Method and device for determining discharge strength of insulator, computer device, readable storage medium and program product

Abstract

The application relates to the technical field of insulator discharge strength monitoring, and provides a determination method and device of insulator discharge strength, computer equipment, a readable storage medium and a program product. The method comprises the following steps: performing feature extraction on a local discharge ultraviolet image of an insulator to obtain an image feature; according to an image feature, a detection distance corresponding to the local discharge ultraviolet image, a detection gain, a temperature environment parameter, a humidity environment parameter and a pulse current peak prediction model, a pulse current prediction peak value corresponding to the local discharge ultraviolet image is obtained; and according to the relationship between the pulse current prediction peak value and a discharge strength grade, the discharge strength of the insulator is obtained. The method can reduce the prediction error of the discharge strength of the insulator.

Description

Technical Field

[0001] This application relates to the field of insulator discharge intensity monitoring technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for determining insulator discharge intensity. Background Technology

[0002] Insulator discharge typically manifests as partial discharge or flashover, which can lead to insulator aging, breakdown, and power facility failures, thereby causing power outages or safety accidents. In order to ensure the stability and safety of the power system, it is necessary to monitor the discharge intensity of insulators.

[0003] Currently, ultraviolet images of insulators can be obtained using ultraviolet detectors. A regression model based on least squares regression and support vector machines is then established to process and analyze the ultraviolet images, revealing the insulator's discharge status and determining its discharge intensity. However, existing methods for determining insulator discharge intensity do not consider the influence of temperature and humidity environments. In practical applications, the regression models exhibit poor adaptability and robustness, resulting in significant prediction errors in insulator discharge intensity. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for determining the discharge intensity of an insulator, in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides a method for determining the discharge intensity of an insulator, comprising:

[0006] Feature extraction is performed on the ultraviolet images of partial discharge of insulators to obtain image features;

[0007] Based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model corresponding to the partial discharge ultraviolet image, the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image is obtained.

[0008] The discharge intensity of the insulator is obtained based on the relationship between the predicted peak value of the pulse current and the discharge intensity level.

[0009] In one embodiment, the feature extraction of the partial discharge ultraviolet image of the insulator to obtain image features includes:

[0010] Based on the partial discharge ultraviolet image, a spot image is obtained;

[0011] The area of ​​the light spot pixels in the light spot image is obtained based on the pixel values ​​of the contour points in the light spot image.

[0012] The light spot image is subjected to graphic fitting to determine the geometric shape of the light spot in the light spot image;

[0013] Based on the described light spot geometry, the light spot size parameters are obtained;

[0014] Image features are obtained based on the spot pixel area and the spot size parameters.

[0015] In one embodiment, obtaining the spot image based on the partial discharge ultraviolet image includes:

[0016] The partial discharge ultraviolet image is subjected to grayscale processing and filtering noise reduction processing to obtain a noise-reduced grayscale image;

[0017] The denoised grayscale image is binarized to obtain a binarized image;

[0018] The binarized image is traversed according to the edge detection algorithm to obtain the spot image.

[0019] In one embodiment, the step of performing graphic fitting on the spot image to determine the spot geometry of the spot image includes:

[0020] Based on the pixel coordinates of the contour points in the light spot image, an ellipse fitting is performed on the light spot contour in the light spot image to obtain an elliptical light spot contour.

[0021] The eccentricity of the light spot is obtained based on the major and minor axes of the elliptical light spot profile.

[0022] When the eccentricity of the light spot is greater than the threshold, the geometric shape of the light spot in the light spot image is determined to be elliptical.

[0023] When the eccentricity of the light spot is less than a threshold, the maximum inscribed circle of the light spot in the light spot image is fitted to obtain the maximum inscribed circle of the light spot in the light spot image; based on the maximum inscribed circle of the light spot in the light spot image, the circularity of the light spot in the light spot image is obtained; when the eccentricity of the light spot is less than a threshold and the circularity of the light spot is greater than a threshold, the geometric shape of the light spot in the light spot image is determined to be circular.

[0024] In one embodiment, before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance corresponding to the partial discharge ultraviolet image, the detection gain, the temperature environment parameters, the humidity environment parameters, and the pulse current peak prediction model, the method further includes:

[0025] The distance from the detector's optical system to the sensor is obtained; the detector is used to acquire visible light from the insulator; the optical system and the sensor are components of the detector;

[0026] A scaling factor is obtained based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator;

[0027] The detection distance corresponding to the partial discharge ultraviolet image is obtained based on the distance and the scaling factor.

[0028] In one embodiment, before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance corresponding to the partial discharge ultraviolet image, the detection gain, the temperature environment parameters, the humidity environment parameters, and the pulse current peak prediction model, the method further includes:

[0029] Obtain image feature samples, detection distance samples, detection gain samples, temperature environment parameter samples, and humidity environment parameter samples corresponding to the partial discharge ultraviolet image samples to form an initial training set;

[0030] Based on the initial training set, multiple sub-training sets are obtained;

[0031] For any subset of training sets, a first decision tree is constructed and a splitting process is performed at least once; wherein, during each split, a subset of features is randomly selected at each node of the first decision tree, and the optimal feature is selected for splitting according to a set evaluation criterion to obtain each newly formed child node; the splitting process is repeated for each newly formed child node until the stopping condition is met to obtain a second decision tree; the steps of constructing the second decision tree are repeated until a predetermined number of second decision trees is reached.

[0032] Based on the second decision tree corresponding to multiple sub-training sets, a pulse current peak prediction model is obtained.

[0033] Secondly, this application also provides an apparatus for determining the discharge intensity of an insulator, comprising:

[0034] The image feature acquisition module is used to extract features from the ultraviolet images of partial discharge of insulators to obtain image features;

[0035] The peak prediction acquisition module is used to obtain the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model.

[0036] The discharge intensity acquisition module is used to obtain the discharge intensity of the insulator based on the relationship between the predicted peak value of the pulse current and the discharge intensity level.

[0037] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the above-described method.

[0038] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, the computer program being executed by a processor using the methods described above.

[0039] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that is executed by a processor using the methods described above.

[0040] The aforementioned method, apparatus, computer equipment, computer-readable storage medium, and computer program product for determining the discharge intensity of an insulator extract features from the partial discharge ultraviolet image of the insulator to obtain image features; based on the image features, the detection distance, detection gain, temperature environmental parameters, humidity environmental parameters, and pulse current peak prediction model corresponding to the partial discharge ultraviolet image, the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image is obtained; based on the relationship between the predicted peak value of the pulse current and the discharge intensity level, the discharge intensity of the insulator is obtained. This application comprehensively considers multiple factors of insulator discharge, including the image features of the partial discharge ultraviolet image, detection distance, detection gain, temperature environmental parameters, and humidity environmental parameters, performs feature mapping on multi-dimensional factors, and combines them with the pulse current peak prediction model to comprehensively reflect the discharge status of the insulator, reduce the error caused by a single discharge factor, and reduce the prediction error of the insulator discharge intensity. Attached Figure Description

[0041] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0042] Figure 1 This is a diagram illustrating the application environment of a method for determining the discharge intensity of an insulator in one embodiment.

[0043] Figure 2 This is a flowchart illustrating a method for determining the discharge intensity of an insulator in one embodiment;

[0044] Figure 3 This is a schematic diagram illustrating the process of obtaining image features in one embodiment;

[0045] Figure 4 This is a structural block diagram of a device for determining the discharge intensity of an insulator in one embodiment;

[0046] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0048] This application provides a method for determining the discharge intensity of an insulator. This embodiment can be executed by a computer device, such as... Figure 1 As shown, the computer device can acquire a partial discharge ultraviolet image of an insulator, as well as the corresponding detection distance, detection gain, temperature environmental parameters, and humidity environmental parameters, thereby obtaining the discharge intensity of the insulator. It is understood that this computer device can be implemented through a server, a terminal, or an interactive system between a terminal and a server. In this embodiment, the method includes... Figure 2 The steps shown are as follows:

[0049] Step S201: Extract features from the partial discharge ultraviolet image of the insulator to obtain image features.

[0050] Insulators generate ultraviolet (UV) radiation during discharge, especially during arcing and partial discharge, where the UV signal is significant and intense. A dual-channel UV / visible detector can be used to acquire visible light images and partial discharge UV images of the insulator. Based on the acquired visible light images, the detection distance can be estimated using image processing and visual geometry principles. Feature extraction can be performed on the partial discharge UV images of the insulator, segmenting the light spots in the discharge region to obtain image features. Simultaneously, the detector's detection gain, as well as temperature and humidity environmental parameters, can be acquired.

[0051] Step S202: Based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model corresponding to the partial discharge ultraviolet image, the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image is obtained.

[0052] The detection range is the distance between the target insulator and the detector (monocular ranging method), which can be represented by the parameter d. The detection gain is the detection gain of the ultraviolet detector (imager), which can be represented by the parameter g. The peak pulse current is the peak pulse current generated by partial discharge (corona discharge, spark, or arc) of the insulator. During the corona discharge stage, the pulse signal is a narrow pulse; during the arc discharge stage, the pulse signal is a wide pulse. Furthermore, the peak pulse signal changes positively with increasing discharge time and discharge intensity (discharge quantity), characterizing the discharge intensity. The temperature and humidity environmental parameters represent the ambient temperature and humidity over a specific time period, which can be represented by time(T, H), where T represents temperature and H represents humidity.

[0053] The pulse current peak prediction model can comprehensively analyze image features, detection distance, detection gain, temperature environment parameters, and humidity environment parameters corresponding to the partial discharge ultraviolet image to predict the pulse current peak corresponding to the partial discharge ultraviolet image.

[0054] The area of ​​the light spot pixels S in the image features, the detection distance d corresponding to the partial discharge ultraviolet image, the detection gain g, the temperature environment parameter and the humidity environment parameter time(T,H) can be input into the pulse current peak prediction model f, i.e., y=f(d,g,S,time(T,H)), to obtain the pulse current prediction peak y corresponding to the partial discharge ultraviolet image.

[0055] Step S203: Based on the relationship between the predicted peak value of the pulse current and the discharge intensity level, the discharge intensity of the insulator is obtained.

[0056] Based on the arc and degree of danger generated by partial discharge of insulators, a correspondence between the peak value of pulse current and the discharge intensity level can be established. That is, based on the impact of the corona and arc generated by the discharge of insulators on the safety of the electrical equipment of the insulators, the peak value of pulse current at the corresponding stage can be divided into different discharge intensity levels (danger levels).

[0057] For example, when the peak pulse current is within the first set range [Int1~Int2), it indicates the start of corona discharge, corresponding to hazard level 1, and the insulator's ultraviolet discharge intensity level output is level 1; when the peak pulse current is within the second set range [Int2~Int3), the peak discharge current further increases, indicating the start of arcing, which poses a danger to the safety of the insulator equipment, corresponding to hazard level 2, and the insulator's ultraviolet discharge intensity level output is level 2; and when the peak pulse current exceeds Int3, it indicates that the current insulator discharge intensity is too high, causing serious safety hazards, requiring immediate intervention and handling, corresponding to hazard level 3, and the insulator's ultraviolet discharge intensity level output is level 3.

[0058] In the event of insulator discharge, the discharge intensity level and hazard level increase accordingly with the continuous increase and peak value of the pulse current. Especially when the insulator is under high discharge intensity, there is a non-linear relationship between the peak pulse current and the discharge intensity level. For example, the discharge intensity level increases exponentially with the peak pulse current.

[0059] The discharge intensity level corresponding to the predicted peak value of the pulse current can be obtained based on the relationship between the predicted peak value of the pulse current and the discharge intensity level, thereby determining the discharge intensity of the insulator.

[0060] The aforementioned method for determining insulator discharge intensity comprehensively considers multiple factors related to insulator discharge, including image features of partial discharge ultraviolet images, detection distance, detection gain, temperature environmental parameters, and humidity environmental parameters. It performs feature mapping on these multi-dimensional factors and combines this with a pulse current peak prediction model to comprehensively reflect the insulator discharge status, reducing errors caused by single discharge factors and minimizing the prediction error of insulator discharge intensity. This provides strong support for power system fault early warning, helps to promptly detect insulator discharge faults, prevents power safety accidents, ensures the stable and reliable operation of the power grid, and provides important technical means for the health management and fault prevention of power equipment. Furthermore, the insulator discharge intensity determination method of this application is a non-contact insulator discharge intensity detection method, which does not require direct contact with power equipment, ensuring operational safety. It is particularly suitable for monitoring high-voltage equipment and supports real-time online monitoring, real-time acquisition of equipment operating status, timely detection and location of insulator discharge faults, and early warning, improving the safety and reliability of equipment operation.

[0061] In one embodiment, feature extraction is performed on the partial discharge ultraviolet image of the insulator to obtain image features. Specific steps are as follows: Figure 3 As shown: Step S301, obtain a spot image based on the partial discharge ultraviolet image; Step S302, obtain the spot pixel area of ​​the spot image based on the pixel values ​​of the contour points in the spot image; Step S303, perform graphic fitting on the spot image to determine the spot geometry; Step S304, obtain the spot size parameters based on the spot geometry; Step S305, obtain image features based on the spot pixel area and spot size parameters.

[0062] Image processing can be performed on the ultraviolet images of partial discharge to obtain the spot image of the discharge area.

[0063] The spot area in the spot image is white, and the pixel value of the white area is set to 1. The other areas outside the white area are black, and the pixel value of the black area is set to 0. The number of pixels with a pixel value of 1 in the spot image is counted, which is taken as the spot pixel area of ​​the spot image.

[0064] Based on statistical analysis of the shape of light spots in partial discharge ultraviolet images in actual practice, it is known that the shape of light spots in partial discharge ultraviolet images is generally circular or elliptical. Therefore, ellipse fitting or circular fitting can be performed on the light spot image, and the geometric shape of the light spot image is determined to be close to a circle or an ellipse based on the light spot eccentricity (eccentricity) and the light spot circularity.

[0065] When the light spot geometry is elliptical, the major axis and minor axis of the ellipse are used as the light spot size parameters Mx and My (XY axis size parameters); when the light spot geometry is circular, the diameter of the circle is used as the light spot size parameters Mx and My.

[0066] The pixel area and size of the light spot are used as image features.

[0067] In this embodiment, a spot image is obtained from a partial discharge ultraviolet image; the spot pixel area and spot size parameters are obtained from the spot image, thereby obtaining more accurate image features.

[0068] In one embodiment, a spot image is obtained from a partial discharge ultraviolet image. The specific steps are as follows: grayscale processing and filtering noise reduction processing are performed on the partial discharge ultraviolet image to obtain a noise-reduced grayscale image; binarization processing is performed on the noise-reduced grayscale image to obtain a binarized image; the binarized image is traversed according to an edge detection algorithm to obtain a spot image.

[0069] Partial discharge ultraviolet images can be processed for grayscale and filtered for noise reduction to obtain a denoised grayscale image. The filtering and noise reduction process can employ median filtering, mean filtering, or dynamic threshold filtering (adaptive filtering). Taking median filtering as an example, median filtering is applied to randomly scattered noise points and scattering points at the edge of the spot in the partial discharge ultraviolet image to eliminate noise. The filter template size can be set, typically 3*3 or 5*5. Median filtering is performed on each pixel of the partial discharge ultraviolet image according to the filter template. The neighborhood of a pixel corresponds to multiple pixels (8 pixels for a 3*3 template), eliminating salt-and-pepper noise points (impulse noise points) while preserving the contour edge information of the spot.

[0070] The denoised grayscale image is binarized to obtain a binarized image; the binarized image is then traversed using an edge detection algorithm such as the Canny detection algorithm to extract the light spots in the discharge area, thus obtaining a light spot image.

[0071] In this embodiment, the partial discharge ultraviolet image is processed to obtain a high-precision spot image.

[0072] In one embodiment, a pattern fitting is performed on the spot image to determine the spot geometry. The specific steps are as follows: based on the pixel coordinates of the contour points of the spot image, an ellipse fitting is performed on the spot contour in the spot image to obtain an elliptical spot contour; based on the major and minor axes of the elliptical spot contour, the spot eccentricity is obtained; when the spot eccentricity is greater than a threshold, the spot geometry of the spot image is determined to be elliptical; when the spot eccentricity is less than a threshold, the maximum inscribed circle fitting is performed on the spot in the spot image to obtain the maximum inscribed circle of the spot image; based on the maximum inscribed circle of the spot image, the spot circularity is obtained; when the spot eccentricity is less than a threshold and the spot circularity is greater than a threshold, the spot geometry of the spot image is determined to be circular.

[0073] Based on the pixel coordinates of the contour points in the spot image, Principal Component Analysis (PCA) or ellipse fitting algorithms can be used to fit the spot contour to an ellipse. For example, based on the least squares method, the parameters of the ellipse are determined by minimizing the sum of squares of the distances from the contour points to the fitted ellipse, thus obtaining an elliptical spot contour. Alternatively, ellipse fitting can be performed using the Hough transform method or the Open Source Computer Vision Library (OpenCV) to obtain an elliptical spot contour.

[0074] Based on the major axis a and minor axis b of the elliptical light spot profile, the eccentricity E of the light spot is obtained as shown in equation (1).

[0075] E=sqrt(1-(b² / a²)) (1)

[0076] The closer the eccentricity E of the light spot is to 0, the closer the geometric shape of the light spot in the image is to a circle; the closer the eccentricity E of the light spot is to 1, the closer the geometric shape of the light spot in the image is to an ellipse.

[0077] A threshold Ev can be preset according to the actual situation. When the spot eccentricity E is greater than the threshold Ev, the spot geometry of the spot image is determined to be elliptical.

[0078] When the eccentricity E of the light spot is less than the threshold Ev, the maximum inscribed circle of the light spot in the light spot image is fitted. By traversing all pixels in the light spot image (considering each pixel as a candidate point for the center of the circle), a pixel is found that minimizes the sum of the distances from the pixel to all contour points. This pixel is then used as the center of the circle, and the maximum inscribed circle of the light spot in the light spot image is obtained by calculating the maximum inscribed circle of the triangle.

[0079] The circularity C of the light spot image can be obtained based on the area F of the largest inscribed circle of the light spot and the maximum distance lmax from the center point of the light spot image to the contour point, as shown in Equation (2).

[0080] C = F / (π * lmax) 2 (2)

[0081] The closer the circularity C of the light spot is to 1, the closer the geometric shape of the light spot in the image is to a circle.

[0082] When the eccentricity E of the light spot is less than the threshold Ev and the circularity C of the light spot is greater than the threshold Ev, the geometric shape of the light spot in the light spot image is determined to be circular.

[0083] In this embodiment, the geometric shape of the light spot image is determined to be circular or elliptical based on the light spot eccentricity and light spot circularity, which can more accurately determine the geometric shape of the light spot image.

[0084] In one embodiment, before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on image features, the detection distance corresponding to the partial discharge ultraviolet image, the detection gain, temperature environment parameters, humidity environment parameters, and the pulse current peak prediction model, the method provided in this application further includes: obtaining the distance from the optical system of the detector to the sensor; the detector is used to obtain the visible light image of the insulator; the optical system and the sensor are components of the detector; a scaling factor is obtained based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator; and the detection distance corresponding to the partial discharge ultraviolet image is obtained based on the distance and the scaling factor.

[0085] A dual-channel ultraviolet / visible light detector can be used to acquire visible light images and partial discharge ultraviolet images of the insulator. The detection distance can be calculated by using the actual size of the insulator in the visible light image. For example, when the detector camera captures an image of the insulator, there is a similar relationship between the actual height H of the insulator and its pixel height h in the visible light image. This height can be determined using an open computer vision library, and the pixel value of that height can be obtained and used as the pixel height h of the insulator in the visible light image.

[0086] The detector can be viewed as a projection device, and the actual height of the insulator is proportional to the height of its projection in the image. Using a virtual triangle as an analogy, the base of the triangle represents the distance D from the detector to the insulator, and the height of the triangle is the actual height H of the insulator. The optical system inside the detector projects this triangle onto the image sensor within the detector, forming a new triangle. The base of this new triangle is the distance from the optical system to the sensor (considered as the focal length f, which is known), and the height of this new triangle is the pixel height h of the insulator in the visible light image. Using the principle of perspective, and given the angle between the detector and the insulator, the distance D from the detector to the insulator can be obtained based on this proportional relationship.

[0087] Specifically, the distance f from the detector's optical system to the sensor can be obtained; based on the pixel height h occupied by the insulator in the visible light image and the actual height H of the insulator, the scaling factor h / H can be obtained; based on the distance f and the scaling factor, the detection distance corresponding to the partial discharge ultraviolet image can be obtained as D=fH / h.

[0088] In this embodiment, a scaling factor can be obtained based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator; the detection distance corresponding to the partial discharge ultraviolet image can be obtained quickly and accurately based on the distance from the detector's optical system to the sensor and the scaling factor.

[0089] In one embodiment, before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on image features, the detection distance corresponding to the partial discharge ultraviolet image, the detection gain, temperature environment parameters, humidity environment parameters, and the pulse current peak prediction model, the method provided in this application further includes: acquiring a visible light image; obtaining a depth map corresponding to the visible light image based on the visible light image and a deep learning model; and obtaining the detection distance corresponding to the partial discharge ultraviolet image based on the pixel values ​​of the depth map.

[0090] It can acquire visible light images and partial discharge ultraviolet images of insulators based on a dual-channel ultraviolet / visible light detector.

[0091] Visible light images can be input into a deep learning model to obtain the corresponding depth map. The deep learning model can be a convolutional neural network model, a monocular depth estimation model, or a depth prediction transformer (DPT). The deep learning model can be pre-trained using a labeled visible light image dataset.

[0092] The total pixel value of the depth map can be obtained from the pixel values ​​of the depth map.

[0093] The pixel values ​​in the depth map typically represent the distance from the camera to each pixel in the scene. Therefore, the total pixel values ​​of the depth map can be used as the detection distance corresponding to the partial discharge ultraviolet image.

[0094] In this embodiment, a depth map is obtained based on the visible light image and the deep learning model; based on the pixel values ​​of the depth map, the detection distance corresponding to the partial discharge ultraviolet image is obtained quickly and accurately.

[0095] In one embodiment, before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on image features, detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model, the method provided in this application further includes: acquiring image feature samples, detection distance samples, detection gain samples, temperature environment parameter samples, and humidity environment parameter samples corresponding to the partial discharge ultraviolet image samples to form an initial training set; obtaining multiple sub-training sets based on the initial training set; constructing a first decision tree for any sub-training set and performing at least one splitting process; wherein, during each split, a feature subset is randomly selected at each node of the first decision tree, and the optimal feature is selected for splitting according to a set evaluation criterion to obtain each newly formed sub-node; repeating the splitting process for each newly formed sub-node until the stopping condition is met to obtain a second decision tree; repeating the step of constructing the second decision tree until a predetermined number of second decision trees is reached; and obtaining the pulse current peak prediction model based on the second decision trees corresponding to the multiple sub-training sets.

[0096] The system can acquire image feature samples, detection distance samples, detection gain samples, temperature environmental parameter samples, and humidity environmental parameter samples corresponding to partial discharge ultraviolet image samples to form an initial training set. The initial training set contains N samples, each of which includes one image feature sample, one detection distance sample, one detection gain sample, one temperature environmental parameter sample, and one humidity environmental parameter sample corresponding to a partial discharge ultraviolet image sample.

[0097] Bootstrap sampling can be used to randomly select a sample from the initial training set and add it to a subset of the training set. This sampling process with replacement can be repeated multiple times (e.g., 100 times), generating a different subset each time, resulting in multiple subsets. Because sampling is done with replacement, some samples may be selected multiple times, while others may not be selected at all.

[0098] For multiple sub-training sets, a random forest regression model can be used for training. By constructing multiple decision trees (regression trees) and integrating the prediction results of multiple decision trees to perform the regression task, a pulse current peak prediction model can be obtained.

[0099] A random forest regression model can be built using the RandomForestRegressor class in the Scikit-learn library. The parameters of the random forest regression model can be set, including the number of decision trees, the maximum depth of the decision trees, the minimum number of samples required for a node split, the minimum number of samples required for a leaf node, and the number of features required for a node split. The number of decision trees refers to the different branches of the tree. The maximum depth of the decision trees controls the depth of decision tree generation. The minimum number of samples required for a node split controls node splitting; if the number of samples for a node is less than this value, no further splitting will occur.

[0100] For any subset of training data, initialize an empty first decision tree and perform at least one split. During each split, randomly select a feature subset at each node of the first decision tree. For example, if the total number of feature subsets is 8, randomly select 3-4 feature subsets. Based on a set evaluation criterion, such as minimizing the mean squared error, select the optimal feature for splitting, resulting in a newly formed child node. For example, divide the feature subset corresponding to the optimal feature into two feature subsets, which form a new child node. Repeat the splitting process for each newly formed child node until a stopping condition is met, resulting in a second decision tree. The stopping condition can be that the number of samples in a node is less than a set minimum number of samples, the node reaches a preset maximum tree depth, or the impurity of the node decreases to a set purity threshold. Repeat the steps of constructing the second decision tree until the predetermined number of second decision trees is reached.

[0101] Based on the second decision trees corresponding to multiple sub-training sets, a pulse current peak prediction model is obtained. Each second decision tree provides a predicted value for the target variable of the observed sample. The prediction result of the pulse current peak prediction model is the aggregation of the prediction results of all second decision trees for each sample, including the average or weighted average. After obtaining the pulse current peak prediction model, the validation set derived from the initial training set can be used to optimize and improve the pulse current peak prediction model, further refining its parameters.

[0102] In this embodiment, samples are randomly selected from the initial training set with replacement to construct a sub-training set for each decision tree. This makes the samples random and allows for a certain degree of repetition, ensuring the diversity of the sub-training set. This significantly improves the resolution of the feature space, forms a more accurate and smooth decision boundary, and increases the differences between different second decision trees.

[0103] For any given subset of training data, a predetermined number of second decision trees are constructed. Based on the second decision trees corresponding to multiple subsets of training data, a high-performance pulse current peak prediction model is obtained. Each second decision tree is trained independently on randomly selected subsamples, resulting in diverse sample types and high tolerance for noise, reducing the risk of overfitting and improving the robustness and accuracy of the pulse current peak prediction model.

[0104] Furthermore, this embodiment does not require a supervision mechanism when processing large-scale initial training sets, especially for nonlinear data, such as image feature samples, detection distance samples, detection gain samples, temperature environmental parameter samples, and humidity environmental parameter samples in this embodiment. When making decision-making features, a small number of randomly selected features are considered, and the best splitting features are recursively selected to reduce redundancy between features, further increase the differences between the second decision trees, and improve the generalization ability, prediction ability, and performance of the pulse current peak prediction model.

[0105] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0106] Based on the same inventive concept, this application also provides an insulator discharge intensity determination device for implementing the above-described method for determining insulator discharge intensity. The solution provided by this device is similar to the solution described in the above-described method. Therefore, the specific limitations in one or more embodiments of the insulator discharge intensity determination device provided below can be found in the limitations of the insulator discharge intensity determination method described above, and will not be repeated here.

[0107] In one exemplary embodiment, such as Figure 4 As shown, a device for determining the discharge intensity of an insulator is provided, wherein:

[0108] The image feature acquisition module 401 is used to extract features from the partial discharge ultraviolet image of the insulator to obtain image features;

[0109] The peak prediction acquisition module 402 is used to obtain the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters and pulse current peak prediction model.

[0110] The discharge intensity acquisition module 403 is used to obtain the discharge intensity of the insulator based on the relationship between the predicted peak value of the pulse current and the discharge intensity level.

[0111] In one embodiment, the image feature acquisition module 401 is further configured to: obtain a spot image based on the partial discharge ultraviolet image; obtain the spot pixel area of ​​the spot image based on the contour point pixel values ​​in the spot image; perform graphic fitting on the spot image to determine the spot geometry of the spot image; obtain spot size parameters based on the spot geometry; and obtain image features based on the spot pixel area and the spot size parameters.

[0112] In one embodiment, the image feature acquisition module 401 is further configured to: perform grayscale processing and filtering noise reduction processing on the partial discharge ultraviolet image to obtain a noise-reduced grayscale image; perform binarization processing on the noise-reduced grayscale image to obtain a binarized image; and traverse the binarized image according to an edge detection algorithm to obtain the spot image.

[0113] In one embodiment, the image feature acquisition module 401 is further configured to: perform elliptical fitting on the light spot contour in the light spot image based on the pixel coordinates of the contour points of the light spot image to obtain an elliptical light spot contour; obtain the light spot eccentricity based on the major axis and minor axis of the elliptical light spot contour; when the light spot eccentricity is greater than a threshold, determine the light spot geometry of the light spot image as elliptical; when the light spot eccentricity is less than a threshold, perform maximum incircle fitting on the light spot in the light spot image to obtain the maximum incircle of the light spot image; obtain the light spot circularity of the light spot image based on the maximum incircle of the light spot image; when the light spot eccentricity is less than a threshold and the light spot circularity is greater than a threshold, determine the light spot geometry of the light spot image as circular.

[0114] In one embodiment, the device further includes a detection distance acquisition module, configured to: acquire the distance from the optical system of the detector to the sensor; the detector is used to acquire a visible light image of the insulator; the optical system and the sensor are components of the detector; obtain a scaling factor based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator; and obtain the detection distance corresponding to the partial discharge ultraviolet image based on the distance and the scaling factor.

[0115] In one embodiment, the device further includes a detection distance acquisition module, configured to: acquire image feature samples, detection distance samples, detection gain samples, temperature environment parameter samples, and humidity environment parameter samples corresponding to partial discharge ultraviolet image samples, forming an initial training set; obtain multiple sub-training sets based on the initial training set; construct a first decision tree for any sub-training set and perform at least one splitting process; wherein, during each split, a feature subset is randomly selected at each node of the first decision tree, and the optimal feature is selected for splitting according to a set evaluation criterion, resulting in each newly formed sub-node; repeat the splitting process for each newly formed sub-node until a stopping condition is met, resulting in a second decision tree; repeat the steps of constructing the second decision tree until a predetermined number of second decision trees is reached; and obtain a pulse current peak prediction model based on the second decision trees corresponding to the multiple sub-training sets.

[0116] Each module in the aforementioned device for determining the discharge intensity of insulators can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0117] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores data for embodiments of the method for determining insulator discharge intensity. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for determining insulator discharge intensity.

[0118] Those skilled in the art will understand that Figure 5The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0119] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0120] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0121] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0122] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0123] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0124] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0125] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining the discharge intensity of an insulator, characterized in that, The method includes: The spot image is obtained based on the partial discharge ultraviolet image of the insulator; The area of ​​the light spot pixels in the light spot image is obtained based on the pixel values ​​of the contour points in the light spot image. Based on the eccentricity and circularity of the light spot, the light spot image is subjected to graphic fitting to determine the geometric shape of the light spot in the light spot image; Based on the described light spot geometry, the light spot size parameters are obtained; Image features are obtained based on the spot pixel area and the spot size parameters; The distance from the detector's optical system to the sensor is obtained; the detector is used to acquire a visible light image of the insulator; the optical system and the sensor are components of the detector; A scaling factor is obtained based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator; The detection distance corresponding to the partial discharge ultraviolet image is obtained based on the distance and the scaling factor. Based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model corresponding to the partial discharge ultraviolet image, the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image is obtained. The discharge intensity of the insulator is obtained based on the relationship between the predicted peak value of the pulse current and the discharge intensity level.

2. The method according to claim 1, characterized in that, The process of obtaining the spot image based on the partial discharge ultraviolet image of the insulator includes: The partial discharge ultraviolet image of the insulator is processed by grayscale and filtered for noise reduction to obtain a noise-reduced grayscale image. The denoised grayscale image is binarized to obtain a binarized image; The binarized image is traversed according to the edge detection algorithm to obtain the spot image.

3. The method according to claim 1, characterized in that, The step of performing graphic fitting on the light spot image based on the light spot eccentricity and light spot circularity to determine the light spot geometry includes: Based on the pixel coordinates of the contour points in the light spot image, an ellipse fitting is performed on the light spot contour in the light spot image to obtain an elliptical light spot contour. The eccentricity of the light spot is obtained based on the major and minor axes of the elliptical light spot profile. When the eccentricity of the light spot is greater than the threshold, the geometric shape of the light spot in the light spot image is determined to be elliptical. When the eccentricity of the light spot is less than a threshold, the maximum inscribed circle of the light spot in the light spot image is fitted to obtain the maximum inscribed circle of the light spot in the light spot image; based on the maximum inscribed circle of the light spot in the light spot image, the circularity of the light spot in the light spot image is obtained; when the eccentricity of the light spot is less than a threshold and the circularity of the light spot is greater than a threshold, the geometric shape of the light spot in the light spot image is determined to be circular.

4. The method according to claim 1, characterized in that, Before obtaining the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and the pulse current peak prediction model, the method further includes: Obtain image feature samples, detection distance samples, detection gain samples, temperature environment parameter samples, and humidity environment parameter samples corresponding to the partial discharge ultraviolet image samples to form an initial training set; Based on the initial training set, multiple sub-training sets are obtained; For any subset of training sets, a first decision tree is constructed and a splitting process is performed at least once; wherein, during each split, a subset of features is randomly selected at each node of the first decision tree, and the optimal feature is selected for splitting according to a set evaluation criterion to obtain each newly formed child node; the splitting process is repeated for each newly formed child node until the stopping condition is met to obtain a second decision tree; the steps of constructing the second decision tree are repeated until a predetermined number of second decision trees is reached. Based on the second decision tree corresponding to multiple sub-training sets, a pulse current peak prediction model is obtained.

5. A device for determining the discharge intensity of an insulator, characterized in that, The device includes: The image feature acquisition module is used to obtain a spot image based on the partial discharge ultraviolet image of the insulator; obtain the spot pixel area based on the outline point pixel values ​​in the spot image; perform graphic fitting on the spot image based on the spot eccentricity and spot circularity to determine the spot geometry; obtain spot size parameters based on the spot geometry; obtain image features based on the spot pixel area and the spot size parameters; acquire the distance from the detector's optical system to the sensor; the detector is used to acquire the visible light image of the insulator; the optical system and the sensor are components of the detector; obtain a scaling factor based on the pixel height occupied by the insulator in the visible light image and the actual height of the insulator; and obtain the detection distance corresponding to the partial discharge ultraviolet image based on the distance and the scaling factor. The peak prediction acquisition module is used to obtain the predicted peak value of the pulse current corresponding to the partial discharge ultraviolet image based on the image features, the detection distance, detection gain, temperature environment parameters, humidity environment parameters, and pulse current peak prediction model. The discharge intensity acquisition module is used to obtain the discharge intensity of the insulator based on the relationship between the predicted peak value of the pulse current and the discharge intensity level.

6. The apparatus according to claim 5, characterized in that, The image feature acquisition module is also used to perform grayscale processing and filtering noise reduction processing on the partial discharge ultraviolet image of the insulator to obtain a noise-reduced grayscale image; and to perform binarization processing on the noise-reduced grayscale image to obtain a binarized image. The binarized image is traversed according to the edge detection algorithm to obtain the spot image.

7. The apparatus according to claim 5, characterized in that, The image feature acquisition module is further configured to: fit an ellipse to the contour of the light spot image based on the pixel coordinates of the contour points of the light spot image to obtain an elliptical light spot contour; obtain the eccentricity of the light spot based on the major and minor axes of the elliptical light spot contour; determine the geometric shape of the light spot image as elliptical when the eccentricity of the light spot is greater than a threshold; fit the maximum inscribed circle of the light spot in the light spot image to obtain the maximum inscribed circle of the light spot image when the eccentricity of the light spot is less than a threshold; obtain the circularity of the light spot image based on the maximum inscribed circle of the light spot image; and determine the geometric shape of the light spot image as circular when the eccentricity of the light spot is less than a threshold and the circularity of the light spot is greater than a threshold.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images