Power transmission line infrared temperature measurement method combined with dynamic threshold

Abstract

The invention relates to the technical field of power transmission line monitoring, and discloses a power transmission line infrared temperature measurement method combined with a dynamic threshold, and the method comprises the following steps: S1, collecting an infrared image, environmental parameters and geographical position information of a power transmission line through an unmanned plane carrying an infrared imaging device and a GPS positioning device, the GPS positioning device is used for recording geographic position information of the power transmission line in real time, and collected data are transmitted to the ground monitoring system through wireless communication; s2, performing denoising and standardization processing on the infrared image acquired in the step S1, and extracting temperature texture features of the infrared image; through an infrared temperature measurement method combined with a dynamic threshold value, a deep learning model and regional characteristic calculation are introduced, and accurate fault identification in multiple environments is realized. The Gaussian filtering and the convolutional neural network are adopted to improve the accuracy of temperature feature extraction and classification, and the judgment reference is dynamically adjusted through a dynamic threshold formula to reduce the false alarm rate and the missing report rate.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission line monitoring, and particularly to an infrared temperature measurement method for transmission lines combined with dynamic thresholds. Background Art

[0002] As a core part of the power system, the monitoring of the operating state of transmission lines has always been an important research direction in the power industry. Temperature anomalies are usually the early manifestations of transmission line faults, so infrared temperature measurement technology is widely used in line monitoring. However, existing technical solutions face multiple challenges in practical applications, which limit their monitoring effects and the reliability of fault determination.

[0003] The core of infrared temperature measurement technology lies in extracting line temperature information from infrared images and combining set temperature thresholds for fault determination. However, existing technologies mostly adopt fixed threshold determination methods. This method usually sets temperature thresholds based on historical experience data or experimental results in standardized environments, and it is difficult to adapt to the complex and changeable environmental conditions of transmission lines. For example, in high-temperature areas, the relatively high ambient temperature may cause the surface temperature of a normally operating line to approach the fixed threshold, thus triggering false alarms. In low-temperature or high-wind-speed areas, the fixed threshold may ignore the heat dissipation effect, resulting in missed detection of actual anomalies. This single static determination method has obvious limitations in application scenarios with significant geographical environment differences. Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides an infrared temperature measurement method for transmission lines combined with dynamic thresholds, which solves the problems of poor adaptability of fixed thresholds, low accuracy in identifying temperature anomalies, and lack of dynamic adjustment in the alarm mechanism in existing infrared temperature measurement methods for transmission lines.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: An infrared temperature measurement method for transmission lines combined with dynamic thresholds, comprising the following steps: S1. Use a drone equipped with an infrared imaging device and a GPS positioning device to collect infrared images, environmental parameters, and geographical location information of the transmission line. Among them, the GPS positioning device is used to record the geographical location information of the transmission line in real time, and the collected data is transmitted to the ground monitoring system through wireless communication; S2. Denoise and standardize the infrared images collected in step S1, and extract their temperature texture features; S3. Use a deep learning model to analyze the infrared images processed in step S2, and output temperature anomaly scores and fault classification results; S4. Generate region-adaptive dynamic threshold values according to the geographical location information and environmental parameters collected in step S1, as well as historical data; S5. Compare the temperature anomaly score obtained in step S3 with the dynamic threshold value generated in step S4 to determine the operating state of the transmission line, and trigger an alarm according to the degree of anomaly.

[0006] Preferably, the environmental parameters in step S1 include environmental temperature, humidity, wind speed, and light intensity. In step S1, the unmanned aerial vehicle flies along a preset inspection path that covers the key nodes and high-risk areas of the transmission line.

[0007] Preferably, when denoising the collected infrared images in step S2, the Gaussian filtering algorithm is used to smooth the pixel values.

[0008] Preferably, the deep learning model in step S3 is a convolutional neural network, and the convolutional neural network includes the following parts: Convolutional layer, which is used to extract the temperature texture features of the infrared images processed in step S2; Pooling layer, which is used to reduce the dimension of the extracted temperature texture features; Fully connected layer, which is used to classify the features after dimensionality reduction and generate feature vectors; Output layer, which is used to output the temperature anomaly score and the fault classification result.

[0009] Preferably, the formula for generating the dynamic threshold value in step S4 is: ; Where, is the dynamic threshold value, is the reference threshold value, is the adjustment factor, which is calculated based on the environmental parameters and regional characteristics collected in step S1, and the includes environmental temperature, humidity, wind speed, light intensity, and the geographical location of the line.

[0010] Preferably, the calculation formula of the adjustment factor is as follows: ; Where, is the environmental temperature, is the humidity, is the wind speed, is the light intensity, , , , are the weight coefficients obtained by training based on historical data.

[0011] Preferably, after comparing the temperature anomaly score obtained in step S3 with the dynamic threshold value in step S4 in step S5, the determination results include the following: If the temperature anomaly score is less than or equal to the dynamic threshold value, it is determined to be in a normal state; If the temperature anomaly score is greater than the dynamic threshold value but less than the sum of the dynamic threshold value and the preset temperature threshold difference, it is determined to be in a slightly abnormal state; If the temperature anomaly score is greater than the sum of the dynamic threshold value and the preset temperature threshold difference, it is determined to be in a serious fault state and an alarm is triggered.

[0012] Preferably, the alarm in step S5 includes the geographical location of the faulty line, the infrared image of the abnormal area, environmental parameters, and the fault classification result.

[0013] The present invention provides an infrared temperature measurement method for transmission lines combined with a dynamic threshold. It has the following beneficial effects: 1. The present invention adopts an infrared temperature measurement method combined with a dynamic threshold value. By collecting geographical location information, environmental parameters, and infrared images in real time, and introducing a deep learning model and regional characteristics to calculate the dynamic threshold value, it achieves the technical effect of accurately identifying line temperature anomalies and dynamically adapting to different environments. Compared with the existing technologies that use fixed temperature thresholds or single analysis methods, it solves the problems of high fault misjudgment rate and poor adaptability caused by environmental differences.

[0014] 2. The present invention performs denoising processing on the infrared image through Gaussian filtering and normalization, and combines a convolutional neural network to extract and classify temperature texture features at multiple levels, achieving the technical effects of improving the accuracy of temperature anomaly scores and the reliability of fault classification. In the existing technologies, regular models or simple threshold judgments are often used, which are difficult to capture complex temperature change laws. The present invention makes up for the deficiencies of these methods in feature recognition ability.

[0015] 3. The present invention provides a flexible fault determination benchmark through the real-time generation formula of the dynamic threshold value, combined with historical data, real-time environmental parameters, and line characteristics, achieving the technical effect of dynamically adjusting the alarm sensitivity. Compared with the fixed strategies in the existing technologies that cannot adjust the threshold value according to environmental changes, the present invention effectively reduces the false alarm rate and missed alarm rate, and is particularly suitable for the monitoring requirements of transmission lines in multi-region and multi-climate conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0018] Please refer to the attached Figure 1 , the embodiment of the present invention provides an infrared temperature measurement method for transmission lines combined with a dynamic threshold, including the following steps: S1. Use a drone equipped with an infrared imaging device and a GPS positioning device to collect infrared images, environmental parameters, and geographical location information of the transmission line. Among them, the GPS positioning device is used to record the geographical location information of the transmission line in real time, and the collected data is transmitted to the ground monitoring system through wireless communication; First of all, step S1 is the basic data collection link to implement the present invention, which is used to provide multi-source data support for the analysis of the operating state of the transmission line, including infrared images, environmental parameters, and geographical location information. To ensure the accuracy and reliability of subsequent data processing and model analysis, the collected data should cover the key nodes and high-risk areas of the transmission line and ensure real-time and continuity.

[0019] Generally, as the main tool for data collection, the drone has the characteristics of high flexibility, wide inspection range, and strong real-time feedback ability. To meet the precise data collection requirements, the drone is equipped with an infrared imaging device, an environmental sensor, and a GPS positioning device. Among them, the GPS positioning device can provide accurate geographical location information to ensure the accurate spatial distribution of each data point.

[0020] In this embodiment, the drone flies along a preset inspection path, and the path covers the key nodes, high-risk areas, and other line parts to be monitored of the transmission line. The drone collects the following data in real time: Infrared images of the transmission line, recording the temperature distribution of the line; Environmental parameters, including environmental temperature, humidity, wind speed, and light intensity; Geographical location information, used to calibrate the position coordinates of each data point.

[0021] To achieve real-time and reliable data transmission, the collected data is transmitted to the ground monitoring system through the wireless communication module of the drone for subsequent processing and analysis.

[0022] Specifically, the infrared imaging device is used to capture infrared images of the transmission line and generate heat map data of the temperature distribution. These image data are represented in the form of pixel values, and each pixel value corresponds to the surface temperature of a certain point on the line. To ensure the quality of image collection, the infrared imaging device has high resolution and can accurately measure the temperature gradient change over a long distance. In addition, the infrared imaging device supports multi-angle imaging modes and can cover different perspective line areas through the attitude adjustment of the drone.

[0023] As an option, the collection of environmental parameters is completed by a multi-functional sensor mounted on the drone.

[0024] Ambient temperature It is collected by a digital temperature sensor, and the unit of measurement is degrees Celsius (°C).

[0025] Humidity It is obtained by a capacitive humidity sensor, and the unit of measurement is percentage (%).

[0026] Wind speed It is measured by an ultrasonic anemometer, and the unit is meters per second (m / s).

[0027] Illuminance It is measured by a photoelectric sensor, and the unit is watts per square meter (W / m 2 )

[0028] In a possible implementation, to ensure complete path coverage, the inspection path is preset through historical fault records and line operation characteristic analysis, and is dynamically adjusted according to real-time environmental conditions (such as wind speed and obstacle distribution). The inspection path adopts a grid distribution design to ensure that the inspection data of all key nodes and high-risk areas can be completely collected.

[0029] For the collection of geographical location information, the sampling frequency of the GPS positioning device is to record multiple times per second, and the positioning accuracy can reach the centimeter level, outputting longitude and latitude coordinates and altitude information. These data provide time and space tags for each frame of infrared image and environmental parameters, facilitating the generation of dynamic threshold values and fault location in subsequent steps.

[0030] In some embodiments, the quality of data collection is further optimized in the following ways: In complex terrains (such as mountains and high-voltage corridors), a high-precision inertial navigation system (INS) is combined with GPS to ensure the stability of the UAV flight and the accuracy of data collection.

[0031] In a strong wind environment, the UAV corrects its flight attitude in real time through an attitude control algorithm to reduce the impact of vibration on image clarity.

[0032] Under low light conditions, the photosensitive intensity of the infrared imaging device will be automatically increased to ensure that the collected infrared image data still has a high signal-to-noise ratio.

[0033] Specifically, the collected data needs to meet the following constraints: The ambient temperature range should cover the typical operating conditions of the transmission line, such as -40°C to 85°C; The humidity range should support 0% to 100%; The upper limit of wind speed measurement needs to reach 30 m / s to meet the inspection requirements in extreme weather; The illuminance range can reach 1000 W / m2 , to cover strong sunlight conditions.

[0034] In a possible implementation, the redundancy design during data acquisition is also crucial. For example, the drone will perform multiple repeated acquisitions during the flight path to avoid data loss or error accumulation. The data from these repeated acquisitions is processed through a fusion algorithm in the ground monitoring system to obtain higher acquisition accuracy.

[0035] Extending to the preliminary processing of data, the edge computing module of the drone can perform preliminary analysis on the infrared image data during flight, including denoising preprocessing and image quality assessment. These processes can filter out low-quality or incomplete data, thereby reducing the computational pressure on subsequent data analysis.

[0036] The above data acquisition method provides multi-modal data of infrared images, environmental parameters, and geographical information, laying a solid foundation for denoising processing, dynamic threshold generation, and fault analysis in subsequent steps.

[0037] S2. Denoise and standardize the infrared image collected in step S1, and extract its temperature texture features; Step S2 is to denoise and standardize the obtained infrared image after the data acquisition in step S1, providing high-quality and standardized data input for subsequent deep learning analysis. Infrared images usually contain a large amount of environmental noise and equipment errors, and these interferences will significantly affect the extraction and analysis of temperature texture features. Therefore, denoising and standardization are key steps to ensure data reliability and consistency.

[0038] Generally, the noise of infrared images mainly comes from the external environment (such as uneven light, wind speed changes) and the inherent errors of infrared imaging devices. To cope with these interferences, a series of denoising techniques are used to preprocess the images. At the same time, the standardization process adjusts the pixel values of the images through a normalization algorithm to ensure that the range and distribution of the input data meet the requirements of the deep learning model.

[0039] In this embodiment, the collected infrared image is first denoised. The denoising uses the Gaussian filtering algorithm, and its core idea is to perform weighted averaging on the pixels in the image through a Gaussian kernel, thereby reducing the interference of random noise on the temperature texture features. The specific formula of Gaussian filtering is as follows: ; Where represents the pixel value after filtering, x and y are the coordinates of the current pixel point, and σ is the standard deviation of the filter, which is used to control the smoothness of the filtering.

[0040] In a possible implementation, the standard deviation σDynamically adjust according to the resolution and noise level of the image. For example, for infrared images with higher resolution and more noise, a larger σ value is taken to improve the denoising ability; while for images with less noise, a smaller σ value is selected to retain more temperature detail features. In some embodiments, the size of the filtering window is set to 3×33 or 5×55 to ensure a balance between the filtering effect and the computational efficiency.

[0041] After denoising, the image is normalized. The purpose of normalization is to adjust the image pixel values to the form of zero mean and unit standard deviation, thereby reducing the influence of the imaging device and environmental conditions on the data distribution. Specifically, the normalization adopts the following formula: ; where, is the normalized pixel value, is the original pixel value, is the mean of the pixel values, is the standard deviation of the pixel values. The image processed by the above formula has a higher uniformity in pixel value distribution, which is beneficial to the feature extraction and training process of the deep learning model.

[0042] As an option, the mean and the standard deviation in the normalization process can be obtained by calculating the entire image, or can be dynamically adjusted according to the pixel values of the local area. In some embodiments, in order to improve the adaptability to large temperature difference scenarios, the mean and standard deviation are calculated separately for the central region and the edge region of each frame of the image, and then the normalization process is performed separately for each region. This local normalization method can enhance the overall balance of the image while maintaining the temperature characteristics.

[0043] Specifically, the combined effect of denoising and normalization is particularly significant in the following scenarios: When the transmission line is located in mountainous areas or complex terrains, the wind speed and light intensity fluctuate greatly, and there are often a large number of random noises in the infrared image. Through the Gaussian filtering algorithm, these random noises can be effectively removed.

[0044] When the transmission line crosses multiple temperature zones (such as high-temperature environment and low-temperature environment), the temperature values in different regions vary greatly. The normalization process can adjust the overall distribution of the image, making it easier for the deep learning model to learn the key features of the temperature texture.

[0045] In a possible implementation, the image preprocessing module operates in real time. The infrared image is immediately transmitted to the processing module for denoising and normalization after being collected, so as to ensure that the processed data can be quickly input into the subsequent analysis steps.

[0046] In some embodiments, to improve the robustness of data processing, the following auxiliary processing is also added: Perform edge enhancement on the filtered image to retain the detailed changes in the temperature gradient, facilitating the identification of hot spots by the deep learning model; Perform dynamic contrast enhancement on the standardized image to improve the clarity of the boundary between the low-temperature and high-temperature regions; In some implementation environments, multi-thread processing technology is used to perform the denoising and standardization steps simultaneously to improve the processing efficiency.

[0047] Through the above method, this step can significantly improve the quality and consistency of the image, providing more accurate and reliable data input for subsequent deep learning analysis.

[0048] S3. Analyze the processed infrared image in step S2 using a deep learning model, and output the temperature anomaly score and the fault classification result; Step S3 is to analyze the processed data using a deep learning model after completing the denoising and standardization processing of the infrared image in step S2, and output the temperature anomaly score and the fault classification result. The core of this step is to automatically identify the abnormal regions in the infrared image through the multi-layer feature extraction ability of the deep learning model, and quantify the analysis results into the temperature anomaly score and the specific fault classification result.

[0049] Generally, the feature extraction and analysis of infrared images require layer-by-layer dimensionality reduction and feature refinement of high-dimensional data to achieve accurate identification of abnormal regions. Due to its highly non-linear mapping ability and automatic feature learning characteristics, the deep learning model has become an ideal tool for analyzing infrared images. In the present invention, a convolutional neural network is selected as the main structure of the deep learning model, and efficient extraction and classification of image features are achieved through multi-layer convolution, pooling, and fully connected operations.

[0050] In this embodiment, the convolutional neural network includes the following modules: a convolutional layer, a pooling layer, a fully connected layer, and an output layer. First, local feature extraction is performed on the infrared image through the convolutional layer. The convolutional layer completes the weighted operation between pixels by sliding the convolutional kernel on the image, and the specific calculation formula is as follows: ; Where represents the convolutional output value, is the weight of the convolutional kernel, is the input pixel value of the infrared image, is the bias value, and m×n is the size of the convolutional kernel. The result of the convolutional operation is a feature map, which is used to represent the temperature texture features of the infrared image.

[0051] Specifically, the size of the convolution kernel can be adjusted according to the resolution of the infrared image, such as 33×3 or 55×5. In some embodiments, multiple convolution kernels are used to process the input image simultaneously to extract features of multiple different scales. These features include temperature gradient changes, the distribution of hot spots, and local outliers.

[0052] The pooling layer follows the convolution layer and is mainly used to reduce the dimensionality of the convolution features. Through pooling operations, data redundancy can be effectively reduced, the computational complexity can be lowered, while key feature information is retained. Generally, max pooling or average pooling is used to reduce the dimensionality of the feature map. Taking max pooling as an example, its formula is: ; where represents the output value after pooling, , ,…, represent the pixel values within the pooling window.

[0053] As an option, the size of the pooling window can be set to 2×22 or 33×3. For example, when processing high-resolution infrared images, using a larger pooling window can further reduce the computational requirements.

[0054] The fully connected layer is used to map the feature maps generated by convolution and pooling operations into specific high-dimensional vectors, further refining the features and performing classification. The output vector of the fully connected layer can be understood as the feature representation of the infrared image for subsequent classification operations. In a possible implementation, the fully connected layer uses the ReLU activation function to increase the nonlinear expression ability of the model, and its formula is: ; The output layer is responsible for generating the temperature anomaly score and the fault classification result. Specifically, the temperature anomaly score is output through a regression model and is used to quantify the degree of anomaly in the infrared image; the fault classification result is output through the Softmax activation function and is used to classify the image into multiple state categories, such as normal, slightly abnormal, or severely faulty.

[0055] In some embodiments, the number of categories for fault classification can be adjusted according to actual application requirements. For example, for the monitoring of a single type of transmission line, the classification results can include normal, slightly overheated, and high-temperature overload; for a complex system with multiple parallel lines, more classification categories can be introduced to cover a wider range of fault types.

[0056] In a possible implementation, to improve the generalization ability and robustness of the model, data augmentation techniques are adopted during the training process. For example, by rotating, flipping, or adding pseudo-noise to the infrared images, more diverse training samples are generated. In addition, the Xavier initialization method is used for the weight initialization of the model to ensure the gradient convergence speed during the training process.

[0057] In the model output part, to improve the interpretability of the anomaly score, the anomaly region can be visually marked according to the numerical range of the temperature anomaly score. For example, the score value is superimposed on the infrared image, and the high-score region is highlighted with enhanced color for quickly locating the potential risk areas of the transmission line.

[0058] The above method can ensure the efficient recognition and classification of temperature anomalies in infrared images. The output temperature anomaly score and fault classification result provide a basic basis for the generation of subsequent dynamic threshold values and fault alarms.

[0059] S4. Generate a region-adaptive dynamic threshold value according to the geographical location information, environmental parameters, and historical data collected in step S1; Step S4 is to generate a region-adaptive dynamic threshold value according to the geographical location information, environmental parameters, and historical temperature data of the transmission line after the temperature anomaly score and fault classification result are generated in step S3. The role of the dynamic threshold value is to provide an adaptive reference value for fault determination by combining the real-time environment and regional characteristics, thereby improving the accuracy and robustness of fault determination.

[0060] Generally, the temperature of the transmission line is affected by environmental factors, the characteristics of the line itself, and operating conditions. These influencing factors vary significantly in different geographical regions and environmental conditions. Therefore, the generation of the dynamic threshold value needs to fully consider the real-time collected environmental parameters, the characteristics of the region where the line is located, and historical temperature data.

[0061] In this embodiment, the dynamic threshold value is calculated using the following formula: ; where, is the region-adaptive dynamic threshold value; is the reference threshold value, representing the upper limit of the normal operating temperature of the line under standard environmental conditions; is the adjustment factor, representing the correction of the threshold value by the regional characteristics; is the environmental parameter, including the environmental temperature , humidity , wind speed , light intensity and the geographical location information of the line.

[0062] Specifically, the reference threshold value can be determined through experimental data or historical statistical data. For example, under the operating conditions of different types of transmission lines, the typical range may be between 70°C and 90°C.

[0063] Adjustment factor is calculated based on the environmental parameters and the regional characteristic fitting model. In one possible implementation, the calculation formula of the adjustment factor is: ; Where: , , , are weight coefficients, obtained by training based on historical data, reflecting the contribution degree of each environmental parameter to the threshold value; is the environmental temperature, in degrees Celsius (°C); is the humidity, in percentage (%); is the wind speed, in meters per second (m / s); is the light intensity, in watts per square meter (W / m 2 ).

[0064] As an option, the weight coefficient , , , can be dynamically adjusted according to the historical environmental characteristics and temperature distribution law of the region where the line is located. For example, in high-temperature regions, the weight can be set to a larger value; while in coastal areas with significant humidity changes, the weight may be higher.

[0065] In some embodiments, the generation of the dynamic threshold value also needs to consider the specific operation history of the line. For example, for a line with an overheating operation record, its dynamic threshold value can be appropriately reduced to enhance the early fault warning ability.

[0066] In another implementation, the calculation of the adjustment factor can further introduce the historical temperature distribution characteristics of the geographical region. Fit the historical temperature curve of the region where the line is located through a regression model, and correct the threshold value in combination with the real-time environmental parameters. The specific formula is: ; Where: is the real-time temperature of the current environment; is the historical average temperature of the corresponding season in this region; is a correction factor used to balance the impacts of historical temperature and real-time temperature.

[0067] In a possible implementation, the generation of the dynamic threshold can be divided into static and dynamic parts. The static part generates a basic threshold based on regional characteristics and historical data, and the dynamic part is adjusted according to the real-time collected environmental parameters. For example: Under high wind speed conditions, the wind speed will have a greater impact on the heat dissipation of the line, and at this time the dynamic threshold will be appropriately increased.

[0068] Under strong light conditions, the light intensity will cause the surface temperature of the line to rise, and the dynamic threshold will be appropriately decreased to enhance sensitivity.

[0069] In some embodiments, to improve the real-time performance of the generation of the dynamic threshold, the calculation of some adjustment factors can be completed in the edge computing module of the unmanned aerial vehicle. This can reduce the delay of data transmission and processing and improve the response speed of the system.

[0070] The generation of the dynamic threshold plays an important role in the accuracy of fault determination. By fully combining environmental parameters and regional characteristics, this step can provide an adaptive determination benchmark for the temperature anomaly score, effectively reducing the risk of misjudgment caused by environmental fluctuations and improving the sensitivity of fault detection at the same time.

[0071] S5. Compare the temperature anomaly score obtained in step S3 with the dynamic threshold generated in step S4, determine the operating state of the transmission line, and trigger an alarm according to the degree of anomaly.

[0072] Step S5 is to compare the temperature anomaly score obtained in step S3 with the dynamic threshold after completing the generation of the region-adaptive dynamic threshold in step S4, so as to determine the operating state of the transmission line and trigger an alarm according to the degree of anomaly. The core of this step is to use the dynamic threshold as a judgment benchmark, compare the temperature anomaly score with it, automatically classify the line state, and generate an alarm signal to indicate the fault location and its severity at the same time.

[0073] Generally, the operating state of the transmission line is divided into three situations: normal, slightly abnormal, and severely faulty. The dynamic threshold is the key judgment benchmark. It combines the impacts of real-time environmental parameters, geographical location information, and historical data. After comparing with the temperature anomaly score, it can accurately reflect the operating state of the line. In addition, the alarm signal also needs to include complete fault information, such as the fault location, images of the abnormal area, and relevant environmental parameters, to provide data support for subsequent maintenance.

[0074] In this embodiment, the determination conditions of the operating state are specifically as follows: If the temperature anomaly score is less than or equal to the dynamic threshold , it is determined that the line status is normal; If the temperature anomaly score is greater than the dynamic threshold but less than the dynamic threshold plus the preset temperature threshold difference , it is determined that the line status is slightly abnormal; If the temperature anomaly score is greater than the dynamic threshold plus the preset temperature threshold difference , it is determined that the line status is severely faulty and an alarm is triggered.

[0075] The formula expression of the above determination condition is: ; Where: is the temperature anomaly score, generated by step S3, and is used to quantify the temperature anomaly degree of the line; is the dynamic threshold, generated by step S4; is the preset temperature threshold difference, used to divide the slightly abnormal and severely faulty states, and its specific value is set according to the line type and operation requirements, usually in the range of 3°C to 5°C.

[0076] Specifically, the triggering mechanism of the alarm signal is related to the operation status determination result. For the slightly abnormal state, the level of the alarm signal is low, indicating potential hidden dangers but not requiring emergency treatment; for the severely faulty state, the level of the alarm signal is high, requiring the operation and maintenance personnel to immediately conduct fault troubleshooting and repair.

[0077] In a possible implementation manner, the alarm signal includes the following contents: Geographical location of the faulty line: Determine the precise coordinates of the fault point through the GPS data collected in step S1; Infrared image of the abnormal area: Directly mark the hot spot positions in the abnormal area for quick positioning; Environmental parameters: Include the environmental temperature, humidity, wind speed and light intensity at the fault point; Fault classification result: Generated by the classification model in step S3, and is used to indicate the fault type, such as high temperature overload or poor contact.

[0078] As an option, the alarm signal can be transmitted to the ground monitoring system in various ways. For example, the drone can send the alarm signal to the control center in real time through the wireless communication network; in areas with complex terrain or poor signal coverage, the complete alarm information can be sent after the signal is restored through the storage relay method.

[0079] In some embodiments, to enhance the practicality of alarms, trend analysis can also be combined with multi-frame data. For example, for a slight abnormal state, the system will monitor its change trend in multiple time frames: if the abnormal score continues to rise, it will be automatically upgraded to a severe fault state; if the score gradually decreases, it will return to the normal state.

[0080] In another implementation, the generation of alarm signals also combines regional environmental risk assessment. For example, in the high-temperature season or extreme weather conditions, the system can appropriately lower the dynamic threshold to improve the alarm sensitivity; in the case of low-load operation, the threshold is appropriately increased to reduce false alarms.

[0081] The visualization of alarm signals is also an important part of this step. In some embodiments, the ground monitoring system will superimpose and display the alarm signals on the real-time images of the transmission lines. The high-temperature areas are marked with colors (such as red or orange), and the fault range is further prompted by the bounding boxes of the hot spots. This visualization method can significantly improve the response efficiency of the operation and maintenance personnel.

[0082] In a possible implementation, to improve the reliability of fault determination, specific dynamic thresholds and alarm conditions can also be set for different types of transmission lines. For example, for high-voltage lines, the temperature threshold for severe faults may be higher, while for low-voltage lines, the temperature threshold is appropriately lowered to enhance sensitivity.

[0083] Through the above method, step S5 can accurately determine the operating state of the transmission line by combining the dynamic threshold and the temperature anomaly score, and at the same time provide comprehensive fault information through multi-dimensional alarm signals, which provides an important guarantee for the efficient operation and maintenance of the transmission line.

[0084] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An infrared temperature measurement method for transmission lines combined with a dynamic threshold, characterized in that It includes the following steps: S1. Use a drone equipped with an infrared imaging device and a GPS positioning device to collect infrared images, environmental parameters, and geographical location information of the transmission line. Among them, the GPS positioning device is used to record the geographical location information of the transmission line in real time, and the collected data is transmitted to the ground monitoring system through wireless communication; S2. Denoise and standardize the infrared images collected in step S1, and extract their temperature texture features; S3. Use a deep learning model to analyze the infrared images processed in step S2, and output the temperature anomaly score and the fault classification result; S4. Generate a region-adaptive dynamic threshold according to the geographical location information and environmental parameters collected in step S1, as well as historical data; S5. Compare the temperature anomaly score obtained in step S3 with the dynamic threshold generated in step S4, determine the operating state of the transmission line, and trigger an alarm according to the degree of abnormality.

2. The infrared temperature measurement method for transmission lines combining dynamic thresholds according to claim 1, characterized in that In step S1, the environmental parameters include environmental temperature, humidity, wind speed, and light intensity. In step S1, the drone flies along a preset inspection path, and the path covers the key nodes and high-risk areas of the transmission line.

3. A method for infrared temperature measurement of transmission lines combining dynamic thresholds according to claim 1, characterized in that In step S2, when denoising the collected infrared images, the Gaussian filter algorithm is used to smooth the pixel values.

4. A method for infrared temperature measurement of a transmission line combining a dynamic threshold according to claim 1, characterized in that, In step S3, the deep learning model is a convolutional neural network, and the convolutional neural network includes the following parts: Convolutional layer, used to extract the temperature texture features of the infrared images processed in step S2; Pooling layer, used to reduce the dimension of the extracted temperature texture features; Fully connected layer, used to classify the features after dimensionality reduction and generate feature vectors; Output layer, used to output the temperature anomaly score and the fault classification result.

5. A method for infrared temperature measurement of transmission lines combining dynamic thresholds according to claim 1, characterized in that, The formula for generating the dynamic threshold in step S4 is: ； Among them, is the dynamic threshold value, is the reference threshold value, is the adjustment factor, which is calculated based on the environmental parameters and regional characteristics collected in step S1, and the includes environmental temperature, humidity, wind speed, light intensity, and the geographical location of the line.

6. The infrared temperature measurement method for transmission lines combining dynamic thresholds according to claim 5, characterized in that The calculation formula for the adjustment factor is as follows: ; Among them, is the ambient temperature, is the humidity, is the wind speed, is the light intensity, , , , are the weight coefficients obtained by training according to historical data.

7. A method for infrared temperature measurement of a transmission line combining a dynamic threshold according to claim 1, characterized in that, After comparing the temperature anomaly score obtained in step S3 with the dynamic threshold in step S4 in step S5, the determination results include the following: If the temperature anomaly score is less than or equal to the dynamic threshold, it is determined to be in a normal state; If the temperature anomaly score is greater than the dynamic threshold but less than the dynamic threshold plus the preset temperature threshold difference, it is determined to be in a slight anomaly state; If the temperature anomaly score is greater than the dynamic threshold plus the preset temperature threshold difference, it is determined to be in a serious fault state and an alarm is triggered.

8. A method for infrared temperature measurement of transmission lines combining dynamic thresholds according to claim 1, characterized in that, The alarm in step S5 includes the geographical location of the faulty line, the infrared image of the abnormal area, environmental parameters, and the fault classification result.

Images