determination module 405, and a result determination module 406, wherein:

[0188] The image acquisition module 401 is used to acquire a visible light image of the transmission line to be analyzed, and to preprocess the visible light image to obtain a preprocessed visible light image.

[0189] The similarity determination module 402 is used to extract the image features corresponding to the preprocessed visible light image and determine the similarity between the image features and the preset image features.

[0190] The first identification module 403 is used to input the preprocessed visible light image into the trained defect identification model when the similarity is less than or equal to the similarity threshold, so as to obtain the first confidence level of the transmission line under each preset defect type.

[0191] The second identification module 404 is used to acquire X-ray images of the transmission line, input the X-ray images into the trained defect identification model, and obtain the second confidence level of the transmission line under each preset defect type.

[0192] The confidence level determination module 405 is used to determine the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level.

[0193] The result determination module 406 is used to select the defect types with a target confidence level greater than the preset confidence level from each preset defect type, and use them as the target defect types of the transmission line. Based on the target defect types, the defect identification results of the transmission line are obtained.

[0194] In an exemplary embodiment, the confidence level determination module 405 is further configured to obtain a first weight corresponding to a first confidence level and a second weight corresponding to a second confidence level, and determine the confidence level difference between the first confidence level and the second confidence level; if the confidence level difference is less than or equal to a confidence level difference threshold, the first confidence level and the second confidence level are weighted and summed according to the first weight and the second weight to obtain the target confidence level of the transmission line under each preset defect type; if the confidence level difference is greater than the confidence level difference threshold, the first weight and the second weight are corrected to obtain a first target weight corresponding to the first confidence level and a second target weight corresponding to the second confidence level, and the first confidence level and the second confidence level are weighted and summed according to the first target weight and the second target weight to obtain the target confidence level of the transmission line under each preset defect type.

[0195] In an exemplary embodiment, the confidence determination module 405 is further configured to input the visible light image and the X-ray image into the cross-modal attention model respectively to obtain a first feature contribution corresponding to the visible light image and a second feature contribution corresponding to the X-ray image; based on the first feature contribution and the second feature contribution, query the correspondence between the feature contribution and the correction coefficient respectively to obtain a first correction coefficient corresponding to the first feature contribution and a second correction coefficient corresponding to the second feature contribution; and perform correction processing on the first weight and the second weight according to the first correction coefficient and the second correction coefficient to obtain a first target weight corresponding to the first confidence and a second target weight corresponding to the second confidence.

[0196] In an exemplary embodiment, the similarity determination module 402 is further configured to input the preprocessed visible light image into a texture feature extraction model to obtain the texture features corresponding to the preprocessed visible light image; input the preprocessed visible light image into a shape feature extraction model to obtain the shape features corresponding to the preprocessed visible light image; input the preprocessed visible light image into a color feature extraction model to obtain the color features corresponding to the preprocessed visible light image; input the preprocessed visible light image into a component structure feature extraction model to obtain the component structure features corresponding to the preprocessed visible light image; the component structure features are used to represent the structural features of the transmission line components of the transmission line; and based on the texture features, shape features, color features, and component structure features, the image features corresponding to the preprocessed visible light image are obtained.

[0197] In an exemplary embodiment, the similarity determination module 402 is further configured to determine a first similarity between the texture feature and a preset texture feature; determine a second similarity between the shape feature and a preset shape feature; determine a third similarity between the color feature and a preset color feature; determine a fourth similarity between the component structure feature and a preset component structure feature; and perform a fusion process on the first similarity, second similarity, third similarity and fourth similarity to obtain the similarity between the image feature and the preset image feature.

[0198] In an exemplary embodiment, the transmission line defect identification device further includes an image processing module for inputting the processed X-ray image into the trained defect identification model to obtain a second confidence level of the transmission line under each preset defect type.

[0199] Each module in the aforementioned transmission line defect identification device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0200] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data such as visible light images and X-ray images. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for identifying defects in power transmission lines.

[0201] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0202] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0203] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described method embodiments.

[0204] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0205] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0206] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0207] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for identifying defects in transmission lines, characterized in that, The method includes: A visible light image of the transmission line to be analyzed is acquired, and the visible light image is preprocessed to obtain a preprocessed visible light image. Extract the image features corresponding to the preprocessed visible light image, and determine the similarity between the image features and preset image features; If the similarity is less than or equal to the similarity threshold, the preprocessed visible light image is input into the trained defect recognition model to obtain the first confidence level of the transmission line under each preset defect type. Obtain an X-ray image of the transmission line, input the X-ray image into the trained defect recognition model, and obtain the second confidence level of the transmission line under each preset defect type; Based on the first confidence level and the second confidence level, the target confidence level of the transmission line under each preset defect type is determined; From the preset defect types, defect types with a target confidence level greater than a preset confidence level are selected as the target defect types of the transmission line. Based on the target defect types, the defect identification result of the transmission line is obtained.

2. The method according to claim 1, characterized in that, The step of determining the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level includes: Obtain the first weight corresponding to the first confidence level and the second weight corresponding to the second confidence level, and determine the confidence level difference between the first confidence level and the second confidence level; When the confidence difference is less than or equal to the confidence difference threshold, the first confidence and the second confidence are weighted and summed according to the first weight and the second weight to obtain the target confidence of the transmission line under each preset defect type. If the confidence difference is greater than the confidence difference threshold, the first weight and the second weight are corrected to obtain the first target weight corresponding to the first confidence and the second target weight corresponding to the second confidence. The first confidence and the second confidence are then weighted and summed according to the first target weight and the second target weight to obtain the target confidence of the transmission line under each preset defect type.

3. The method according to claim 2, characterized in that, The step of correcting the first weight and the second weight to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level includes: The visible light image and the X-ray image are respectively input into a cross-modal attention model to obtain the first feature contribution value corresponding to the visible light image and the second feature contribution value corresponding to the X-ray image; Based on the first feature contribution and the second feature contribution respectively, query the correspondence between feature contribution and correction coefficient to obtain the first correction coefficient corresponding to the first feature contribution and the second correction coefficient corresponding to the second feature contribution; The first weight and the second weight are corrected according to the first correction coefficient and the second correction coefficient to obtain the first target weight corresponding to the first confidence level and the second target weight corresponding to the second confidence level.

4. The method according to claim 1, characterized in that, The step of extracting the image features corresponding to the preprocessed visible light image includes: The preprocessed visible light image is input into the texture feature extraction model to obtain the texture features corresponding to the preprocessed visible light image. The preprocessed visible light image is input into the shape feature extraction model to obtain the shape features corresponding to the preprocessed visible light image; The preprocessed visible light image is input into the color feature extraction model to obtain the color features corresponding to the preprocessed visible light image. The preprocessed visible light image is input into the component structure feature extraction model to obtain the component structure features corresponding to the preprocessed visible light image; the component structure features are used to represent the structural features of the transmission line components of the transmission line; Based on the texture features, shape features, color features, and component structure features, the image features corresponding to the preprocessed visible light image are obtained.

5. The method according to claim 4, characterized in that, The preset image features include preset texture features, preset shape features, preset color features, and preset component structure features; Determining the similarity between the image features and preset image features includes: Determine the first similarity between the texture feature and the preset texture feature; Determine a second similarity between the shape feature and the preset shape feature; Determine the third similarity between the color feature and the preset color feature; A fourth similarity is determined between the structural features of the component and the preset structural features of the component; The first similarity, the second similarity, the third similarity, and the fourth similarity are fused to obtain the similarity between the image features and the preset image features.

6. The method according to any one of claims 1 to 5, characterized in that, Before inputting the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type, the method further includes: The X-ray image is subjected to multi-scale decomposition processing to obtain the high-frequency image component and the low-frequency image component in the X-ray image; The high-frequency image component is denoised to obtain the denoised high-frequency image component, and the low-frequency image component is smoothed to obtain the smoothed low-frequency image component. The denoised high-frequency image components and the smoothed low-frequency image components are reconstructed to obtain the processed X-ray image. The step of inputting the X-ray image into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type includes: The processed X-ray image is input into the trained defect recognition model to obtain the second confidence level of the transmission line under each preset defect type.

7. A transmission line defect identification device, characterized in that, The device includes: The image acquisition module is used to acquire a visible light image of the transmission line to be analyzed, and to preprocess the visible light image to obtain a preprocessed visible light image. The similarity determination module is used to extract the image features corresponding to the preprocessed visible light image and determine the similarity between the image features and preset image features; The first identification module is used to input the preprocessed visible light image into the trained defect identification model when the similarity is less than or equal to the similarity threshold, so as to obtain the first confidence level of the transmission line under each preset defect type. The second identification module is used to acquire X-ray images of the transmission line, input the X-ray images into the trained defect identification model, and obtain the second confidence level of the transmission line under each preset defect type. A confidence level determination module is used to determine the target confidence level of the transmission line under each preset defect type based on the first confidence level and the second confidence level; The result determination module is used to select the defect type with a target confidence level greater than a preset confidence level from the preset defect types, and use it as the target defect type of the transmission line. Based on the target defect type, the defect identification result of the transmission line is obtained.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images