Method and system for identifying abnormal region of power transmission line based on infrared thermal image feature fusion

Abstract

The invention relates to the technical field of power system monitoring, and discloses an infrared thermal image feature fusion-based power transmission line abnormal area identification method, which comprises the following steps of: acquiring infrared thermal image data of a power transmission line, and preprocessing the data; the invention further provides an infrared thermal image feature fusion power transmission line abnormal area identification system, which comprises a data acquisition module, an infrared thermal image feature fusion module, an infrared thermal image feature fusion module, an infrared thermal image feature fusion module, an infrared thermal image feature fusion module and an infrared thermal image feature fusion module, and is characterized in that the data acquisition module is used for acquiring infrared thermal image data of a power transmission line; and the preprocessing module is used for preprocessing the collected infrared thermal image data. According to the method, through combination of multi-scale wavelet transform, nonlinear dynamic modeling, graph neural network optimization and reinforcement learning, the abnormal region identification precision is improved, the robustness and adaptive ability of the system in a complex environment are enhanced, and it is ensured that the abnormal region can still be efficiently and accurately identified under dynamic change and noise interference.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring, and in particular to a method and system for identifying abnormal areas of power transmission lines by integrating infrared thermal image features. Background Art

[0002] With the expansion of the scale of power systems and the increase in power demand, the health monitoring of transmission lines has become increasingly important. Equipment failure or temperature abnormalities in transmission lines may cause serious power accidents, affecting social production and residents' lives. Therefore, timely detection and handling of abnormal conditions of transmission lines is crucial to ensuring the stability and safety of power systems. Infrared thermal imaging technology, as a non-contact temperature detection method, monitors the temperature changes on the surface of equipment in real time through infrared cameras, and can quickly identify potential abnormal areas such as overheating and failures.

[0003] In the prior art, infrared thermal image processing mainly relies on traditional image analysis techniques, such as threshold-based anomaly detection methods and edge detection algorithms. These methods can determine whether the equipment is abnormal through simple temperature thresholds, and perform edge detection on the image to identify potential fault areas. In some simple monitoring scenarios, traditional methods can effectively identify abnormal temperature areas and help maintenance personnel quickly locate faults to a certain extent. These technologies realize the automation of temperature monitoring, reduce the workload of manual inspections, and improve the efficiency of preliminary fault diagnosis.

[0004] Although existing technologies have achieved certain success in certain application scenarios, their shortcomings are also obvious. First, existing threshold-based detection methods lack flexibility and cannot adapt to temperature changes in different environments. Due to the constant changes in equipment and environmental conditions, the accuracy of these methods is greatly reduced when dealing with complex or changing environments. Secondly, existing technologies mostly rely on static features, such as temperature thresholds or the overall brightness distribution of images, ignoring the dynamics and spatiotemporal evolution characteristics of temperature changes. Problems such as overheating and failure of equipment are usually a continuously changing process. Traditional methods fail to consider these spatiotemporal characteristics, resulting in limited recognition accuracy. Furthermore, existing methods have a weak response to subtle changes or local failures in images, and are prone to miss relatively small abnormal areas, affecting the overall performance of the system. Finally, although some modern image processing technologies (such as deep learning) have begun to be applied to infrared thermal image analysis, most existing methods rely on a large amount of labeled data for training, and have poor adaptability to environmental changes, and there is still a lot of room for optimization. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a method and system for identifying abnormal areas of power transmission lines by fusion of infrared thermal imaging features, which solves the problems in the prior art of low accuracy in identifying abnormal areas in complex environments, insufficient capture of dynamic changes, and lack of adaptive optimization.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features, comprising the following steps: Collect infrared thermal imaging data of the transmission line and pre-process the data; The pre-processed infrared thermal image is decomposed into multiple scales by wavelet transform to extract the low-frequency part and the high-frequency part; The extracted multi-scale features are embedded into high-dimensional data through the manifold learning method to obtain a feature set in a low-dimensional space; The spatiotemporal evolution of the abnormal area is modeled based on a nonlinear dynamic model and fused with the embedded feature set; Use graph neural networks to optimize the fused features, update the relationship between nodes, and output the optimized features; Use the optimized features to identify abnormal areas and obtain the locations of abnormal areas; The abnormal area identification strategy is optimized through reinforcement learning and global optimization algorithms to improve the accuracy of the identification results.

[0007] Preferably, the pretreatment comprises: Remove noise from the collected infrared thermal imaging data; Perform contrast enhancement processing on infrared thermal images; Perform histogram equalization on the processed image.

[0008] Preferably, the wavelet transform adopts discrete wavelet transform, the low-frequency part includes the overall temperature distribution characteristics, the high-frequency part includes the detail characteristics, and the extraction process of the low-frequency part and the high-frequency part includes: Perform wavelet transform on the image data to obtain low-frequency and high-frequency features at different scales; The low-frequency and high-frequency parts are denoised, and the parts with significant features are retained.

[0009] Preferably, the high-dimensional data embedding is performed using a principal component analysis method or a t-SNE method, and the feature set after the data embedding includes: Perform dimensionality reduction on high-dimensional data to extract the most representative features; The reduced feature set is used as input for subsequent processing.

[0010] Preferably, the nonlinear dynamic modeling adopts the Lorenz system to describe the dynamic evolution process of the abnormal area, and the dynamic evolution process includes: Based on the nonlinear equation of the Lorenz system, the dynamic evolution of abnormal areas in infrared thermal images over time is simulated. By using the three-dimensional state variables of the Lorenz equation, the temperature change, expansion and shrinkage of the abnormal area are described. In the Lorenz system, by selecting appropriate parameters and adjusting according to the actual infrared thermal imaging data, the model can better match the actual temperature change pattern; The dynamic evolution process obtained by Lorenz system modeling is fused with the multi-scale features extracted from infrared thermal images through a weighted fusion method.

[0011] Preferably, the graph neural network optimization includes: Each image feature node is used as a node of the graph, and the similarity between nodes is represented by the edges of the graph; Normalize the adjacency matrix of the graph, use graph neural network to propagate features, and update node features; The node features are integrated using graph convolution operations to obtain the optimized feature set as the final output.

[0012] Preferably, the abnormal area recognition adopts a support vector machine or a decision tree method, and the recognition results include: Establish a classification model based on training data to automatically identify abnormal areas; Output the location coordinates and temperature information of the identified abnormal area.

[0013] Preferably, the reinforcement learning algorithm includes: Define state space, action space, and reward function; Use Q-learning or deep Q-network to optimize abnormal region identification strategy; The strategy is adjusted according to the feedback reward function to maximize the recognition accuracy.

[0014] Preferably, the global optimization uses a genetic algorithm or a simulated annealing algorithm to globally optimize the identification strategy, and the optimization process includes: Searching for parameters of the identification strategy using genetic algorithms or simulated annealing algorithms; The stability and accuracy of the strategy are optimized through global search.

[0015] The present invention also provides a transmission line abnormal area recognition system using infrared thermal image feature fusion, comprising: Data acquisition module, used to collect infrared thermal imaging data of power transmission lines; A preprocessing module, used for preprocessing the collected infrared thermal imaging data; Feature extraction module, used to perform wavelet transform, multi-scale analysis, data embedding and nonlinear dynamic modeling on the pre-processed infrared thermal imaging data to extract multi-dimensional features; Graph neural network optimization module, used to optimize the graph structure of the extracted features; An abnormal area recognition module is used to locate and identify abnormal areas based on optimized features; Reinforcement learning and global optimization modules are used to optimize abnormal area identification strategies and improve identification accuracy.

[0016] The present invention provides a method and system for identifying abnormal areas of power transmission lines by integrating infrared thermal imaging features. It has the following beneficial effects: 1. The present invention extracts multi-scale features of images through wavelet transform, combines the Lorenz system for nonlinear dynamic modeling, and accurately captures static and dynamic information in the image. Compared with traditional methods that only focus on static features of images, the present invention can dynamically analyze the temperature changes and expansion process of abnormal areas, significantly improving the accuracy and comprehensiveness of abnormal area identification.

[0017] 2. The present invention optimizes the relationship between feature nodes through graph neural networks, so that each feature in the image can better integrate contextual information. Different from traditional feature processing methods, graph neural networks can effectively mine the spatial and temporal dependencies between image features, greatly improving the accuracy of locating abnormal areas, especially in complex environments.

[0018] 3. The present invention optimizes the recognition strategy through reinforcement learning, combined with global optimization methods such as genetic algorithms, so that the system can adapt to different environmental conditions. In traditional methods, the strategy is fixed and easily affected by environmental changes, while the present invention improves the robustness and accuracy of the system in complex environments through continuous learning and adjustment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a flow chart of the method of the present invention; Figure 2 It is a system structure diagram of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] Please see attached Figure 1 The embodiment of the present invention provides a method for identifying abnormal areas of power transmission lines by integrating infrared thermal imaging features, comprising the following steps: S1. Collect and pre-process data: collect infrared thermal imaging data of the transmission line and pre-process the data; This step lays the foundation for subsequent multi-scale feature extraction, dynamic modeling and abnormal area identification. Data acquisition and preprocessing directly affect the performance of subsequent modules and ensure the data accuracy and robustness during subsequent processing steps. In order to overcome the possible influence of environmental noise, image distortion, etc., this step uses a series of effective image processing technologies to improve data quality.

[0022] In this embodiment, the temperature distribution data of the power transmission line is collected by using a high-precision infrared thermal imager. Infrared thermal imagers usually use the principle of infrared radiation to accurately measure and record the temperature information on the surface of power equipment. The collected infrared thermal image is usually a two-dimensional image, and each pixel corresponds to the temperature of different areas on the surface of the equipment.

[0023] After image acquisition, in order to eliminate noise and interference during the acquisition process, the image must be preprocessed. Generally, image preprocessing includes operations such as noise removal, contrast enhancement, and histogram equalization. These operations can significantly improve image quality, eliminate the impact of environmental factors on data, and thus improve the accuracy of subsequent analysis processes.

[0024] During the infrared thermal imaging data acquisition process, due to environmental changes or equipment limitations, the image may contain certain noise. These noises mainly come from the electronic noise of the sensor, interference signals during image transmission, and external environmental factors (such as temperature fluctuations, weather changes, etc.). In order to solve this problem, the present invention uses Gaussian filtering or median filtering to remove noise.

[0025] Gaussian filtering: Generally, Gaussian filtering is used in the image denoising process to smooth out the noise in the image. The formula for the Gaussian filter is: ; in, is the value of the Gaussian filter, is the standard deviation, which indicates the width of the filter and controls the degree of denoising; and are pixel coordinates.

[0026] Median filtering: As an option, median filtering can effectively remove salt and pepper noise and retain image edge information. The basic principle of median filtering is to replace the target pixel value by taking the median of the neighboring pixels.

[0027] The temperature changes of infrared thermal images are often subtle, especially in complex background environments, and abnormal areas may be difficult to identify. Therefore, contrast enhancement can highlight the details in the image, making the abnormal area more prominent. In this embodiment, histogram equalization technology is used to enhance the contrast of the image.

[0028] Histogram equalization is a commonly used image enhancement technique, which aims to make the grayscale range of the image more uniform by adjusting the grayscale distribution of the image. Specifically, this method improves the contrast of the image by equalizing the cumulative distribution function (CDF) of the image, thereby enhancing the local details of the image. The formula is as follows: ; in, is the image grayscale The cumulative distribution function of Gray level in the image The probability density function of : Represents a gray level or brightness value in an image; : Indicates the current accumulated grayscale value.

[0029] While performing contrast enhancement, histogram equalization redistributes the grayscale of the image so that the brightness range of the image is more evenly distributed. Through this processing, the temperature differences in the image will be more obvious and abnormal areas will be easier to detect.

[0030] In one possible implementation, multiple enhancement techniques may be combined to enhance image details. For example, after noise removal, an adaptive contrast enhancement method is used to adaptively adjust contrast in a local area to ensure that abnormal areas are clearly represented under various environmental conditions.

[0031] In this embodiment, the infrared thermal image after the above preprocessing will enter the wavelet transform module. The image after denoising and enhancement has a higher signal-to-noise ratio and clearer temperature distribution information, which can provide high-quality input for subsequent multi-scale wavelet transform and feature extraction. By further decomposing the image, features at different scales are extracted, providing more accurate basic data for subsequent feature fusion and abnormal area identification.

[0032] Through this series of preprocessing, the system can effectively reduce the impact of environmental interference and equipment errors, while improving the detection effect of abnormal areas, thereby ensuring the accuracy and robustness of the entire abnormal area identification method.

[0033] Through accurate acquisition and multi-level preprocessing of infrared thermal imaging data, noise is removed, contrast is enhanced, and grayscale distribution of the image is balanced. This step ensures the efficiency and reliability of subsequent feature extraction and abnormal area identification, and provides accurate input data for the entire system.

[0034] S2. Multi-scale wavelet transform feature extraction: The pre-processed infrared thermal image is decomposed into multiple scales through wavelet transform to extract the low-frequency part and the high-frequency part; Multi-scale wavelet transform can effectively extract information from images at different frequency scales, ensuring that the system can perform effective detection regardless of the size or location of the abnormal area.

[0035] In this embodiment, after the preprocessing steps (such as denoising, contrast enhancement, and histogram equalization), the infrared thermal image already has a higher signal-to-noise ratio and clearer temperature distribution information. The application of wavelet transform at this stage can further explore the deeper local and global features in the image. Through wavelet transform, the low-frequency and high-frequency parts of the image are separated to facilitate more effective processing of features of different scales.

[0036] Generally speaking, wavelet transform, as a classic signal processing tool, is widely used in image analysis, especially in scenarios where information at different scales needs to be captured. Its main advantage is that it can provide localized information in both space and frequency, which is very helpful for identifying abnormal areas in infrared thermal images of power transmission lines, especially edges and local faults.

[0037] Specifically, in this embodiment, the image is first decomposed by discrete wavelet transform (DWT) to extract the low-frequency part and the high-frequency part. The low-frequency part contains the overall temperature change of the image, while the high-frequency part contains the local details of the image and the small fluctuations of the temperature change. The high-frequency part usually corresponds to the hot spots, edges or mutation areas in the image, which is crucial for locating and identifying abnormal areas.

[0038] In one possible implementation, the wavelet transform decomposes the image into multiple scales, thereby capturing features at different scales. The low-frequency portion represents the rough structure of the image, usually areas with relatively uniform temperatures, while the high-frequency portion reflects areas with drastic temperature changes, usually corresponding to areas where problems or failures occur in electrical equipment.

[0039] The specific implementation process includes: The low-frequency part of the wavelet transform generally corresponds to the smooth area of ​​the image. In infrared thermal images, the low-frequency part usually reflects the overall temperature distribution. The temperature changes in these areas are relatively slow, which usually indicates that there is no serious fault in the power equipment. The extraction of the low-frequency part can help the system better understand the overall temperature trend and provide a reference for subsequent abnormal area detection.

[0040] The high frequency part reflects the local details in the image, especially the sudden change area, such as equipment failure, hot spot or current overload. The hot spot area usually shows a sudden temperature rise. By extracting the high frequency part through wavelet transform, the characteristics of these abnormal areas can be effectively highlighted. The high frequency part is usually used to detect those areas with drastic temperature changes and local abnormalities.

[0041] In order to further improve the accuracy of abnormal area identification, multi-scale wavelet transform technology is used to extract features at multiple scales at the same time. Multi-scale transformation can ensure that no matter how big the abnormal area is, the system can effectively capture the corresponding details. Wavelet transform is not limited to the decomposition of one layer, but gradually extracts different levels of detail information through multiple scale conversions.

[0042] In another embodiment, the image is gradually decomposed into smaller and smaller parts by multi-layer wavelet transform to capture detailed features. After each decomposition, the system retains the corresponding features for subsequent processing according to the different scales of temperature change to avoid losing key details.

[0043] Through this process, the low-frequency part and the high-frequency part represent different physical phenomena. The low-frequency part mainly helps to identify the normal state of the equipment, while the high-frequency part helps to identify areas where there may be faults. For infrared thermal imaging data, the low-frequency part reflects the average trend of temperature distribution and can reveal the overall condition, while the high-frequency part reflects fault signs such as abnormal temperature and local overheating.

[0044] In this embodiment, the extracted multi-scale features will be used as input to the subsequent feature fusion and nonlinear dynamic modeling steps. These features provide a solid foundation for the detection of abnormal areas. Through subsequent feature fusion and modeling, the recognition accuracy of abnormal areas can be further improved. Multi-scale wavelet transform provides the system with rich feature information, ensuring that the system can effectively deal with various temperature anomalies in complex environments.

[0045] Through multi-scale wavelet transform, the present invention can extract key features in infrared thermal images from different scales, including the overall temperature changes in the low-frequency part and the local details in the high-frequency part. This step effectively enhances the system's ability to detect abnormal areas in complex environments, improves recognition accuracy, and ensures the ability to capture details at different scales.

[0046] S3, Dimensionality reduction of high-dimensional features: The extracted multi-scale features are embedded into high-dimensional data through the manifold learning method to obtain a feature set in a low-dimensional space; Dimensionality reduction can not only reduce the difficulty of data processing, but also remove redundant information, retain the most representative features, and further improve the efficiency and accuracy of abnormal area identification.

[0047] In this embodiment, when performing dimensionality reduction of high-dimensional features, we use manifold learning technology, especially principal component analysis (PCA) and t-SNE methods to reduce the dimensionality of data. These methods can effectively map high-dimensional data to low-dimensional space while retaining important information in the original data as much as possible.

[0048] Generally speaking, high-dimensional data usually contains a lot of redundant or noisy information, and the purpose of dimensionality reduction is to map data from high-dimensional space to lower-dimensional space, so that important feature information is more concentrated and easier to process later. Dimensionality reduction can significantly reduce the complexity of data, improve computational efficiency, and help avoid overfitting problems.

[0049] Specifically, in this embodiment, we first use the principal component analysis (PCA) method to perform preliminary dimensionality reduction on the features. PCA is a linear dimensionality reduction technique that reduces the dimension of the data by projecting the data onto a new coordinate axis and selecting the most representative principal components. The basic principle of PCA is to perform eigenvalue decomposition on the data covariance matrix and select the principal component with a larger eigenvalue as a new feature, thereby retaining most of the information of the data.

[0050] The calculation process of PCA is as follows: ; in: is the original high-dimensional feature matrix; is the projection matrix consisting of eigenvectors; is the feature matrix after dimensionality reduction.

[0051] In this process, PCA retains the part with larger variance in the data to the greatest extent through the linear combination of eigenvectors, thereby retaining the most important information. The explanatory power of each principal component can be measured by the size of its corresponding eigenvalue. The principal component corresponding to the larger eigenvalue represents the more important changes in the data.

[0052] As an option, if you need to further capture the complex nonlinear relationships in the data, you can use the t-SNE (t-Distributed Stochastic Neighbor Embedding) method, which is particularly suitable for processing complex nonlinear relationships in high-dimensional data. t-SNE calculates the similarity between high-dimensional data points and uses probability distribution to embed the data into a low-dimensional space, so that similar data points in the high-dimensional space are as close as possible in the low-dimensional space.

[0053] The dimensionality reduction process of t-SNE is achieved in the following way: Calculate the conditional probability of each pair of data points to measure the similarity between them; Preserving local structures in high-dimensional data by minimizing the KL divergence between similar points in low-dimensional space.

[0054] Specifically, in this embodiment, PCA is mainly used for preliminary dimensionality reduction, reducing high-dimensional features to lower dimensions and retaining the most representative principal components. However, t-SNE can more effectively maintain local structures when further reducing dimensions, especially in complex nonlinear data. The data after dimensionality reduction reduces noise interference by retaining the most important feature information, which is helpful for subsequent modeling and abnormal area identification.

[0055] In one possible implementation, the reduced-dimensional feature data not only reduces the dimension and computational cost, but also retains the essential characteristics of the data. These reduced-dimensional data will be used as input for subsequent nonlinear dynamics modeling and graph neural network optimization, further improving the accuracy and efficiency of abnormal area identification.

[0056] Through dimensionality reduction, the originally high-dimensional and complex data is compressed into simpler and more structured low-dimensional data, removing redundant information. This process significantly improves the efficiency of subsequent algorithms (such as nonlinear modeling and graph neural network optimization), while also avoiding the dimensionality disaster problem that may occur in high-dimensional space, allowing the entire system to process efficiently at a lower computing cost.

[0057] The feature set after dimensionality reduction will enter the nonlinear dynamic modeling step. In this step, the dimensionality reduction data not only reduces the computational complexity, but also effectively retains the important information in the image, so that the subsequent nonlinear modeling based on the Lorenz system can more accurately and efficiently capture the dynamic evolution of the abnormal area.

[0058] In general, through dimensionality reduction of high-dimensional features, the present invention can significantly improve data processing efficiency while maintaining key image information, providing high-quality input data for subsequent model construction and abnormal area detection.

[0059] By reducing the dimensionality of high-dimensional feature data, the data dimension and computational burden are reduced while retaining the most important information features.

[0060] S4, nonlinear dynamics modeling and feature fusion: model the spatiotemporal evolution of the abnormal area based on the nonlinear dynamics model and fuse it with the embedded feature set; The ability to model the spatiotemporal evolution of abnormal areas is further enhanced through nonlinear dynamic modeling and feature fusion, and the features from different modules are effectively combined to improve the accuracy of subsequent abnormal area identification.

[0061] In this embodiment, we use the Lorenz system as a nonlinear dynamic model to simulate the spatiotemporal evolution of abnormal areas in infrared thermal images. Since the temperature changes in abnormal areas are often affected by multiple factors, the Lorenz system can effectively model these changes and fuse them with the feature set after dimensionality reduction. In this way, the system can not only identify the static features in the image, but also capture the dynamic features of the abnormal area that change over time.

[0062] Generally speaking, when processing infrared thermal images, the changes in abnormal areas are usually nonlinear and dynamic, and may be affected by multiple factors such as temperature, humidity, and current load. These factors may cause a sharp increase or unstable changes in local temperature, and traditional static feature analysis methods often cannot effectively capture these changes. By introducing the Lorenz system, we can better simulate these complex nonlinear processes, especially in the context of multi-dimensional data and dynamic changes, with greater adaptability and accuracy.

[0063] Specifically, the Lorenz system consists of three equations and is used to describe a three-dimensional nonlinear system. Its basic form is as follows: ; ; ; in: , , Respectively represent the three state variables of the Lorenz system. In the present invention, represents the temperature change, represents the speed at which the hot spot expands, It indicates other relevant temperature factors or equipment status; , , are the parameters of the Lorenz system and are usually set as the initial conditions of the system. By adjusting these parameters, different temperature change patterns and device states can be simulated.

[0064] As an option, in this system, we use a numerical solution method to solve the above differential equations and train and optimize the system based on historical data. , and Parameters can better adapt to the changes in abnormal areas under different environments, thereby improving the fitting accuracy of the model.

[0065] In one possible implementation, by fusing with the feature set after dimensionality reduction, the Lorenz model can provide time series data for each abnormal area, which reflects the dynamic characteristics of the area over time. These dynamic features combined with static image features (such as temperature distribution) can provide more comprehensive information for subsequent abnormal area identification.

[0066] Fusion of static and dynamic features: Through weighted fusion, the static features extracted by wavelet transform (such as low-frequency temperature distribution) are combined with the dynamic features obtained by Lorenz system modeling. The weight is set according to the contribution of each feature to the identification of abnormal areas. Dynamic features usually occupy a larger proportion in abnormal areas with more obvious time changes, while static features are more used to describe the overall temperature trend.

[0067] Fusion feature set: Through the fusion feature set, the system can simultaneously consider the temperature change and time evolution process in the image. This fusion not only improves the accuracy of recognition, but also enhances the ability to identify abnormal areas in complex scenes.

[0068] Specifically, the fused feature set can accurately describe two important dimensions of abnormal areas in the image: spatial features (temperature distribution) and temporal features (dynamic changes in abnormal areas). This combination greatly enhances the adaptability of the system in a variety of environments and can effectively identify potential fault areas even when temperature changes are not obvious or device status changes are slow.

[0069] In another implementation, in order to further improve the robustness of the system, the fusion process can also introduce more dynamic modeling methods, such as combining time series analysis (such as ARIMA model) to further optimize the prediction of abnormal areas. By combining multiple models, more accurate results can be obtained under different conditions.

[0070] In this embodiment, nonlinear dynamic modeling and feature fusion provide highly accurate dynamic features, which provide key input for subsequent graph neural network optimization. Through this process, the system can not only extract important features from static images, but also predict the evolution trend of abnormal areas based on historical data and dynamic changes.

[0071] By using the Lorenz system for nonlinear dynamic modeling, the dynamic changes of abnormal areas can be accurately captured, and these dynamic features can be effectively integrated with static features. This step significantly improves the system's ability to identify abnormal areas in complex environments, ensuring that abnormal area detection under different working conditions is highly accurate and robust. Through feature fusion, the system can simultaneously consider the evolution of temperature distribution and temperature change, providing comprehensive and accurate feature data for subsequent abnormal area identification.

[0072] S5. Graph neural network optimization features: Use graph neural network to optimize the fused features, update the relationship between nodes, and output the optimized features; This step uses the graph structure information of the graph neural network to optimize the relationship between feature nodes, further improving the quality and recognizability of feature representation.

[0073] In this embodiment, the main task of the graph neural network optimization is to update the features of each node in the graph through graph convolution operations, so that the features can better reflect the spatial structure and local dependencies of the abnormal area. Specifically, the graph neural network regards each feature in the infrared thermal image as a node, and the relationship between different feature nodes is represented by the edges of the graph. Through the graph convolution operation, the information between nodes can be transmitted, and the node features are continuously updated, so that the network can perform feature optimization on the graph structure.

[0074] In general, graph neural networks can effectively process graph structured data and optimize the relationship between nodes through local information transfer. For infrared thermal images, the advantage of graph neural networks is that they can capture the spatial and temporal dependencies between different temperature changes, especially the local features of abnormal areas in the image. Graph neural networks propagate information between nodes through convolution operations, so that the characteristics of each node not only depend on its own attributes, but also on the attributes of neighboring nodes.

[0075] Specifically, the basic structure of a graph neural network includes an adjacency matrix and a node feature matrix of a graph. The adjacency matrix represents the connection relationship between nodes in the graph, and the node feature matrix represents the features of each node. The graph convolution operation propagates information and updates features based on the adjacency matrix and node features.

[0076] The general form of the graph convolution operation is: ; in: It is The node feature matrix of the layer; is the normalized adjacency matrix, which represents the similarity relationship between nodes; It is The weight matrix of the layer, used to map feature dimensions; Is the activation function. Common activation functions include ReLU or sigmoid. It is The node feature matrix after layer update.

[0077] In one possible implementation, the adjacency matrix It is usually standardized to avoid the influence of degree differences of nodes in the graph on feature propagation. The standardized adjacency matrix can be obtained by the following formula: ; in, is the height matrix of the nodes in the graph, is the original adjacency matrix of the graph. Through this standardization, the propagation process of node features is balanced, avoiding the unbalanced influence between nodes.

[0078] As an option, in some cases, the convolution operation of the graph neural network can also enhance the learning ability of features through multi-layer stacking. Specifically, through multi-layer graph convolution operations, the network can capture a wider range of node relationships and deep feature dependencies, thereby improving the ability to identify abnormal areas in complex scenes.

[0079] Specifically, the main task of graph neural network feature optimization is to update the information through layer-by-layer propagation so that the features of each node more accurately reflect the spatial structure of the abnormal area in the image. For example, in infrared thermal images, temperature changes are usually localized, and abnormal areas usually appear as concentrated hot spots. Graph neural networks can capture this local structure, allowing the network to optimize features for specific abnormal areas and further improve the positioning accuracy of abnormal areas.

[0080] In one possible implementation, in order to further enhance the graph neural network's ability to learn the relationship between nodes, a multi-head attention mechanism can be used, which enables the network to learn the relationship between nodes in parallel in multiple subspaces, thereby improving the network's expressiveness. The multi-head attention mechanism can help the network better handle complex dependencies, especially when processing images with a large number of nodes and complex edge relationships, which can improve the performance of graph neural networks.

[0081] Through the optimization of graph neural networks, the features in the image can be recombined according to the similarities between nodes, so that each feature node can more accurately reflect the spatial distribution and local characteristics of the abnormal area. This process not only improves the accuracy of feature representation, but also enhances the robustness of the system, so that the system can still maintain a high recognition accuracy in complex environments.

[0082] The feature optimization of graph neural networks provides accurate and optimized feature inputs for the subsequent abnormal region identification step. These optimized features can more accurately reflect the abnormal regions in the image and provide high-quality data support for subsequent classification and positioning.

[0083] S6. Abnormal area identification: using the optimized features to identify abnormal areas and obtain the location of the abnormal areas; Abnormal area recognition uses these features to accurately locate possible abnormal areas in the image, such as hot spots with excessive temperatures or equipment failures. This step is the core of the entire system, determining whether the system can efficiently and accurately identify areas that require attention and provide a basis for subsequent power equipment maintenance.

[0084] In this embodiment, the abnormal area recognition uses classic classification algorithms such as support vector machine (SVM) or decision tree. By using these machine learning algorithms, the system can learn the characteristic patterns of the abnormal area based on the optimized feature data, and accurately identify the location of the abnormal area and its related information in the new image. Through this step, the system can filter out the areas that actually need to be paid attention to from the complex infrared thermal image, avoiding the occurrence of false positives and false negatives.

[0085] In general, the identification of abnormal areas depends on the differences between features. There are usually some areas with higher or lower temperatures in the image, which may be related to problems such as overload, local failure, hot spots, etc. of power equipment. Classification algorithms such as support vector machines (SVM) and decision trees can accurately distinguish normal areas from abnormal areas by learning the features of abnormal areas in historical annotated data.

[0086] Specifically, in this embodiment, a classification model is first established by a support vector machine (SVM) or a decision tree. The input of the model is the feature data extracted and optimized in the previous steps, including low-frequency and high-frequency features after wavelet transformation, dynamic features in the nonlinear dynamics model, and features optimized by graph neural network. These features are used to describe information such as temperature distribution and change trend in different areas of the image.

[0087] The goal of a support vector machine (SVM) is to classify data into two categories by finding a hyperplane that maximizes the margin between categories. During training, SVM solves the following optimization problem: ; The constraints are: ; in: is the normal vector of the hyperplane; is the bias of the hyperplane; It is The feature vector of sample points; It is The labels of the samples; It is a slack variable that allows some sample points to violate the classification boundary to improve the robustness of the model; is a regularization parameter used to control the complexity of the model.

[0088] As an option, the decision tree algorithm can also be used for abnormal area identification. The decision tree forms a tree structure by learning the relationship between the feature values ​​and the target values ​​in the training data, and finally determines which category the sample belongs to through the leaf nodes. During the training process, the decision tree selects the optimal split feature through information gain or Gini index, and gradually builds the tree structure. The learning process of the decision tree can be expressed by the following formula: ; in: It is a dataset The information entropy indicates the purity of the data; It is a dataset By Features The subset obtained after segmentation; and Represents the data set and subset size; Is a subset Information entropy.

[0089] Specifically, the training process of support vector machines and decision trees requires a large amount of labeled data in order to learn the boundaries between normal and abnormal areas. Through these machine learning methods, the system can extract the features of abnormal areas from the training data, generate a classification model, and then classify new images in actual applications to accurately identify abnormal areas.

[0090] In one possible implementation, in order to improve the accuracy of abnormal area identification, the identification process can also be combined with a multi-classification algorithm, for example, based on the traditional two-classification model, it can be expanded into a multi-classification problem to handle different types of abnormal areas. Through the multi-classification model, the system can not only identify common abnormal areas, but also distinguish different types of abnormalities (such as overload, fault, hot spot, etc.), further improving the accuracy and adaptability of the system.

[0091] Generally, the identification result of the abnormal area is not just a location, but can also include the temperature information, fault type and fault severity of the area. This information will help maintenance personnel to more accurately judge the equipment status and take timely and effective maintenance measures.

[0092] As an option, in addition to using SVM and decision trees, other advanced deep learning algorithms (such as convolutional neural networks (CNNs)) can also be applied to this task, especially in more complex scenarios. By combining deep learning models, the system is able to automatically extract richer representations at the feature level, further enhancing the ability to identify abnormal areas.

[0093] After the abnormal area is identified in step S6, the system passes the identified abnormal area and its location information to the subsequent abnormal area positioning and optimization step. This step will further optimize the detection results of the abnormal area and provide more accurate information for subsequent maintenance decisions.

[0094] S7. Strategy optimization to improve recognition accuracy: Optimize the abnormal area recognition strategy through reinforcement learning and global optimization algorithms to improve the accuracy of recognition results; Through strategy optimization, combined with reinforcement learning and global optimization algorithms, the existing recognition strategy is optimized so that the system can still maintain high accuracy and robustness in a changing environment.

[0095] In this embodiment, we use reinforcement learning and global optimization algorithms (such as genetic algorithms or simulated annealing algorithms) to optimize the strategy for abnormal area identification. Through reinforcement learning, the system can adjust the strategy based on feedback in the process of continuous trial and error, and further improve the search range and stability of the strategy through the global optimization algorithm, thereby improving the accuracy of abnormal area identification.

[0096] Generally speaking, reinforcement learning is a machine learning method that optimizes strategies by interacting with the environment. The system adjusts its behavior by continuously performing actions and receiving reward signals from the environment in order to achieve the maximum long-term return. In the present invention, the main goal of reinforcement learning is to optimize the recognition strategy so that the system can more accurately identify and locate abnormal areas.

[0097] Specifically, we define the state space, action space, and reward function of the system: State space: The state space represents the environmental state of the system at each moment. In the present invention, the state space can be the feature information of each pixel in the image, or the overall features of the image block. Each state represents the current abnormal area distribution and other related information of the image.

[0098] Action space: The action space represents the different recognition strategies that the system can adopt. Each action corresponds to the adjustment of the abnormal area recognition strategy, such as selecting different thresholds, adjusting feature weights, etc.

[0099] Reward function: The reward function is used to evaluate whether the action selected by the system helps improve recognition accuracy. Specifically, if the recognition result is correct, the system will receive a positive reward; if the recognition is wrong, it will receive a negative reward. In this way, the system gradually learns strategies that can improve recognition accuracy.

[0100] In one possible implementation, reinforcement learning methods can use Q-learning or deep Q-network (DQN) for policy optimization. Q-learning is a value function-based method that selects the optimal action by learning the Q value of each state-action pair. The update formula of Q value is as follows: ; in: is in state Take action Q value; It is the system that is taking action Rewards received after is a discount factor used to control the impact of future rewards; is the learning rate, which is used to control the speed of Q value update; is in state The maximum Q value for taking the best action.

[0101] Through this update formula, the Q value gradually converges, and eventually the system can learn the optimal strategy, thereby improving the recognition accuracy of abnormal areas.

[0102] As an option, in addition to Q-learning, we can also use deep Q network (DQN). DQN can handle more complex state spaces and high-dimensional data by using deep neural networks to approximate the Q value function. Through neural network training, the system can be optimized in a large-scale state space, thereby effectively improving the accuracy of abnormal area recognition.

[0103] In a possible implementation, in order to further improve the effect of strategy optimization, the present invention also introduces a global optimization algorithm, such as a genetic algorithm or a simulated annealing algorithm. These global optimization algorithms optimize the parameters of the strategy through global search, avoiding the problem of local optimal solutions, thereby ensuring the global optimality of the identification strategy.

[0104] Genetic algorithm: Genetic algorithm is an optimization algorithm that simulates the natural selection process. It generates new strategy combinations through operations such as selection, crossover, and mutation, and evaluates their advantages and disadvantages based on fitness, gradually evolving the optimal solution. The main steps of genetic algorithm include: Selection: Select excellent strategies for crossover operation based on the fitness value of each strategy; Crossover: Generate a new strategy by combining two strategies; Mutation: Randomly mutate certain strategies to increase diversity; Fitness evaluation: Calculate the fitness value of each strategy and evaluate its contribution to the accuracy of abnormal area recognition.

[0105] Simulated annealing algorithm: The simulated annealing algorithm is a global optimization algorithm based on probabilistic search, which can jump out of the local optimal solution during the search process to find the global optimal solution. Its basic principle is to simulate the physical annealing process, gradually reduce the search range by controlling the temperature, and finally converge to the optimal solution.

[0106] Specifically, by combining reinforcement learning and global optimization algorithms, the present invention can continuously adjust the abnormal area identification strategy to ensure that high recognition accuracy can be maintained in different scenarios. This optimization process can not only improve the accuracy of the model, but also enhance the robustness of the system, so that it can still effectively identify abnormal areas in the face of complex and dynamically changing environments.

[0107] After strategy optimization, the system will obtain a more accurate and robust abnormal area identification strategy. This optimized strategy will provide guidance for subsequent abnormal area positioning and response, ensuring that abnormal areas can be accurately located and responded to in a timely manner under various environmental conditions.

[0108] The power transmission line abnormal area recognition system using infrared thermal image feature fusion described below and the power transmission line abnormal area recognition method using infrared thermal image feature fusion described above can be referred to each other.

[0109] Please see attached Figure 2 The present invention also provides a transmission line abnormal area recognition system based on infrared thermal image feature fusion, comprising: Data acquisition module, used to collect infrared thermal imaging data of power transmission lines; A preprocessing module, used for preprocessing the collected infrared thermal imaging data; Feature extraction module, used to perform wavelet transform, multi-scale analysis, data embedding and nonlinear dynamic modeling on the pre-processed infrared thermal imaging data to extract multi-dimensional features; Graph neural network optimization module, used to optimize the graph structure of the extracted features; An abnormal area recognition module is used to locate and identify abnormal areas based on optimized features; Reinforcement learning and global optimization modules are used to optimize abnormal area identification strategies and improve identification accuracy.

[0110] The system of this embodiment can be used to execute the above method embodiments, and its principles and technical effects are similar, which will not be repeated here.

[0111] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying abnormal areas of power transmission lines based on infrared thermal image feature fusion, characterized in that: The following steps are involved: Collect infrared thermal imaging data of the transmission line and pre-process the data; The pre-processed infrared thermal image is decomposed into multiple scales by wavelet transform to extract the low-frequency part and the high-frequency part; The extracted multi-scale features are embedded into high-dimensional data through the manifold learning method to obtain a feature set in a low-dimensional space; The spatiotemporal evolution of the abnormal area is modeled based on a nonlinear dynamic model and fused with the embedded feature set; Use graph neural networks to optimize the fused features, update the relationship between nodes, and output the optimized features; Use the optimized features to identify abnormal areas and obtain the locations of abnormal areas; The abnormal area identification strategy is optimized through reinforcement learning and global optimization algorithms to improve the accuracy of the identification results.

2. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1 is characterized in that: The pre-processing comprises: Remove noise from the collected infrared thermal imaging data; Perform contrast enhancement processing on infrared thermal images; Perform histogram equalization on the processed image.

3. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1 is characterized in that: The wavelet transform adopts discrete wavelet transform, the low-frequency part includes the overall temperature distribution characteristics, the high-frequency part includes the detail characteristics, and the extraction process of the low-frequency part and the high-frequency part includes: Perform wavelet transform on the image data to obtain low-frequency and high-frequency features at different scales; The low-frequency and high-frequency parts are denoised, and the parts with significant features are retained.

4. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1 is characterized in that: The high-dimensional data embedding is performed using a principal component analysis method or a t-SNE method, and the feature set after the data embedding includes: Perform dimensionality reduction on high-dimensional data to extract the most representative features; The reduced feature set is used as input for subsequent processing.

5. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1, characterized in that: The nonlinear dynamic modeling adopts the Lorenz system to describe the dynamic evolution process of the abnormal area. The dynamic evolution process includes: Based on the nonlinear equation of the Lorenz system, the dynamic evolution of abnormal areas in infrared thermal images over time is simulated. By using the three-dimensional state variables of the Lorenz equation, the temperature change, expansion and shrinkage of the abnormal area are described. In the Lorenz system, by selecting appropriate parameters and adjusting according to the actual infrared thermal imaging data, the model can better match the actual temperature change pattern; The dynamic evolution process obtained by Lorenz system modeling is fused with the multi-scale features extracted from infrared thermal images through a weighted fusion method.

6. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1, characterized in that: The graph neural network optimization includes: Each image feature node is used as a node of the graph, and the similarity between nodes is represented by the edges of the graph; Normalize the adjacency matrix of the graph, use graph neural network to propagate features, and update node features; The node features are integrated using graph convolution operations to obtain the optimized feature set as the final output.

7. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1, characterized in that: The abnormal area recognition adopts support vector machine or decision tree method, and the recognition results include: Establish a classification model based on training data to automatically identify abnormal areas; Output the location coordinates and temperature information of the identified abnormal area.

8. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1, characterized in that: The reinforcement learning algorithm includes: Define state space, action space, and reward function; Use Q-learning or deep Q-network to optimize abnormal region identification strategy; The strategy is adjusted according to the feedback reward function to maximize the recognition accuracy.

9. The method for identifying abnormal areas of power transmission lines by fusion of infrared thermal image features according to claim 1, characterized in that: The global optimization uses a genetic algorithm or a simulated annealing algorithm to globally optimize the identification strategy, and the optimization process includes: Searching for parameters of the identification strategy using genetic algorithms or simulated annealing algorithms; The stability and accuracy of the strategy are optimized through global search.

10. The abnormal area identification system of power transmission lines based on infrared thermal image feature fusion is characterized by: The method for identifying abnormal areas of power transmission lines using the infrared thermal image feature fusion described in any one of claims 1 to 9 comprises: Data acquisition module, used to collect infrared thermal imaging data of power transmission lines; A preprocessing module, used for preprocessing the collected infrared thermal imaging data; Feature extraction module, used to perform wavelet transform, multi-scale analysis, data embedding and nonlinear dynamic modeling on the pre-processed infrared thermal imaging data to extract multi-dimensional features; Graph neural network optimization module, used to optimize the graph structure of the extracted features; An abnormal area recognition module is used to locate and identify abnormal areas based on optimized features; Reinforcement learning and global optimization modules are used to optimize abnormal area identification strategies and improve identification accuracy.

Images