Power engineering defect diagnosis method based on AI image recognition and blockchain technology

Abstract

The present invention discloses a method for diagnosing defects in electric power engineering based on AI image recognition and blockchain technology, which relates to the technical field of diagnosing defects in electric power engineering, and includes the following steps: according to the monitoring requirements and on-site conditions of the power lines, the initial flight rate of the drone is set to ensure that the drone can cover all key areas of the power lines during the inspection process while maintaining the inspection efficiency. The present invention can intelligently adjust the flight rate of the drone, so that it can stay in potential abnormal areas for a longer time, ensure detailed collection of temperature data, and thus more accurately capture subtle temperature changes, especially minor temperature anomalies caused by arcs and electrical overloads. This intelligent adjustment mechanism ensures that during the inspection process, both the inspection efficiency of the entire process can be guaranteed and the accurate monitoring of key areas can be ensured, reducing the risk of loss of details or missed temperature anomalies due to excessive flight speeds.

Description

Technical Field

[0001] The present invention relates to the technical field of power engineering defect diagnosis, and in particular to a power engineering defect diagnosis method based on AI image recognition and blockchain technology. Background Art

[0002] The application of AI image recognition technology in power engineering defect diagnosis improves equipment monitoring and management capabilities through high-precision automated detection. Using image processing and deep learning algorithms, AI can accurately identify various defects on the surface of power equipment (such as power transmission and transformation equipment, lines, and photovoltaic modules), such as cracks, wear, etc., and can even detect potential problems caused by changes in the external environment. This technology, combined with drone inspections and GIS (geographic information system), realizes fully automated on-site operation and maintenance inspections, greatly reducing manual intervention and improving operating efficiency. For example, in the defect identification of photovoltaic modules, AI image recognition technology has increased detection accuracy by more than 10% and operating efficiency by more than 50%. This automated diagnosis not only saves time, but also improves the accuracy of defect detection, further ensuring the safe operation of equipment.

[0003] Blockchain technology provides a reliable solution for data management and sharing in power engineering. The distributed storage and tamper-proof characteristics of blockchain ensure the security and traceability of the status data of power transmission and transformation equipment. Under this technical framework, information such as inspection data, defect records, and diagnostic reports are recorded in a chain, avoiding the risk of data tampering and ensuring real-time data sharing and transparency among multiple parties. By combining with AI image recognition technology, blockchain not only ensures the integrity of equipment data, but also effectively solves the problems of data transmission delay and privacy protection, and builds a transparent, secure and efficient equipment management network. This technology integration improves efficiency and ensures the safe operation of equipment during the construction of power engineering, and also provides a feasible solution for data management in complex engineering projects.

[0004] When arcing or electrical overload occurs in power lines, it may cause equipment damage or cause fire. AI image recognition combined with infrared thermal imaging technology can effectively detect abnormally high temperatures of wires caused by local overheating caused by arcing or electrical overload, which is crucial to ensuring the safety of the power system. To detect arcs and overloads, existing technologies usually use drones equipped with thermal imaging equipment to collect temperature data of power lines in real time at a constant flight rate to quickly identify overheating problems. This method is suitable for obvious temperature anomalies, but when the temperature anomaly is not obvious, the monitoring accuracy will depend on the image resolution and the dwell time at each monitoring point. If the drone continues to maintain a faster flight rate, the equipment will not be able to stay at each monitoring point for enough time for accurate collection, resulting in reduced image resolution or loss of details, and unable to accurately capture the subtle temperature differences on the surface of the wire. In particular, minor temperature anomalies caused by arcing or electrical overloads may gradually deteriorate if not discovered in time, leading to more serious equipment damage or fire accidents.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a power engineering defect diagnosis method based on AI image recognition and blockchain technology to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a method for diagnosing defects in power engineering based on AI image recognition and blockchain technology, comprising the following steps:

[0008] According to the monitoring requirements and on-site conditions of the power lines, the initial flight speed of the drone is set to ensure that the drone can cover all key areas of the power lines during the inspection while maintaining inspection efficiency;

[0009] At the initial flight speed, the thermal imaging device carried by the drone starts to collect temperature data of the power lines in real time to generate corresponding infrared thermal imaging images;

[0010] Pre-process the collected infrared thermal imaging images to improve data quality and enhance the visibility of potential temperature anomalies;

[0011] Extract key features related to power line temperature anomalies from preprocessed infrared thermal imaging images, further analyze the extracted features, and quantify the analyzed features and convert them into numerical data that can be processed by the machine learning model, providing input for the subsequent machine learning model's potential temperature anomaly prediction;

[0012] The quantized data is passed as input to a pre-trained machine learning model, which predicts potential abnormalities in the power line by analyzing the quantized features and determines whether there is a temperature anomaly.

[0013] When the machine learning model identifies potential power line temperature anomalies, the drone’s target flight rate is dynamically adjusted based on the prediction results, and the initial flight rate is intelligently lowered so that it can stay in suspicious areas longer to collect detailed temperature data. In high-risk areas, it is guaranteed to capture subtle temperature changes more accurately and ensure accurate identification of anomalies.

[0014] All temperature anomaly detection results are recorded in the blockchain to ensure the security, transparency and immutability of the data, and to ensure that each temperature anomaly detection generates a chain record to support subsequent auditing, traceability and data sharing.

[0015] Preferably, the specific steps of collecting temperature data of the power line in real time by thermal imaging equipment and generating corresponding infrared thermal imaging images are as follows:

[0016] Before the drone is equipped with thermal imaging equipment, ensure that the drone is properly started and positioned in the monitoring area of ​​the power line;

[0017] After the drone locates the monitoring area, it starts the thermal imaging equipment and uses the infrared detector to capture the temperature distribution on the surface of the power line in real time. The thermal imaging equipment detects the infrared radiation emitted by the surface of the power line and its surrounding environment, and converts it into temperature data to generate infrared thermal imaging images.

[0018] Preferably, key features related to the temperature anomaly of the power line are extracted from the preprocessed infrared thermal imaging image, wherein the extracted key features include the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area. The extracted temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are further analyzed to generate an infrared texture complexity factor and a hot spot edge fluctuation factor, respectively. The complexity of the temperature distribution texture on the surface of the power line is quantified by the infrared texture complexity factor, and the irregular fluctuation degree of the edge of the high-temperature area is quantified by the hot spot edge fluctuation factor. The temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are converted into numerical data that can be processed by a machine learning model, thereby providing input for the subsequent prediction of potential temperature anomalies by a machine learning model.

[0019] Preferably, the infrared texture complexity factor and the hotspot edge fluctuation factor generated by quantitative analysis of the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are passed as input to a pre-trained machine learning model, and a temperature anomaly assessment coefficient is generated by the machine learning model. The potential anomaly of the power line is predicted based on the temperature anomaly assessment coefficient.

[0020] Preferably, the temperature anomaly assessment coefficient generated when predicting the potential anomaly of the power line by the pre-trained machine learning model is compared and analyzed with the pre-set reference threshold range to determine whether there is a temperature anomaly in the power line. The specific determination steps are as follows:

[0021] If the temperature anomaly assessment coefficient generated during the prediction is greater than the maximum value of the reference threshold range, the temperature anomaly of the power line is determined to be an obvious anomaly, and a high-priority warning is issued to prompt the operation and maintenance personnel to take immediate measures;

[0022] If the temperature anomaly assessment coefficient generated during the prediction is less than the minimum value of the reference threshold range, the temperature anomaly of the power line is determined to be non-abnormal and no further action is required;

[0023] If the temperature anomaly assessment coefficient generated during the prediction is within the reference threshold range, the temperature anomaly of the power line will be judged as a potential anomaly, and a medium-priority warning will be issued, indicating that further tracking, monitoring and analysis are required.

[0024] Preferably, for the pre-processed infrared thermal imaging image, the specific steps of further analyzing the extracted temperature texture distribution to generate the infrared texture complexity factor are as follows:

[0025] First, the temperature gradient change matrix of different scales is extracted from the preprocessed infrared thermal imaging image to quantify the intensity and direction of local temperature change in the image. Calculate the local temperature gradient change tensor , the calculation expression is:

[0026] ,in: Represents the pixel point in the infrared thermal imaging image The temperature value, and Represents pixel points Second order temperature variation in the horizontal and vertical directions, Represents pixel Second-order temperature variation in the diagonal direction;

[0027] After obtaining the local temperature gradient change tensor, the infrared texture complexity factor is calculated. The infrared texture complexity factor quantifies the texture change complexity of the entire image based on feature entropy and information gain. First, the temperature gradient change tensor of each pixel is calculated. Calculate the characteristic entropy. The calculation expression of characteristic entropy is:

[0028] ,in: Pixel The characteristic entropy of is the local temperature gradient change tensor The characteristic value of represents the intensity of temperature change in different directions. i Indicates the direction of temperature change; then, the infrared texture complexity factor of the entire infrared thermal imaging image is calculated by the total information gain of the feature entropy. The calculation expression of the infrared texture complexity factor is:

[0029] ,in: m is the total number of pixels in the image, Represents pixel The absolute value of the temperature gradient change is is the infrared texture complexity factor.

[0030] Preferably, the specific steps of further analyzing the fluctuation degree of the extracted high-temperature region edge to generate the hotspot edge fluctuation factor are as follows:

[0031] In the preprocessed infrared thermal imaging image, the edge of the high temperature area is first extracted, and the edge detection algorithm is used to identify the boundary between the high temperature area and the surrounding area, and the temperature gradient change on the edge of the high temperature area is calculated. The temperature gradient change represents the temperature change rate on the edge of the high temperature area. By accurately calculating the temperature difference of the edge pixels in the image, the temperature gradient change at the edge of the high temperature area is obtained. The calculation expression of the temperature gradient change is:

[0032] ,in, Indicates the pixel point on the edge of the high temperature area The temperature value, Indicates the pixel point on the edge of the high temperature area The temperature value, is the sum of the gradient changes at the edge of one of the high temperature regions, k Indicates the pixel number on the edge of the high temperature area, used to traverse all the pixels in the edge area. x and y is the pixel coordinate in the image;

[0033] Based on the obtained temperature gradient changes, the complexity of edge fluctuations is further quantified to generate the hotspot edge fluctuation factor. The hotspot edge fluctuation factor is calculated by calculating the nonlinear combination of edge gradient changes to capture more complex fluctuation patterns. The calculation expression is:

[0034] ,in, represents the hotspot edge fluctuation factor, is the temperature gradient change of each pixel on the edge of the high temperature area, p andq is a parameter that adjusts the gradient sensitivity, is the exponential magnification factor, It indicates the absolute value of the temperature difference between adjacent pixels in the infrared thermal imaging image, indicating the local temperature change at the edge of the high-temperature area of ​​the power line. N is the total number of edge pixels in the high temperature area.

[0035] Preferably, when the machine learning model identifies a potential power line temperature anomaly, the target flight rate of the drone is dynamically adjusted according to the prediction results. The specific steps are as follows:

[0036] The current temperature anomaly assessment coefficient will be generated by the pre-trained machine learning model and compared with the set reference threshold range. According to the comparison results, the target flight rate of the UAV is dynamically calculated. The calculation expression is:

[0037] ,in: is the target flight speed of the UAV, that is, the actual flight speed after dynamic adjustment. is the initial flight speed of the drone, the standard flight speed set at the beginning of the inspection. The temperature anomaly assessment coefficient is calculated in real time by the machine learning model to measure the temperature anomaly of the power line. is the minimum value of the reference threshold range, indicating the critical value of no abnormal state, is the maximum value of the reference threshold range, indicating the critical value of obvious abnormal state, is the flight speed adjustment coefficient, which is used to control the change range of the target flight speed;

[0038] A new temperature anomaly assessment coefficient is calculated at each monitoring point and adjusted in real time using the flight rate update function. The real-time update formula for the flight rate is as follows:

[0039] ,in: is the updated flight speed at time t, For the The target flight speed at the moment, is the temperature anomaly assessment coefficient at time t, For the The temperature anomaly assessment coefficient at the moment, is the feedback adjustment coefficient, which is used to control the response speed of the UAV flight rate. Indicates positive feedback adjustment.

[0040] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0041] The present invention can intelligently adjust the flight speed of the drone to stay in potential abnormal areas for a longer time, ensuring detailed collection of temperature data, thereby more accurately capturing subtle temperature changes, especially minor temperature anomalies caused by arcing and electrical overload. This intelligent adjustment mechanism ensures both the efficiency of the entire inspection process and the accurate monitoring of key areas during the inspection process, reducing the risk of loss of details or missed temperature anomalies due to excessive flight speed. With the timely discovery of temperature anomalies, the risk of equipment damage or fire accidents is reduced, and the stability and safety of the power system are significantly improved.

[0042] The present invention uses blockchain technology to record all test results, which greatly enhances the security and transparency of the data. Each temperature anomaly data generated by the test will generate a unique chain record, including key information such as the test time, location, and test results, and the reliability and authenticity of the data are ensured by the immutability of the blockchain. This can not only effectively prevent the data from being tampered with or lost, but also provide strong guarantees for subsequent audits, traceability, and data sharing, and improve the standardization and transparency of power engineering management. In addition, the distributed storage characteristics of the blockchain can ensure the real-time sharing of data by multiple parties, promote information exchange and collaborative cooperation between power companies, help accelerate emergency response and decision-making processes, and further improve the intelligent level of power line monitoring and operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0044] Figure 1 This is a method flow chart of the power engineering defect diagnosis method based on AI image recognition and blockchain technology of the present invention. DETAILED DESCRIPTION

[0045] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0046] The present invention provides Figure 1 The power engineering defect diagnosis method based on AI image recognition and blockchain technology shown includes the following steps:

[0047] According to the monitoring requirements and on-site conditions of the power lines, the initial flight speed of the drone is set to ensure that the drone can cover all key areas of the power lines during the inspection while maintaining inspection efficiency;

[0048] The initial flight rate refers to the standard flight speed set by the drone when it starts monitoring the power lines. This flight speed is optimized based on monitoring needs and on-site conditions. Inspections at the initial flight rate ensure that the drone efficiently covers all key areas of the power lines, thereby quickly detecting obvious temperature anomalies, such as obvious hot spots or arcing. The main benefit of this setting is that it significantly improves the overall inspection efficiency, allowing the drone to complete a wide range of inspection tasks in a shorter time, reducing manpower and time costs.

[0049] The initial flight rate is relatively high, and is usually suitable for detecting those abnormal conditions with obvious temperature differences that are easy to identify. For abnormal conditions with small or inconspicuous temperature changes, the initial flight speed is relatively fast, which will cause the thermal imaging equipment to be unable to capture subtle temperature differences and thus be unable to accurately identify these potential hidden dangers. Therefore, the initial flight rate has a significant advantage in improving detection efficiency, but its monitoring accuracy may be limited when facing inconspicuous temperature anomalies.

[0050] At the initial flight speed, the thermal imaging device carried by the drone starts to collect temperature data of the power lines in real time to generate corresponding infrared thermal imaging images;

[0051] The specific steps of collecting temperature data of power lines in real time through thermal imaging equipment and generating corresponding infrared thermal imaging images are as follows:

[0052] Before the drone is equipped with thermal imaging equipment, ensure that the drone is properly started and positioned in the monitoring area of ​​the power line;

[0053] This step includes the take-off operation of the drone, flight path planning, and precise navigation to the predetermined inspection area. Through GPS positioning and automatic flight system, the drone can accurately fly to the predetermined height and position to ensure that the entire area of ​​the power line can be covered. At this time, the flight altitude, flight angle, and flight path of the drone need to be optimized according to factors such as the length, shape, and environmental obstacles of the power line in order to maximize the coverage of the line and avoid obstacles. Accurate positioning and flight path planning can ensure that the drone successfully covers all key areas of the power line to avoid missed inspections. Correct flight preparation ensures the efficiency and high quality of subsequent data collection.

[0054] After the drone locates the monitoring area, it starts the thermal imaging device and uses the infrared detector to capture the temperature distribution on the surface of the power line in real time. The thermal imaging device detects the infrared radiation emitted by the surface of the power line and its surrounding environment, and converts it into temperature data to generate infrared thermal imaging images.

[0055] Infrared thermal imaging images show the temperature differences in different areas, and can effectively identify local overheating or abnormally high temperatures caused by obvious arcs, overloads, short circuits, etc. Thermal imaging equipment usually collects data based on the set temperature threshold, and marks abnormal areas with temperatures higher than the set threshold in the display area to ensure that potential risk points can be captured in real time. This step is the core of data collection. By monitoring the temperature distribution of power lines in real time, thermal imaging equipment provides accurate temperature data. These data provide a key basis for subsequent abnormality detection and fault diagnosis.

[0056] Pre-process the collected infrared thermal imaging images to improve data quality and enhance the visibility of potential temperature anomalies;

[0057] Preprocessing operations include image denoising, contrast enhancement, thermal map smoothing and other technologies. Denoising helps remove noise caused by environmental factors or the equipment itself to avoid misjudgment; contrast enhancement can highlight temperature differences and make small temperature anomalies more obvious; thermal map smoothing helps eliminate high-frequency noise in the image and ensures image smoothness and clarity; preprocessing improves image quality and makes small temperature anomalies easier to identify. Without preprocessing, the noise and interference that may exist in the original thermal imaging image will seriously affect subsequent anomaly detection and analysis.

[0058] Extract key features related to power line temperature anomalies from preprocessed infrared thermal imaging images, further analyze the extracted features, and quantify the analyzed features and convert them into numerical data that can be processed by the machine learning model, providing input for the subsequent machine learning model's potential temperature anomaly prediction;

[0059] Feature extraction is achieved through image analysis methods (such as edge detection, region segmentation, etc.), with the goal of extracting the features that best represent abnormal conditions from a large amount of temperature data, thereby reducing the amount of calculation and improving the efficiency of subsequent processing; feature extraction is to convert complex image data into concise information that can be analyzed, providing efficient input for machine learning models. By extracting key features, we can effectively focus on the most potential abnormal areas, reduce the computational burden, and improve recognition accuracy. Feature quantization is part of data preprocessing to ensure that the data input into the machine learning model is consistent and comparable.

[0060] The key features related to the temperature anomaly of the power line are extracted from the preprocessed infrared thermal imaging images, wherein the extracted key features include the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area. The extracted temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are further analyzed to generate the infrared texture complexity factor and the hot spot edge fluctuation factor respectively. The complexity of the temperature distribution texture on the surface of the power line is quantified by the infrared texture complexity factor, reflecting the hidden local overheating anomaly caused by arc discharge, corrosion or insulation aging on the power line. The irregular fluctuation degree of the edge of the high-temperature area is quantified by the hot spot edge fluctuation factor, which helps to identify the potential temperature anomaly caused by arc discharge, poor contact or local damage on the surface of the power line. The temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are converted into numerical data that can be processed by the machine learning model, thereby providing input for the subsequent machine learning model's prediction of potential temperature anomaly.

[0061] The complexity of the temperature texture distribution in infrared thermal imaging images can usually reflect whether there are potential temperature anomalies in the power line. When the power line is operating normally, the temperature distribution on the surface of the conductor should be relatively uniform, presenting a regular and smooth texture image. If the thermal imaging image shows complex temperature texture distribution, irregular edges, sudden changes in temperature differences in local areas, etc., it may indicate that there are potential overheating, poor contact, arc discharge or insulation aging problems on the surface of the conductor. In particular, when discontinuous high-temperature patches, sudden changes in temperature gradients, or fluctuations in the edges of hot spots appear in the temperature texture image, these abnormal textures are usually precursors to potential faults. Therefore, analyzing the complexity of temperature textures can effectively identify those tiny temperature anomalies that are difficult to detect with the naked eye, providing an important basis for preventive maintenance of power lines.

[0062] For the preprocessed infrared thermal imaging image, the specific steps of further analyzing the extracted temperature texture distribution to generate the infrared texture complexity factor are as follows:

[0063] First, the temperature gradient change matrix of different scales is extracted from the preprocessed infrared thermal imaging image to quantify the intensity and direction of local temperature change in the image. Calculate the local temperature gradient change tensor , the calculation expression is:

[0064] ,in: Represents the pixel point in the infrared thermal imaging image The temperature value, and Represents pixel points Second order temperature variation in the horizontal and vertical directions, Represents pixel Second-order temperature variation in the diagonal direction;

[0065] By constructing a local temperature gradient change tensor, we can capture subtle changes in the local temperature distribution on the power line surface, including edges, patches, and irregular textures. These changes are usually related to problems such as arc discharge, corrosion, or insulation aging, and can reveal hidden local overheating anomalies in the image that are difficult to detect through simple visual inspection.

[0066] After obtaining the local temperature gradient change tensor, the infrared texture complexity factor is calculated. The infrared texture complexity factor quantifies the texture change complexity of the entire image based on feature entropy and information gain. First, the temperature gradient change tensor of each pixel is calculated. Calculate the characteristic entropy. The calculation expression of characteristic entropy is:

[0067] ,in: Pixel The characteristic entropy of is the local temperature gradient change tensor The characteristic value of is the local temperature gradient change tensor The characteristic value is used to describe the temperature variation intensity of the power line surface in different directions. The characteristic value is essentially a numerical value that represents the temperature variation trend of a certain point (pixel) in the infrared thermal imaging image. It represents the temperature variation intensity in different directions. i Indicates the direction of temperature change (horizontal, vertical, diagonal); then, the infrared texture complexity factor of the entire infrared thermal imaging image is calculated by the total information gain of the feature entropy. The calculation expression of the infrared texture complexity factor is:

[0068] ,in: m is the total number of pixels in the image, Represents pixel The absolute value of the temperature gradient change is is the infrared texture complexity factor;

[0069] Characteristic Entropy The information gain part describes the randomness and uncertainty of the local temperature change of each pixel. The information gain part further emphasizes the edge mutation and irregular texture areas through weighted analysis of the temperature gradient change. These areas are often the key features of abnormalities such as arc discharge and insulation aging. By calculating the infrared texture complexity factor , the overall texture complexity of the image can be quantified into a numerical value, which is used to determine whether there is a hidden local overheating anomaly in the power line.

[0070] It can be seen from the infrared texture complexity factor that the larger the performance value of the infrared texture complexity factor generated by further analyzing the temperature texture distribution extracted from the preprocessed infrared thermal imaging image, the greater the hidden danger of potential temperature anomaly in the power line. The infrared texture complexity factor quantifies the complexity of the temperature texture distribution in the infrared thermal imaging image, revealing the intensity and irregularity of the local temperature change on the surface of the power line. When the infrared texture complexity factor is large, it means that the temperature gradient in the image changes significantly and the texture changes more complex, which may be caused by hidden overheating anomalies caused by arc discharge, insulation aging or corrosion; when the infrared texture complexity factor is small, it means that the temperature distribution in the image is more uniform and stable, and the line operation is relatively normal. Therefore, the infrared texture complexity factor is an important reference indicator for assessing the risk of temperature anomaly in power lines, and its value directly reflects the possibility and severity of potential fault hazards in the line.

[0071] The large degree of fluctuation at the edge of the high-temperature area in the infrared thermal imaging image of the power line usually indicates that there is a potential temperature anomaly in the power line. When the power line has problems such as local overheating, poor contact or arcing, the distribution of the high-temperature area is usually irregular, and the edge will show large fluctuations or discontinuities. This fluctuation may be caused by local uneven current, damage to the wire or joint, aging of the insulating material, etc. By analyzing the degree of fluctuation at the edge of the high-temperature area, it can help identify potential temperature anomalies that are difficult to detect with the naked eye, discover hidden dangers of the power line in time, and avoid potential equipment damage or fire accidents. Therefore, the high-temperature area with a large degree of edge fluctuation is an important clue to determine whether there is an abnormality in the power line.

[0072] The specific steps of further analyzing the fluctuation degree of the edge of the extracted high-temperature area to generate the hotspot edge fluctuation factor are as follows:

[0073] In the preprocessed infrared thermal imaging image, the edge of the high temperature area is first extracted, and the edge detection algorithm (such as Canny edge detection) is used to identify the boundary between the high temperature area and the surrounding area, and the temperature gradient change on the edge of the high temperature area is calculated. The temperature gradient change represents the temperature change rate on the edge of the high temperature area, which can effectively reflect the degree of fluctuation of the edge of the high temperature area. By accurately calculating the temperature difference of the edge pixels in the image, the temperature gradient change at the edge of the high temperature area is obtained. The calculation expression of the temperature gradient change is:

[0074] ,in, Indicates the pixel point on the edge of the high temperature area The temperature value, Indicates the pixel point on the edge of the high temperature area The temperature value, is the sum of the gradient changes at the edge of one of the high temperature regions, k Indicates the pixel number on the edge of the high temperature area, used to traverse all the pixels in the edge area. x and y is the pixel coordinate in the image;

[0075] The purpose of this step is to preliminarily capture the intensity of temperature fluctuations on the edge by calculating the gradient change.

[0076] Based on the obtained temperature gradient changes, the complexity of edge fluctuations is further quantified to generate the hotspot edge fluctuation factor. The hotspot edge fluctuation factor is calculated by calculating the nonlinear combination of edge gradient changes to capture more complex fluctuation patterns. The calculation expression is:

[0077] ,in, represents the hotspot edge fluctuation factor, is the temperature gradient change of each pixel on the edge of the high temperature area, p and q is a parameter that adjusts the gradient sensitivity, is the exponential magnification factor, It indicates the absolute value of the temperature difference between adjacent pixels in the infrared thermal imaging image, indicating the local temperature change at the edge of the high-temperature area of ​​the power line. N is the total number of edge pixels in the high temperature area;

[0078] The role of this step is to generate a highly sensitive hotspot edge fluctuation factor by comprehensively considering the nonlinear changes in the edge temperature gradient, further quantify the degree of fluctuation in the high-temperature area, and effectively capture potential temperature anomalies.

[0079] It can be seen from the hotspot edge fluctuation factor that the hotspot edge fluctuation factor generated by further analyzing the fluctuation degree of the edge of the extracted high-temperature area for the preprocessed infrared thermal imaging image is larger, indicating that the hidden danger of potential temperature anomaly in the power line is greater; conversely, the smaller the performance value, the smaller the hidden danger. The hotspot edge fluctuation factor captures those abnormal areas with uneven temperature distribution and drastic boundary changes by quantifying the fluctuation degree of the edge of the high-temperature area in the infrared thermal imaging image. When the fluctuation degree of the edge is large, it usually means that the temperature gradient in the area is unstable, which may be caused by arc discharge, poor contact, overload or insulation aging. Such irregular fluctuations are often important signals of early faults in power lines. Therefore, areas with high hotspot edge fluctuation factors need to be paid special attention to avoid more serious equipment damage or safety accidents. When the hotspot edge fluctuation factor is low, it means that the temperature change at the edge of the high-temperature area is relatively gentle, the temperature distribution tends to be stable, and the potential failure risk is low. Therefore, the size of the hotspot edge fluctuation factor directly reflects the degree of temperature anomaly risk of the power line and is an important indicator for judging whether there are hidden dangers in the equipment.

[0080] The quantized data is passed as input to a pre-trained machine learning model, which predicts potential abnormalities in the power line by analyzing the quantized features and determines whether there is a temperature anomaly.

[0081] The infrared texture complexity factor and hotspot edge fluctuation factor generated by quantitative analysis of the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are passed as input to a pre-trained machine learning model. The temperature anomaly assessment coefficient is generated by the machine learning model, and the potential anomaly of the power line is predicted based on the temperature anomaly assessment coefficient.

[0082] A pre-trained machine learning model means that before it is officially applied to the detection of abnormal temperature of power lines, a large amount of historical data has been used to train the model, so that it has the ability to identify abnormal temperature patterns and evaluate the potential abnormal risks of power lines. The training process includes inputting a large amount of infrared thermal imaging data and related feature parameters (such as infrared texture complexity factor, hot spot edge fluctuation factor, etc.) with abnormal conditions annotated to the model, so that the model can learn and identify the relationship between these features and actual faults. For example, during the training phase, the model will learn the distribution characteristics of high-temperature areas under different conditions (such as different weather environments, wire material types or load levels), the correspondence between edge fluctuation patterns and fault types such as arc discharge and poor contact. Through a large amount of sample data, the model can gradually accurately identify the law of abnormal temperature changes and form a set of evaluation methods for abnormal temperature of power lines. When it is officially applied, this model does not need to be re-learned. It only needs to input the newly collected infrared thermal imaging image features into the model. The model can use the knowledge it has acquired during the training process to quickly analyze and judge the current data and generate the corresponding temperature anomaly evaluation coefficient.

[0083] The reason why this pre-trained machine learning model is important is that the temperature anomaly of power lines is usually affected by many factors, including ambient temperature, load fluctuations, material aging, etc. These factors will lead to a variety of abnormalities. Some anomalies may be more obvious, such as high-temperature hot spots, while others may appear as subtle texture changes or fluctuations at the edge of high-temperature areas. In these cases, manual inspections often have difficulty in timely detecting subtle anomalies, and traditional fixed threshold judgment methods are also prone to missed detections or false detections. Through the training and optimization of the machine learning model, the model can automatically identify different types of abnormal patterns, and quantify the severity of the anomaly based on the input infrared texture complexity factor and hot spot edge fluctuation factor to generate a temperature anomaly assessment coefficient. This assessment coefficient can dynamically reflect the current health status of the power line, help operation and maintenance personnel determine which areas may have potential fault hazards, and prioritize targeted maintenance measures, thereby greatly improving the safety and operation efficiency of the power system. The pre-training process of the model can also be iteratively optimized based on feedback data from actual applications, so that it maintains high accuracy and robustness under different environmental and line conditions.

[0084] The machine learning model is not specifically limited here, and can realize the infrared texture complexity factor and hotspot edge fluctuation factor Comprehensive analysis to generate temperature anomaly assessment coefficients In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; the temperature anomaly evaluation coefficient The resulting calculation formula is:

[0085] , where , are infrared texture complexity factors and hotspot edge fluctuation factor The preset scaling factor of , Both are greater than 0.

[0086] It can be seen from the temperature anomaly assessment coefficient that, for the preprocessed infrared thermal imaging image, the greater the performance value of the infrared texture complexity factor generated by further analyzing the extracted temperature texture distribution, the greater the performance value of the hotspot edge fluctuation factor generated by further analyzing the fluctuation degree of the edge of the extracted high-temperature area. That is, the greater the performance value of the temperature anomaly assessment coefficient generated when predicting the potential anomaly of the power line through the pre-trained machine learning model, the greater the hidden danger of potential temperature anomaly in the power line, and vice versa.

[0087] The temperature anomaly assessment coefficient generated when predicting the potential anomaly of the power line through the pre-trained machine learning model is compared and analyzed with the pre-set reference threshold range to determine whether there is a temperature anomaly in the power line. The specific judgment steps are as follows:

[0088] If the temperature anomaly assessment coefficient generated during the prediction is greater than the maximum value of the reference threshold range, the temperature anomaly of the power line is determined to be an obvious anomaly, and a high-priority warning is issued to prompt the operation and maintenance personnel to take immediate measures;

[0089] If the temperature anomaly assessment coefficient generated during the prediction is less than the minimum value of the reference threshold range, the temperature anomaly of the power line is determined to be non-abnormal and no further action is required;

[0090] If the temperature anomaly assessment coefficient generated during the prediction is within the reference threshold range, the temperature anomaly of the power line will be judged as a potential anomaly (that is, not an obvious anomaly), and a medium-priority warning will be issued, indicating that further tracking, monitoring and analysis are needed.

[0091] When the temperature anomaly assessment coefficient generated by the machine learning model is greater than the maximum value of the reference threshold range, it means that the temperature anomaly of the power line has exceeded the normal range. This anomaly usually manifests itself as significant overheating of the conductor surface, concentrated distribution of local high-temperature areas, or damage to insulating materials. Such obvious anomalies are most likely caused by faults such as arc discharge, overload, and poor contact. If not handled in time, it may cause serious accidents such as equipment damage and fire. Therefore, in this case, the system will judge the temperature status of the power line as "obviously abnormal" and issue a high-priority warning, prompting the operation and maintenance personnel to take immediate measures.

[0092] When the temperature anomaly assessment coefficient is less than the minimum value of the reference threshold range, it means that the temperature distribution of the power line is completely within the normal range and there are no signs of abnormality. This situation shows that the line has good conductivity, the equipment is running stably, and no significant overheating or local temperature mutations are found. Therefore, the system will judge the temperature status of the power line as "no abnormality" and no further action is required. This type of judgment result can effectively reduce misjudgments and unnecessary maintenance, thereby reducing operation and maintenance costs.

[0093] When the temperature anomaly assessment coefficient is between the reference threshold range, it indicates that there are some subtle abnormal changes in the temperature distribution of the power line. This anomaly is usually difficult to identify by the traditional fixed threshold judgment method, and may appear as temperature fluctuations with small local temperature differences, slight changes in the surface texture of the conductor, or irregular fluctuations at the edge of the hot spot. Such anomalies may be signs of early failures, such as insulation aging, tiny arc discharges, or mild poor contact. Although these anomalies will not cause immediate equipment damage, they may gradually deteriorate over time if not taken seriously. Therefore, the system determines these abnormal conditions as "potential anomalies" and issues a medium-priority warning, indicating that further tracking, monitoring, and analysis are required.

[0094] When the machine learning model identifies potential power line temperature anomalies, the drone’s target flight rate is dynamically adjusted based on the prediction results, and the initial flight rate is intelligently lowered so that it can stay in suspicious areas longer to collect detailed temperature data. In high-risk areas, it is guaranteed to capture subtle temperature changes more accurately and ensure accurate identification of anomalies.

[0095] When the machine learning model identifies a potential power line temperature anomaly, the target flight rate of the drone is dynamically adjusted based on the prediction results. The specific steps are as follows:

[0096] The target flight rate of the drone is dynamically adjusted from its initial flight rate according to the temperature anomaly assessment coefficient calculated in real time to cope with different degrees of temperature anomalies. During the flight, the current temperature anomaly assessment coefficient will be generated by the pre-trained machine learning model and compared with the set reference threshold range. According to the comparison results, the target flight rate is dynamically calculated by the following formula, and the calculation expression is:

[0097] ,in: is the target flight speed of the UAV, that is, the actual flight speed after dynamic adjustment. is the initial flight speed of the drone, the standard flight speed set at the beginning of the inspection. The temperature anomaly assessment coefficient is calculated in real time by the machine learning model to measure the temperature anomaly of the power line. is the minimum value of the reference threshold range, indicating the critical value of no abnormal state, is the maximum value of the reference threshold range, indicating the critical value of obvious abnormal state, The flight speed adjustment coefficient is used to control the change range of the target flight speed. It is usually set according to the actual inspection requirements. indicates a more sensitive adjustment;

[0098] This step ensures that when the temperature anomaly assessment coefficient is higher, the target flight rate of the drone is lower, so that it stays in the high-risk area for a longer time and collects more detailed temperature data. If the temperature anomaly assessment coefficient is close to the maximum value of the reference threshold range, the target flight rate will be significantly reduced, even close to the hovering state, to ensure that the subtle temperature changes on the surface of the wire can be accurately captured. When the temperature anomaly assessment coefficient is low, the drone maintains a higher flight rate to improve inspection efficiency.

[0099] In the actual inspection process, the drone needs to dynamically update the flight rate adjustment strategy based on the real-time collected temperature data to ensure the continuous optimization of the target flight rate. To achieve this, a new temperature anomaly assessment coefficient is calculated at each monitoring point and adjusted in real time using the flight rate update function. The real-time update formula for the flight rate is as follows:

[0100] ,in: is the updated flight speed at time t, For the The target flight speed at the moment, is the temperature anomaly assessment coefficient at time t, For the The temperature anomaly assessment coefficient at the moment, is the feedback adjustment coefficient, which is used to control the response speed of the UAV flight rate. It indicates positive feedback adjustment;

[0101] This step introduces a feedback mechanism to ensure that the drone's flight rate can respond to temperature anomalies in real time. If the current temperature anomaly assessment coefficient increases significantly compared to the previous moment, the system will further reduce the target flight rate so that the drone stays in the temperature anomaly area longer to improve monitoring accuracy. At the same time, the feedback adjustment coefficient The introduction of can effectively control the change range of flight speed, prevent frequent fluctuations in flight speed from interfering with inspection tasks, and ensure that the flight path and data collection process of the drone are stable and reliable;

[0102] Positive feedback ( ) means that the system adjusts the flight speed according to the direction of change of the temperature anomaly assessment coefficient. For example, when the temperature anomaly increases, the flight speed is reduced; when the temperature anomaly decreases, the flight speed is increased.

[0103] All temperature anomaly detection results are recorded in the blockchain to ensure data security, transparency and immutability, and to ensure that each temperature anomaly detection generates a chain record to support subsequent auditing, traceability and data sharing;

[0104] All temperature anomaly detection results are recorded in the blockchain to ensure the security, transparency and credibility of temperature anomaly detection data. By recording each test result in the blockchain, the characteristics of immutability and distributed storage are utilized to effectively prevent data from being tampered with or forged. Each test record contains key monitoring information, such as timestamps, test results, drone flight paths, etc., so that the data can be accurately traced and used for subsequent audits. This not only improves the reliability of power line monitoring data, but also provides a convenient and reliable basis for subsequent data sharing, ensuring that all parties can rely on the authenticity and effectiveness of this data when viewing and using it, and enhancing the supervision and compliance of the system.

[0105] Through the above-mentioned power engineering defect diagnosis method based on AI image recognition and blockchain technology, first of all, the accuracy and efficiency of power line temperature anomaly detection can be significantly improved. By intelligently adjusting the flight speed of the drone, it can stay in the potential abnormal area for a longer time to ensure the detailed collection of temperature data, so as to more accurately capture subtle temperature changes, especially the slight temperature anomalies caused by arcing and electrical overload. This intelligent adjustment mechanism ensures that during the inspection process, it can not only ensure the inspection efficiency of the entire process, but also ensure accurate monitoring of key areas, reducing the risk of loss of details or missed temperature anomalies due to excessive flight speed. With the timely detection of temperature anomalies, the risk of equipment damage or fire accidents is reduced, and the stability and safety of the power system are significantly improved.

[0106] Secondly, the use of blockchain technology to record all test results greatly enhances the security and transparency of data. Each test generates a unique chain record of abnormal temperature data, including key information such as test time, location, and test results, and the reliability and authenticity of the data are ensured by the immutability of blockchain. This can not only effectively prevent data from being tampered with or lost, but also provide strong guarantees for subsequent audits, traceability, and data sharing, and improve the standardization and transparency of power engineering management. In addition, the distributed storage characteristics of blockchain can ensure real-time sharing of data by multiple parties, promote information exchange and collaboration between power companies, help accelerate emergency response and decision-making processes, and further improve the level of intelligence in power line monitoring and operation and maintenance.

[0107] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0108] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0109] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A power engineering defect diagnosis method based on AI image recognition and blockchain technology, characterized in that: The following steps are involved: According to the monitoring requirements and on-site conditions of the power lines, the initial flight speed of the drone is set to ensure that the drone can cover all key areas of the power lines during the inspection while maintaining inspection efficiency; At the initial flight speed, the thermal imaging device carried by the drone starts to collect temperature data of the power lines in real time to generate corresponding infrared thermal imaging images; Pre-process the collected infrared thermal imaging images to improve data quality and enhance the visibility of potential temperature anomalies; Extract key features related to power line temperature anomalies from preprocessed infrared thermal imaging images, further analyze the extracted features, and quantify the analyzed features and convert them into numerical data that can be processed by the machine learning model, providing input for the subsequent machine learning model's potential temperature anomaly prediction; The quantized data is passed as input to a pre-trained machine learning model, which predicts potential abnormalities in the power line by analyzing the quantized features and determines whether there is a temperature anomaly. When the machine learning model identifies potential power line temperature anomalies, the drone’s target flight rate is dynamically adjusted based on the prediction results, and the initial flight rate is intelligently lowered so that it can stay in suspicious areas longer to collect detailed temperature data. In high-risk areas, it is guaranteed to capture subtle temperature changes more accurately and ensure accurate identification of anomalies. All temperature anomaly detection results are recorded in the blockchain to ensure data security, transparency and immutability, and to ensure that each temperature anomaly detection generates a chain record to support subsequent auditing, traceability and data sharing; When the machine learning model identifies a potential power line temperature anomaly, the target flight rate of the drone is dynamically adjusted based on the prediction results. The specific steps are as follows: The current temperature anomaly assessment coefficient will be generated by the pre-trained machine learning model and compared with the set reference threshold range. According to the comparison results, the target flight rate of the UAV is dynamically calculated. The calculation expression is: ,in: is the target flight speed of the UAV, that is, the actual flight speed after dynamic adjustment. is the initial flight speed of the drone, the standard flight speed set at the beginning of the inspection. The temperature anomaly assessment coefficient is calculated in real time by the machine learning model to measure the temperature anomaly of the power line. is the minimum value of the reference threshold range, indicating the critical value of no abnormal state, is the maximum value of the reference threshold range, indicating the critical value of obvious abnormal state, is the flight speed adjustment coefficient, which is used to control the change range of the target flight speed; A new temperature anomaly assessment coefficient is calculated at each monitoring point and adjusted in real time using the flight rate update function. The real-time update formula for the flight rate is as follows: in: is the updated flight speed at time t, For the The target flight speed at the moment, is the temperature anomaly assessment coefficient at time t, For the The temperature anomaly assessment coefficient at the moment, is the feedback adjustment coefficient, which is used to control the response speed of the UAV flight rate. Indicates positive feedback adjustment.

2. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 1 is characterized in that: The specific steps of collecting temperature data of power lines in real time through thermal imaging equipment and generating corresponding infrared thermal imaging images are as follows: Before the drone is equipped with thermal imaging equipment, ensure that the drone is properly started and positioned in the monitoring area of ​​the power line; After the drone locates the monitoring area, it starts the thermal imaging equipment and uses the infrared detector to capture the temperature distribution on the surface of the power line in real time. The thermal imaging equipment detects the infrared radiation emitted by the surface of the power line and its surrounding environment, and converts it into temperature data to generate infrared thermal imaging images.

3. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 1 is characterized in that: The key features related to the temperature anomaly of the power line are extracted from the preprocessed infrared thermal imaging images, wherein the extracted key features include the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area. The extracted temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are further analyzed to generate the infrared texture complexity factor and the hot spot edge fluctuation factor respectively. The complexity of the temperature distribution texture on the surface of the power line is quantified by the infrared texture complexity factor. The irregular fluctuation degree of the edge of the high-temperature area is quantified by the hot spot edge fluctuation factor. The temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are converted into numerical data that can be processed by the machine learning model, so as to provide input for the subsequent machine learning model to predict potential temperature anomalies.

4. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 3 is characterized in that: The infrared texture complexity factor and hotspot edge fluctuation factor generated by quantitative analysis of the temperature texture distribution and the fluctuation degree of the edge of the high-temperature area are passed as input to a pre-trained machine learning model. The temperature anomaly assessment coefficient is generated by the machine learning model, and the potential anomaly of the power line is predicted based on the temperature anomaly assessment coefficient.

5. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 4 is characterized in that: The temperature anomaly assessment coefficient generated when predicting the potential anomaly of the power line through the pre-trained machine learning model is compared and analyzed with the pre-set reference threshold range to determine whether there is a temperature anomaly in the power line. The specific judgment steps are as follows: If the temperature anomaly assessment coefficient generated during the prediction is greater than the maximum value of the reference threshold range, the temperature anomaly of the power line is determined to be an obvious anomaly, and a high-priority warning is issued to prompt the operation and maintenance personnel to take immediate measures; If the temperature anomaly assessment coefficient generated during the prediction is less than the minimum value of the reference threshold range, the temperature anomaly of the power line is determined to be non-abnormal and no further action is required; If the temperature anomaly assessment coefficient generated during the prediction is within the reference threshold range, the temperature anomaly of the power line will be judged as a potential anomaly, and a medium-priority warning will be issued, indicating that further tracking, monitoring and analysis are required.

6. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 3 is characterized in that: For the preprocessed infrared thermal imaging image, the specific steps of further analyzing the extracted temperature texture distribution to generate the infrared texture complexity factor are as follows: First, the temperature gradient change matrix of different scales is extracted from the preprocessed infrared thermal imaging image to quantify the intensity and direction of local temperature change in the image. Calculate the local temperature gradient change tensor , the calculation expression is: ,in: Represents the pixel point in the infrared thermal imaging image The temperature value, and Represents pixel points Second order temperature variation in the horizontal and vertical directions, Represents pixel Second-order temperature variation in the diagonal direction; After obtaining the local temperature gradient change tensor, the infrared texture complexity factor is calculated. The infrared texture complexity factor quantifies the texture change complexity of the entire image based on feature entropy and information gain. First, the temperature gradient change tensor of each pixel is calculated. Calculate the characteristic entropy. The calculation expression of characteristic entropy is: ,in: Pixel The characteristic entropy of is the local temperature gradient change tensor The characteristic value of represents the intensity of temperature change in different directions. i Indicates the direction of temperature change; then, the infrared texture complexity factor of the entire infrared thermal imaging image is calculated by the total information gain of the feature entropy. The calculation expression of the infrared texture complexity factor is: ,in: m is the total number of pixels in the image, Represents pixel The absolute value of the temperature gradient change is is the infrared texture complexity factor.

7. The power engineering defect diagnosis method based on AI image recognition and blockchain technology according to claim 6 is characterized in that: The specific steps of further analyzing the fluctuation degree of the edge of the extracted high-temperature area to generate the hotspot edge fluctuation factor are as follows: In the preprocessed infrared thermal imaging image, the edge of the high temperature area is first extracted, and the edge detection algorithm is used to identify the boundary between the high temperature area and the surrounding area, and the temperature gradient change on the edge of the high temperature area is calculated. The temperature gradient change represents the temperature change rate on the edge of the high temperature area. By accurately calculating the temperature difference of the edge pixels in the image, the temperature gradient change at the edge of the high temperature area is obtained. The calculation expression of the temperature gradient change is: ,in, Indicates the pixel point on the edge of the high temperature area The temperature value, Indicates the pixel point on the edge of the high temperature area The temperature value, is the sum of the gradient changes at the edge of one of the high temperature regions, k Indicates the pixel number on the edge of the high temperature area, used to traverse all the pixels in the edge area. x and y is the pixel coordinate in the image; Based on the obtained temperature gradient changes, the complexity of edge fluctuations is further quantified to generate the hotspot edge fluctuation factor. The hotspot edge fluctuation factor is calculated by calculating the nonlinear combination of edge gradient changes to capture more complex fluctuation patterns. The calculation expression is: ,in, represents the hotspot edge fluctuation factor, is the temperature gradient change of each pixel on the edge of the high temperature area, p and q is a parameter that adjusts the gradient sensitivity, is the exponential magnification factor, It indicates the absolute value of the temperature difference between adjacent pixels in the infrared thermal imaging image, indicating the local temperature change at the edge of the high-temperature area of ​​the power line. N is the total number of edge pixels in the high temperature area.

Images