A method and device for detecting insulator leakage current based on video monitoring

Abstract

The application discloses a kind of insulator leakage current detection methods based on video monitoring, comprising the following steps: visible light video detection is carried out to the insulator to be detected, obtains first monitoring video, carries out thermal imaging video detection to insulator, obtains second monitoring video;Discharge trace size identification is carried out to first monitoring video, obtains discharge trace area, and first leakage current information is obtained by classification;Second monitoring video is identified, and second leakage current information is obtained;According to first leakage current information and second leakage current information, leakage current information is calculated and obtained, as the leakage current detection result of insulator.The application also discloses corresponding method.Implementation of the present application can improve the accuracy of insulator leakage current detection, improve detection effect.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method and device for detecting leakage current in insulators based on video monitoring. Background Technology

[0002] Insulators are indispensable key components in power systems, their primary function being to isolate conductors from the ground or other conductive structures to prevent electrical leakage or short circuits. However, insulators are susceptible to environmental factors during long-term outdoor use, such as the deposition of pollutants like rainwater, fog, dust, and salt. These pollutants form conductive channels on the insulator surface, leading to leakage current. If the leakage current is too high, it will cause the insulator to break down or flashover, seriously affecting the safety and stability of the power system.

[0003] Currently, insulator leakage current detection generally uses sensors, which have low accuracy and suffer from technical problems such as low detection accuracy and poor detection effect. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and device for detecting leakage current of insulators based on video monitoring, which can improve the accuracy of insulator leakage current detection and improve the detection effect.

[0005] To address the aforementioned technical problems, as a first aspect of the present invention, a method for detecting insulator leakage current based on video surveillance is provided, the method comprising:

[0006] Step S1: Perform visible light video detection on the insulator to be inspected to obtain a first monitoring video, and perform thermal imaging video detection on the insulator to obtain a second monitoring video;

[0007] Step S2: Identify the size of the discharge traces in the first monitoring video, obtain the area of ​​the discharge traces, and classify them to obtain the first leakage current information;

[0008] Step S3: Identify the second monitoring video to obtain the second leakage current information;

[0009] Step S4: Calculate the leakage current information based on the first leakage current information and the second leakage current information, and use it as the leakage current detection result of the insulator.

[0010] Preferably, step S1 further includes:

[0011] Under visible light, video detection and acquisition are performed on the insulator to be tested to obtain the first monitoring video;

[0012] Thermal imaging video was captured on the insulator to obtain a second monitoring video.

[0013] Preferably, step S2 further includes:

[0014] Construct a discharge trace detector;

[0015] The first monitoring video is input into the discharge trace identifier, and the discharge trace area is obtained by the identification output.

[0016] Preferably, in step S2, constructing the discharge trace detector includes:

[0017] Collect a set of first monitoring videos of insulator samples, and mark the area of ​​discharge traces in the first monitoring videos of the samples to obtain a set of discharge trace areas of the samples.

[0018] A convolutional neural network is used to train a discharge trace identifier using the first set of monitoring videos and the set of discharge trace areas. The discharge trace identifier is trained using gradient descent, and the training rule is as follows:

[0019]

[0020] Where, θ t Here, η represents the model parameters at the t-th iteration, and η is the step size. For gradient.

[0021] Preferably, in step S2, the classification to obtain the first leakage current information includes:

[0022] Based on the historical leakage current detection data of the insulators, the sample discharge trace area set and the sample first leakage current information set are collected;

[0023] A mapping relationship is constructed between the set of sample discharge trace areas and the set of sample first leakage current information to obtain a first leakage current classifier;

[0024] The discharge trace area is input into the first leakage current classifier to classify and obtain the first leakage current information.

[0025] Preferably, step S3 further includes:

[0026] Collect a second set of monitoring videos of the insulator samples, and collect a second set of leakage current information for the samples;

[0027] Using a convolutional neural network, a second leakage current identifier is trained using the sample second monitoring video set and the sample second leakage current information set.

[0028] The second monitoring video is input into the second leakage current identifier to identify and obtain the second leakage current information.

[0029] Preferably, step S4 further includes:

[0030] The first leakage current information and the second leakage current information are weighted and calculated to obtain leakage current information, which is used as the leakage current detection result of the insulator.

[0031] As another aspect of the present invention, an insulator leakage current detection device based on video monitoring is also provided, including a processor and a memory, wherein the memory stores a computer program that can be executed by the processor; when the processor executes the computer program, it executes the insulator leakage current detection method based on video monitoring as described in the present invention.

[0032] Implementing the embodiments of the present invention has the following beneficial effects:

[0033] This invention provides a video monitoring-based method for detecting insulator leakage current. By employing both visible light video and thermal imaging video monitoring, it comprehensively acquires information on discharge traces and temperature anomalies on the insulator surface, significantly improving detection accuracy and avoiding the errors of single monitoring methods. Furthermore, both thermal imaging and visible light detection maintain good stability in complex environments, enhancing the environmental adaptability of the detection method.

[0034] Meanwhile, the video monitoring method enables continuous and real-time detection, which can promptly identify and warn when leakage current increases sharply, thus improving system safety.

[0035] In addition, the flexible deployment of video surveillance expands the monitoring coverage, enabling it to cover remote or difficult-to-maintain areas and reduce blind spots.

[0036] In summary, the solution of this invention achieves automatic classification and calculation of leakage current information, enabling more efficient and accurate leakage current monitoring. It meets the intelligent requirements of power systems and provides comprehensive and reliable technical support for the efficient operation and maintenance of power systems. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, obtaining other drawings based on these drawings without creative effort still falls within the scope of the present invention.

[0038] Figure 1 This is a schematic diagram of the main flow of an insulator leakage current detection method based on video monitoring provided by the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.

[0040] like Figure 1 The diagram shows the main flow of an embodiment of a video surveillance-based insulator leakage current detection method provided by the present invention. In this embodiment, the method includes at least the following steps:

[0041] Step S1: Perform visible light video detection on the insulator to be inspected to obtain a first monitoring video, and perform thermal imaging video detection on the insulator to obtain a second monitoring video;

[0042] Step 1 of the method provided in this embodiment of the invention includes:

[0043] Under visible light, video detection and acquisition are performed on the insulator to be tested to obtain the first monitoring video;

[0044] Thermal imaging video was captured on the insulator to obtain a second monitoring video.

[0045] In this embodiment of the invention, the insulator to be inspected is first subjected to visible light video inspection, that is, a visible light camera is used to acquire an image of the insulator's appearance, generating a clear first monitoring video. Visible light video inspection can record the external characteristics of the insulator, such as the visible range and shape changes of discharge traces. Discharge traces refer to the marks left by corona discharge phenomena generated on the surface of the insulator. These traces are usually due to possible contamination or micro-cracks on the surface of the insulator.

[0046] In this embodiment of the invention, thermal imaging video detection is performed on the insulator to obtain a second monitoring video. Thermal imaging video is a visual image generated by capturing the temperature distribution on and around the insulator surface using a thermal imaging camera. Thermal imaging technology can detect areas of abnormal temperature, especially localized overheating or abnormal temperature distribution, which is usually caused by leakage current.

[0047] Thermal imaging cameras can detect the infrared radiation emitted by objects and convert it into a visible temperature image, or "thermal map." This thermal map can reveal potential temperature anomalies on insulators, thereby identifying the magnitude of leakage current causing the temperature anomalies.

[0048] Leakage current can cause localized temperature increases. Monitoring the temperature changes of the insulator using a second monitoring video can effectively detect these abnormal areas of temperature rise, improving the accuracy of leakage current detection.

[0049] Step S2: Identify the size of the discharge traces in the first monitoring video, obtain the area of ​​the discharge traces, and classify them to obtain the first leakage current information;

[0050] In this embodiment of the invention, a detailed image analysis is first performed on the first monitoring video to identify the size of possible discharge traces on the surface of the insulator, i.e., the area of ​​the discharge traces.

[0051] After identifying the area of ​​the discharge trace, it is classified using a preset classification algorithm to obtain preliminary leakage current information, i.e., the first leakage current information. The larger the area of ​​the discharge trace, the greater the first leakage current information.

[0052] Step S2 in the method provided in this embodiment of the invention includes:

[0053] In this embodiment of the invention, a discharge trace identifier is constructed specifically for identifying discharge traces on the surface of insulators. This identifier, based on advanced image processing algorithms and deep learning technology, is capable of accurately extracting the features of discharge traces from video images. The discharge trace identifier is trained on a large number of labeled insulator discharge samples to recognize the characteristics of various discharge traces, such as the area of ​​the discharge traces.

[0054] The construction of the discharge trace detector includes:

[0055] Collect a set of first monitoring videos of insulator samples, and mark the area of ​​discharge traces in the first monitoring videos of the samples to obtain a set of discharge trace areas of the samples.

[0056] A convolutional neural network is used to train a discharge trace recognizer using the first set of monitoring videos and the set of discharge trace areas of the samples.

[0057] In this embodiment of the invention, a large number of sample monitoring videos of insulators are first collected, namely, video samples of insulators taken under different conditions, recording possible discharge traces on their surfaces. To facilitate subsequent model training, the area of ​​visible discharge traces in each frame of these sample videos is manually or semi-automatically marked. The marking of discharge trace areas includes outlining the discharge traces and calculating their areas, generating a sample discharge trace area set as real reference data for training.

[0058] For example, a convolutional neural network (CNN) is used to train the discharge trace detector. A CNN is a deep learning model widely used in image recognition. It extracts and identifies local features in images through multiple convolutional and pooling layers, making it particularly suitable for processing complex graphics and shapes in images. The collected first set of monitoring videos and the set of sample discharge trace areas are input into the CNN. Through repeated iterations with a large number of training samples, the detector's accuracy is gradually optimized, enabling it to automatically detect discharge traces of different shapes in the videos.

[0059] The operation of the convolutional layer in the discharge trace detector for extracting convolutional features from the insulator image in the first monitoring video is as follows:

[0060] S(i,j)=(I*K)(i,j)=∑ m ∑ n I(i+m,j+n)·K(m,n);

[0061] Where S(i,j) is the output after convolution, I(i,j) is the pixel value of the insulator image, K(m,n) is the weight of the convolution kernel, and (i,j) is the coordinate of the pixel position in the insulator image.

[0062] The discharge trace detector is trained using gradient descent, and the training rules are as follows:

[0063]

[0064] Where, θ t Here, η represents the model parameters at the t-th iteration, and η is the step size. For gradient.

[0065] During training, the convolutional neural network adjusts its internal parameters by comparing the differences between the recognition results and the labeled data, and is trained through gradient descent to gradually improve the accuracy of identifying the area of ​​discharge traces. The ultimately trained discharge trace identifier can accurately identify the area information of discharge traces from actual videos, providing real-time and reliable input data for insulator leakage current detection. Through this convolutional neural network-based training process, the discharge trace identifier not only has the ability to efficiently process large amounts of video data, but also possesses strong generalization ability, maintaining high recognition accuracy in different environments.

[0066] The first monitoring video of the insulator currently being collected is input into the discharge trace identifier that has been trained, and the output shows the area of ​​the discharge trace.

[0067] The establishment of the discharge trace detector enables automated and high-precision identification of the area of ​​traces caused by current leakage on the surface of insulators through video detection. Through an efficient and accurate image processing workflow, the area information of discharge traces can be obtained in real time, providing important basic data for further analysis and monitoring of leakage current.

[0068] In this embodiment of the invention, obtaining the first leakage current information by classification includes:

[0069] Based on the historical leakage current detection data of the insulators, the sample discharge trace area set and the sample first leakage current information set are collected;

[0070] A mapping relationship is constructed between the set of sample discharge trace areas and the set of sample first leakage current information to obtain a first leakage current classifier;

[0071] The discharge trace area is input into the first leakage current classifier to classify and obtain the first leakage current information.

[0072] In this embodiment of the invention, based on historical leakage current detection data of insulators, a large number of insulator samples with discharge traces are collected. The set of discharge trace areas and the corresponding leakage current data for each sample are obtained, generating a first leakage current information set for the sample. The sample discharge trace area set includes the size of the discharge trace for each sample, while the first leakage current information set includes the corresponding leakage current value, such as 10mA. The collection of this data lays the data foundation for establishing a relationship model between discharge traces and leakage current.

[0073] A mapping relationship is constructed between the set of sample discharge trace areas and the set of sample first leakage current information to generate a first leakage current classifier. The mapping relationship is essentially finding the correspondence between discharge trace area and leakage current in the data; that is, the size of the discharge trace area corresponds to different leakage current levels within different ranges. The first leakage current classifier is the mapping table between the set of sample discharge trace areas and the set of sample first leakage current information.

[0074] The currently identified discharge trace area is input into the first leakage current classifier. The classifier maps and classifies the area according to a pre-established mapping relationship and outputs the corresponding first leakage current information. For example, the first leakage current information of the sample discharge trace area that is closest to the current discharge trace area (the difference between the two is the smallest) is output as the first leakage current information.

[0075] The initial leakage current information, determined based on the area of ​​the discharge traces, provides a reference for further insulator health assessment. This classification model based on historical data makes leakage current detection more efficient and accurate, enabling dynamic assessment of the insulator's condition and improving the accuracy and practicality of the detection.

[0076] Step S3: Identify the second monitoring video to obtain the second leakage current information;

[0077] In this embodiment of the invention, the second monitoring video is identified and analyzed to identify leakage current based on the temperature anomaly of the insulator in the thermal imaging video, thereby obtaining second leakage current information.

[0078] Step 3 in the method provided in this embodiment of the invention includes:

[0079] Collect a second set of monitoring videos of the insulator samples, and collect a second set of leakage current information for the samples;

[0080] Using a convolutional neural network, a second leakage current identifier is trained using the sample second monitoring video set and the sample second leakage current information set.

[0081] The second monitoring video is input into the second leakage current identifier to identify and obtain the second leakage current information.

[0082] In this embodiment of the invention, a second set of monitoring videos of insulator samples is collected. This second set of monitoring videos is a collection of thermal imaging videos of insulators taken by a thermal imaging camera, recording the temperature distribution on the insulator surface under different environmental conditions and pollution states. Correspondingly, the actual leakage current data of each video segment is collected to form a second set of leakage current information. This second set of leakage current information records the leakage current level of the insulator under different temperature states, providing accurate reference data for the model.

[0083] Furthermore, a convolutional neural network (CNN) was used to train the second leakage current identifier. A CNN is a deep learning model highly suited for image processing. Through multi-layer convolution and pooling operations on images, it can extract features from temperature images, such as the size, location, and shape of abnormal temperature regions. These features are related to leakage current information. The sample set of second monitoring videos and the sample set of second leakage current information were input into the CNN for training. Through repeated iterations and error correction, the network gradually learned the relationship between different temperature image features and leakage current levels.

[0084] After training, the second monitoring video of the actual detection is input into the second leakage current identifier. Based on its understanding of temperature image characteristics, the identifier analyzes the temperature distribution in the video, especially the size of the overheated area and the temperature change pattern, and outputs the second leakage current information. The second leakage current information is the leakage current judgment result obtained after thermal imaging video analysis, which can reflect the correlation between the real-time temperature state of the insulator and the current leakage level.

[0085] The second leakage current identifier can efficiently and accurately map temperature features in video images to leakage current levels, realizing intelligent correlation analysis between temperature anomalies and leakage current, and providing more comprehensive technical support for insulator condition monitoring.

[0086] Step S4: Calculate the leakage current information based on the first leakage current information and the second leakage current information, and use it as the leakage current detection result of the insulator.

[0087] In this embodiment of the invention, the first leakage current information and the second leakage current information are combined to calculate the final leakage current information, which serves as the leakage current detection result for the insulator. The first leakage current information is the leakage current level inferred based on the size and distribution of discharge traces in visible light video, reflecting the visible discharge situation on the insulator surface. The second leakage current information originates from temperature anomalies detected in thermal imaging video; the leakage current causing the temperature change is obtained through thermal image analysis of the insulator surface. These two sources of information reveal the state of the insulator's leakage current from different perspectives, complementing each other and contributing to a more complete judgment.

[0088] Step S4 in the method provided in this embodiment of the invention includes:

[0089] The first leakage current information and the second leakage current information are weighted and calculated to obtain leakage current information, which is used as the leakage current detection result of the insulator.

[0090] In this embodiment of the invention, the first leakage current information and the second leakage current information are weighted and calculated to generate the final leakage current information, which serves as the leakage current detection result of the insulator.

[0091] The formula for weighted calculation of the first leakage current information and the second leakage current information is as follows:

[0092] C final =w1·C1+w2·C2;

[0093] Among them, C final For leakage current information, w1 and w2 are weights, the sum of w1 and w2 is 1, and C1 and C2 are the first leakage current information and the second leakage current information.

[0094] During the weighted calculation process, appropriate weights w1 and w2 are assigned to the first and second leakage current information. The weight allocation is dynamically adjusted based on environmental conditions, historical detection data, and reliability analysis. For example, under good daylight conditions, the discharge traces in the visible light video are more obvious, thus assigning a higher weight to the first leakage current information, such as w1 and w2 of 0.6 and 0.4, respectively. Conversely, under low light or severe weather conditions, thermal imaging detection is more reliable, leading to a preference for increasing the weight of the second leakage current information, such as w1 and w2 of 0.4 and 0.6, respectively. This weighted calculation method ensures that the final leakage current information accurately reflects the actual condition of the insulator under various operating conditions.

[0095] The calculated leakage current information is the insulator leakage current detection result that integrates visible light and thermal imaging data. This result provides a more comprehensive reflection of the leakage current situation, helping to improve the accuracy and real-time performance of the detection, and providing maintenance personnel with a reliable basis for judging the health status of insulators. Through this weighted calculation method, the detection system can quickly adapt to environmental changes and achieve accurate monitoring and early warning of leakage current, providing an important guarantee for the safe operation of the power system.

[0096] This invention also provides an insulator leakage current detection device based on video surveillance, including a processor and a memory. The memory stores a computer program executable by the processor. When the processor executes the computer program, it performs the insulator leakage current detection method based on video surveillance as described in this invention. Further details can be found in conjunction with the foregoing description. Figure 1 The description of that will not be repeated here.

[0097] Implementing the embodiments of the present invention has the following beneficial effects:

[0098] This invention provides a video monitoring-based method for detecting insulator leakage current. By employing both visible light video and thermal imaging video monitoring, it comprehensively acquires information on discharge traces and temperature anomalies on the insulator surface, significantly improving detection accuracy and avoiding the errors of single monitoring methods. Furthermore, both thermal imaging and visible light detection maintain good stability in complex environments, enhancing the environmental adaptability of the detection method.

[0099] Meanwhile, the video monitoring method enables continuous and real-time detection, which can promptly identify and warn when leakage current increases sharply, thus improving system safety.

[0100] In addition, the flexible deployment of video surveillance expands the monitoring coverage, enabling it to cover remote or difficult-to-maintain areas and reduce blind spots.

[0101] In summary, the solution of this invention achieves automatic classification and calculation of leakage current information, enabling more efficient and accurate leakage current monitoring. It meets the intelligent requirements of power systems and provides comprehensive and reliable technical support for the efficient operation and maintenance of power systems.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0103] Obviously, the embodiments described above are merely some embodiments of the present invention, not all embodiments. The accompanying drawings show preferred embodiments of the present invention, but do not limit the patent scope of the present invention. The present invention can be implemented in many different forms; rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the patent protection scope of this invention.

Claims

1. A video monitoring-based insulator leakage current detection method, characterized in that, The method comprises: Step S1, visible light video detection is performed on the insulator to be detected to obtain a first monitoring video, and thermal imaging video detection is performed on the insulator to obtain a second monitoring video; Step S2, discharge trace size recognition is performed on the first monitoring video to obtain a discharge trace area, and first leakage current information is obtained through classification; Step S3, the second monitoring video is recognized to obtain second leakage current information; Step S4, leakage current information is calculated and obtained according to the first leakage current information and the second leakage current information, which is taken as a leakage current detection result of the insulator; The step S2 further comprises: A discharge trace recognizer is constructed; The first monitoring video is input into the discharge trace recognizer, and a discharge trace area is obtained through recognition output; In the step S2, the discharge trace recognizer is constructed, which comprises: A sample first monitoring video set of the insulator is collected, and a sample discharge trace area set is obtained through identification of discharge trace areas in the sample first monitoring video; A convolutional neural network is used to train the discharge trace recognizer by using the sample first monitoring video set and the sample discharge trace area set, wherein the discharge trace recognizer is trained by using gradient descent in the training process, and the training rule is: ； wherein, is the model parameter at the tth iteration, is the step size, is the gradient; In the step S2, the first leakage current information is obtained through classification, which comprises: According to historical leakage current detection data of the insulator, a sample discharge trace area set and a sample first leakage current information set are collected; A mapping relationship of the sample discharge trace area set and the sample first leakage current information set is constructed to obtain a first leakage current classifier; The discharge trace area is input into the first leakage current classifier to obtain the first leakage current information through classification; The step S3 further comprises: A sample second monitoring video set of the insulator is collected, and a sample second leakage current information set is collected; A convolutional neural network is used to train a second leakage current recognizer by using the sample second monitoring video set and the sample second leakage current information set; The second monitoring video is input into the second leakage current recognizer to obtain the second leakage current information through recognition.

2. The method of claim 1, wherein, The step S1 further comprises: The first monitoring video is obtained through video detection collection of the insulator to be detected under visible light; The second monitoring video is obtained through thermal imaging video collection of the insulator.

3. The method of claim 2, wherein, The step S4 further comprises: The first leakage current information and the second leakage current information are weighted calculated to obtain the leakage current information, which is taken as the leakage current detection result of the insulator.

4. A video monitoring-based insulator leakage current detection device, characterized by, A processor and a memory are included, and the memory stores a computer program that can be executed by the processor; when the processor executes the computer program, the video monitoring based insulator leakage current detection method in any one of claims 1 to 3 is executed.

Images