Arc morphology fractal quantitative characterization method and application thereof

Abstract

The application discloses an arc shape fractal quantitative characterization method, which comprises the following steps: arc image gray processing; image denoising processing; arc image equalization, which divides the arc into a central combustion area and a radiation area; arc shape segmentation, which provides an accurate segmentation threshold in arc image binary processing based on a gray segmentation value, and segments the arc central combustion area and the arc radiation area; image binary processing; edge feature extraction; and adoption of a fractal dimension to quantitatively characterize the arc shape. The application can accurately reflect the complexity and irregularity of the arc shape by image processing to extract arc shape features and calculate the fractal dimension, and can establish a quantitative relationship between the arc shape and the arc power through the characterization of the arc, thereby helping to monitor or evaluate the operation safety and reliability of electrical equipment and systems.

Description

Technical Field

[0001] This invention relates to the field of health monitoring technology for electrical contact components, and in particular to a fractal quantitative characterization method for electric arc morphology and its application. Background Technology

[0002] Electric arcing is widespread in electrical equipment and systems, significantly impacting their stability, lifespan, and safety. For example, in pantograph-catenary systems, the contact pressure between the pantograph and the overhead contact line fluctuates due to vibration and installation errors during train operation. When contact pressure and contact area decrease, contact separation may occur, creating an ionized gas channel and generating an electric arc. The high temperature of the arc causes welding and oxidation on the surfaces of the pantograph and contact wire. At high speeds, shear forces exacerbate oxide shedding, causing severe wear, increasing maintenance costs, and even triggering equipment failure. Furthermore, electric arcs distort current and voltage waveforms, affecting the system's current-carrying capacity and operational reliability. Therefore, arc characterization is fundamental to the health monitoring of electrical contact components.

[0003] Patent document CN113406455A proposes a method for characterizing arc state using arc voltage data. Arc generation is accompanied by voltage fluctuations, but voltage fluctuations do not necessarily indicate the formation of an arc, nor do they directly reflect arc parameters, as contact resistance or load fluctuations can also cause interface voltage changes. In contrast, direct analysis of arc morphology is more accurate. The paper "Research on Arc Morphology Characteristics and Temperature Field of Switchgear for High-Current Vacuum Switches" proposes an arc morphology edge detection method, using wavelet methods to extract arc morphology and enabling observation of the arc formation, combustion, and extinction processes. The paper "Diagnosis of Arc Morphology in Vacuum Switches and Its Model Visualization" proposes using arc area to characterize the arc and studying the dynamic development process of the arc through changes in arc area. The existing technology has the following problems: (1) The brightness of the center of the electric arc is high, and the gray value gradually decreases from the center to the edge. In the process of contour feature extraction, if the threshold is large, the obtained electric arc contour area is small. That is, the electric arc contour extraction is affected by the gray value threshold; (2) The electric arc contour is irregular. The contour details of the images taken by high-speed cameras with different pixels are different. The electric arc contour features are related to the camera pixels; (3) The existing electric arc image processing technology is for the purpose of extracting the electric arc contour and is used to observe the shape of the electric arc. It cannot establish the relationship between the shape and the electric arc energy. The electric arc energy is an important physical quantity that causes surface ablation. Summary of the Invention

[0004] Purpose of the invention: In order to overcome the shortcomings of the prior art, the first objective of this invention is to provide a fractal quantitative characterization method for electric arc morphology. By extracting the characteristics of electric arc morphology through image processing and calculating the fractal dimension, the complexity and irregularity of the electric arc morphology can be accurately reflected. By characterizing the electric arc, a quantitative relationship between the electric arc morphology and the electric arc power can be established, thereby helping to monitor or evaluate the operational safety and reliability of electrical equipment and systems. The second objective is to provide the application of the above-mentioned fractal quantitative characterization method of electric arc morphology in railway pantograph-catenary systems, thereby improving the safety and reliability of railway systems.

[0005] Technical solution: The fractal quantitative characterization method for electric arc morphology disclosed in this invention includes the following steps: S1. Grayscale processing of electric arc image; S2, Image denoising processing; S3. Arc image equalization divides the arc into a central burning area and a radiation area; S4. Arc morphology segmentation: An image processing method based on grayscale segmentation values ​​provides an accurate segmentation threshold in the binarization of arc images, segmenting the arc center combustion zone and the arc radiation zone. S5. Image binarization processing; S6. Edge feature extraction; S7. Fractal dimension is used to quantitatively characterize the electric arc morphology.

[0006] Furthermore, S1 uses the HIS grayscale transformation method and an adaptive threshold function to perform grayscale processing on the arc image; First, the arc image is converted from RGB mode to HSI mode, and the brightness is separated as an image component to form a grayscale image; then, the adaptive grayscale threshold for different arc shapes is calculated using an adaptive threshold function.

[0007] Furthermore, S2 uses wavelet transform threshold denoising to denoise the grayscale image.

[0008] Furthermore, S3 transforms the histogram of the original image into a more uniform distribution, and enhances image contrast by adjusting the dynamic range of pixel grayscale values.

[0009] Furthermore, S4 includes the following steps: S4-1. Obtain grayscale height maps of different arc shapes; S4-2. Set multiple different segmentation grayscale values; S4-3. Use different segmentation gray values ​​to segment the electric arc image, remove pixels with gray values ​​below the gray value, and retain pixels with gray values ​​above the gray value, thereby eliminating messy low gray value pixels in the electric arc radiation area and retaining the burning area in the center of the electric arc. S4-4. Binarize and extract edge features from the segmented arc image; S4-5. Calculate the fractal dimension of the arc edge under different segmentation gray values ​​using the box dimension method. S4-6. Select the gray value with the smallest fractal dimension as the optimal gray value of the central arc-burning region, and select the gray value with the largest fractal dimension as the optimal gray value of the radiation region.

[0010] Furthermore, S5 uses the maximum inter-class variance method, combined with the segmented gray values, to find suitable binarization thresholds for the central burning region and the arc radiation region of the electric arc image, and performs image binarization processing.

[0011] Furthermore, S6 extracts edge features from the binarized image using the Sobel operator.

[0012] Furthermore, S7 includes the following steps: S7-1. Obtain the grayscale contour curve edge of the electric arc image. Cover the grayscale contour curve edge of the electric arc image with boxes of different sizes. Calculate the fractal dimension of the edge. The side length of the box is 𝑟. The total number of boxes required to cover the entire edge is 𝑁𝑟. S7-2. Record different values ​​of ln(r)A and ln(Nr)A, fit a straight line in double logarithmic coordinates, and the opposite of its slope is the fractal dimension of the arc edge. S7-3 Calculate the fractal dimension of different morphological arcs under different segmentation gray values; S7-4. Perform nonlinear fitting on the calculation results; S7-5. Calculate the mean fractal dimension of the central combustion region and the radiation region of electric arcs with different morphologies: S7-6. The arc power is effectively characterized by the shape dimension of the arcing zone, and the arc morphology is effectively characterized by the shape dimension of the radiation zone.

[0013] The above-mentioned fractal quantitative characterization method of electric arc morphology was applied to the railway pantograph-catenary system.

[0014] Beneficial effects: Compared with the prior art, the advantages of the present invention are: (1) By extracting the morphological features of the electric arc and performing fractal quantitative characterization, the energy release process of the electric arc can be reflected more accurately, overcoming the limitations of traditional methods that rely solely on current and voltage measurements; (2) The electric arc image is processed by using techniques such as HIS grayscale transformation, adaptive threshold function, wavelet transform denoising and histogram equalization, which significantly improves the image quality and contrast and enhances the distinction between the arc center burning area and the radiation area. (3) By establishing the relationship between scale and measurement, the fractal dimension is used to characterize the morphological features of the electric arc, which is not affected by the size of the camera pixels and ensures the objectivity of the characterization; (4) By calculating the fractal dimension of different arc shapes under different segmentation gray values, the relationship between arc shape characteristics and arc power can be revealed, which helps to improve the operational safety and reliability of electrical equipment and systems. Attached Figure Description

[0015] Figure 1 Flowchart for extracting arc morphology features; Figure 2 To convert the electric arc image to grayscale; Figure 3 A schematic diagram of image equalization for a blade-shaped electric arc; Figure 4 A schematic diagram of segmenting an electric arc image for different grayscale values; Figure 5 To binarize the electric arc image; Figure 6 This describes the edge feature extraction process using the Sobel gradient operator. Figure 7 This is a schematic diagram of the fractal characteristics of an electric arc; Figure 8 This is a schematic diagram for calculating fractal dimension; Figure 9 Curves showing the relationship between arc power and the morphological dimensions of the arc zone and radiation zone. Detailed Implementation

[0016] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0017] The present invention provides a fractal quantitative characterization method for electric arc morphology, which is specifically divided into two parts: an electric arc morphology feature extraction method and a fractal quantitative characterization method for current morphology.

[0018] Methods for extracting arc morphology features, such as Figure 1 As shown, it includes the following steps: Arc image grayscale processing: The arc image is converted to grayscale using the HIS grayscale transformation method and an adaptive threshold function, such as... Figure 2 As shown.

[0019] First, the arc image is converted from RGB mode to HSI mode, and the brightness (I) is separated as an image component to form a grayscale image. To enhance the grayscale effect, an adaptive threshold function is used to calculate the adaptive grayscale threshold for different arc shapes. The arc shapes are divided into three types: blade-shaped, star-shaped, and comet-tail-shaped, and their threshold ranges are set as follows: blade-shaped arc [0.0039, 1], star-shaped arc [0.11, 0.6], and comet-tail-shaped arc [0.0047, 0.9961].

[0020] Image denoising: Wavelet transform thresholding is used to denoise grayscale images in order to improve image quality.

[0021] Arc image equalization: The histogram of the original image is transformed into a more uniform distribution. By adjusting the dynamic range of pixel grayscale values, image contrast is enhanced, thereby strengthening image feature information. The electric arc is divided into a central combustion region and a radiation region, and the blade-shaped electric arc image is equalized, such as... Figure 3 As shown.

[0022] Arc morphology segmentation: A grayscale segmentation-based image processing method is proposed to provide an accurate segmentation threshold in the binarization of electric arc images, thereby effectively segmenting the central combustion zone and the arc radiation zone. The steps for selecting the optimal grayscale segmentation value are as follows: The first step is to obtain grayscale height maps of different arc shapes, including images of blade-shaped, star-shaped, and comet-tail-shaped arcs.

[0023] The second step, as Figure 4 As shown, nine different grayscale values ​​are set for segmentation: The values ​​are 30, 60, 90, 120, 150, 180, 210, 220, and 225 respectively.

[0024] The third step involves segmenting the arc image using different grayscale values ​​(SG), removing pixels with grayscale values ​​below a certain threshold, and retaining pixels with grayscale values ​​above a certain threshold. This process eliminates cluttered low-grayscale pixels in the arc radiation area while preserving the central burning area of ​​the arc.

[0025] The fourth step is to binarize and extract edge features from the segmented arc image.

[0026] The fifth step is to use the box dimension method to calculate the fractal dimension of the arc edge under different segmentation gray values.

[0027] Step 6: Select the gray value with the smallest fractal dimension as the optimal gray value for the central arc-burning region: The arc radius is 150° for blade-shaped arcs, 180° for star-shaped arcs, and 150° for comet-tail-shaped arcs.

[0028] At the same time, the gray value with the largest fractal dimension is selected as the optimal gray value for the radiation region: The values ​​for blade-shaped arc, star-shaped arc, and comet-tail-shaped arc are all 30.

[0029] Image binarization processing: By using the Otsu's method and combining it with the segmentation of gray values, a suitable binarization threshold is found for the central burning region and the arc radiation region of the electric arc image, and then the image is binarized.

[0030] A threshold is set to divide the image pixels into two parts: those greater than the threshold and those less than the threshold. The pixels greater than the threshold are set to 255, representing the arc itself, while the pixels less than the threshold are set to 0, representing the black background. This facilitates edge extraction of the arc image. Figure 5 As shown.

[0031] Edge feature extraction: The edges of an image are defined by calculating the magnitude of the image gradient. Edge feature extraction is performed on the binarized image. This invention uses the Sobel operator for edge feature extraction of the binarized image. The processing flow is as follows: Figure 6 As shown.

[0032] In the fractal quantitative characterization method of current morphology, fractal features describe the self-similarity of geometric figures, meaning that local structures remain similar to the overall structure even after magnification. The edge contour of an electric arc exhibits this property, such as... Figure 7 As shown in the figure. Therefore, fractal dimension can be used to quantitatively characterize the arc morphology. The specific method is as follows: Obtain the grayscale contour edges of the electric arc image. Cover the grayscale contour edges of the electric arc image with boxes of different sizes, and calculate the fractal dimension of the edges. The side length of the box is 𝑟, and the total number of boxes required to cover the entire edge is 𝑁𝑟.

[0033] Record different ln( r ) A and ln( N r ) A The value is obtained by fitting a straight line in a double logarithmic coordinate system. The negative of the slope of this line is the fractal dimension of the arc edge. Figure 8 As shown.

[0034] Calculate the fractal dimension of different arc shapes under different segmentation gray values.

[0035] The calculation results were nonlinearly fitted using an Allometric function model, y=axb. After five iterations, the convergence condition was met. The fitting relationship is as follows: The fitting equation for the relationship between arc power and the fractal dimension of the combustion region at the arc center is as follows: D z =0.4695 P 0.2165 goodness of fit R1 2 =0.98478; The functional relationship between arc power and the fractal dimension of the radiation region is fitted by the following equation: D r =0.31312 P 0.30201 goodness of fit R² 2 =0.94605.

[0036] Calculate the mean fractal dimension of the central combustion region and the radiation region of blade-shaped, star-shaped, and comet-tail-shaped electric arcs: The mean dimension of the combustion zone at the center of the blade-shaped electric arc is 1.4421, and the mean dimension of the radiation zone is 1.4947. The mean shape dimension of the combustion zone at the center of the star-shaped electric arc is 1.4653, and the mean dimension of the radiation zone is 1.5390. The mean dimension of the combustion zone at the center of the comet-tail-shaped electric arc is 1.5021, and the mean dimension of the radiation zone is 1.5785.

[0037] 6) The relationship between arc power and the morphological dimensions of the arc zone and radiation zone is as follows: Figure 9 As shown, the arc power can be effectively characterized by the morphological dimension of the arc zone, and the arc morphology can be effectively characterized by the morphological dimension of the radiation zone.

[0038] The above-mentioned fractal quantitative characterization method of electric arc morphology is applied to the railway pantograph-catenary system: In railway pantograph-catenary systems, electric arcing is easily generated due to high current and high-speed operation. These arcs are mainly classified into three types: blade-shaped, star-shaped, and comet-tail-shaped. Electric arcs not only affect the normal operation of electrical equipment but can also have a significant impact on the safety and reliability of the entire railway system. The following steps were performed to accurately extract and quantitatively characterize the morphological features of the electric arcs.

[0039] First, during the operation of the railway pantograph-catenary system, high-frequency cameras are used to capture real-time images of the electric arc generated within the system. The acquired arc images are converted from RGB mode to HSI mode, and the luminance (I) component is extracted to generate a grayscale image. Next, wavelet transform threshold denoising is applied to the grayscale image. Subsequently, histogram equalization is performed on the denoised grayscale image to enhance image contrast and strengthen the characteristic information of the arc's central combustion zone and radiation zone.

[0040] Based on this, a grayscale height map of the electric arc is obtained, and nine different segmentation grayscale values ​​are set (e.g., 30, 60, 90, 120, 150, 180, 210, 220, 225). These grayscale values ​​are used for segmentation, removing pixels with grayscale values ​​lower than the selected values ​​and retaining pixels with grayscale values ​​higher than the selected values. Then, the arc edge features are extracted, and the fractal dimension of the arc edge under different segmentation grayscale values ​​is calculated using the box-counting method. The grayscale value with the smallest fractal dimension is selected as the optimal grayscale value for the central arcing region, and the grayscale value with the largest fractal dimension is selected as the optimal grayscale value for the radiation region.

[0041] By combining the Otsu's method and grayscale segmentation, suitable binarization thresholds for the central burning region and the arc radiation region in the electric arc image are found. Image binarization is then performed, dividing pixel values ​​into two parts: those greater than and those less than the threshold, to facilitate edge extraction. The Sobel operator is then used to extract edge features from the binarized image, reducing edge localization errors and discontinuities.

[0042] By utilizing the grayscale contour lines covering the arc image, the fractal dimension of the edges is calculated. A straight line is fitted in a double logarithmic coordinate system, and the inverse of the slope is obtained, which is the fractal dimension. The fractal dimensions of different arc shapes under different segmented grayscale values ​​are calculated, and an Allometric function model is used to perform nonlinear fitting on different fractal dimensions. Finally, the mean fractal dimension of the burner zone in the center of the pantograph-catenary system is calculated to be 1.4421, and the mean in the radiating zone is 1.4947; the mean fractal dimension of the burner zone in the center of the star-shaped arc is 1.4653, and the mean in the radiating zone is 1.5390; the mean fractal dimension of the burner zone in the center of the comet-tail-shaped arc is 1.5021, and the mean in the radiating zone is 1.5785. Through the obtained equation relating arc power and fractal dimension, the mean fractal dimension of the radiating region of different arc shapes differs significantly, allowing for better differentiation of arc shapes. The fractal dimension of the radiating region can effectively characterize the arc shape. The arc power and the fractal dimension of the central arcing zone are consistent, and the arc power can be effectively characterized by the fractal dimension of the arcing zone.

[0043] This application example verifies the effectiveness of the arc morphology feature extraction method and the fractal quantitative characterization method in railway pantograph-catenary systems.

Claims

1. A fractal quantitative characterization method for electric arc morphology, characterized in that, Includes the following steps: S1. Grayscale processing of electric arc image; S2, Image denoising processing; S3. Arc image equalization divides the arc into a central burning area and a radiation area; S4. Arc morphology segmentation: An image processing method based on grayscale segmentation values ​​provides an accurate segmentation threshold in the binarization of arc images, segmenting the arc center combustion zone and the arc radiation zone. S5. Image binarization processing; S6. Edge feature extraction; S7. Use fractal dimension to quantitatively characterize the electric arc morphology; S1 uses the HIS grayscale transformation method and an adaptive threshold function to perform grayscale processing on the electric arc image; First, the arc image is converted from RGB mode to HSI mode, and the brightness is separated as an image component to form a grayscale image; Adaptive grayscale thresholds for different arc morphologies are calculated using an adaptive threshold function. S4 includes the following steps: S4-1. Obtain grayscale height maps of different arc shapes; S4-2. Set multiple different segmentation grayscale values; S4-3. Use different segmentation gray values ​​to segment the electric arc image, remove pixels with gray values ​​below the gray value, and retain pixels with gray values ​​above the gray value, thereby eliminating messy low gray value pixels in the electric arc radiation area and retaining the burning area in the center of the electric arc. S4-4. Binarize and extract edge features from the segmented arc image; S4-5. Calculate the fractal dimension of the arc edge under different segmentation gray values ​​using the box dimension method. S4-6. Select the gray value with the smallest fractal dimension as the optimal gray value of the central arc-burning region, and select the gray value with the largest fractal dimension as the optimal gray value of the radiation region. S7 includes the following steps: S7-1. Obtain the grayscale contour curve edge of the electric arc image. Cover the grayscale contour curve edge of the electric arc image with boxes of different sizes. Calculate the fractal dimension of the edge. The side length of the box is 𝑟. The total number of boxes required to cover the entire edge is 𝑁𝑟. S7-2. Record different values ​​of ln(r)A and ln(Nr)A, fit a straight line in double logarithmic coordinates, and the opposite of its slope is the fractal dimension of the arc edge. S7-3 Calculate the fractal dimension of different morphological arcs under different segmentation gray values; S7-4. Perform nonlinear fitting on the calculation results; S7-5. Calculate the mean fractal dimension of the central combustion region and the radiation region of electric arcs with different morphologies: S7-6. The arc power is effectively characterized by the shape dimension of the arcing zone, and the arc morphology is effectively characterized by the shape dimension of the radiation zone.

2. The fractal quantitative characterization method for electric arc morphology according to claim 1, characterized in that: S2 uses wavelet transform threshold denoising to denoise grayscale images.

3. The fractal quantitative characterization method for electric arc morphology according to claim 1, characterized in that: S3 transforms the histogram of the original image into a more uniform distribution, enhancing image contrast by adjusting the dynamic range of pixel grayscale values.

4. The fractal quantitative characterization method for electric arc morphology according to claim 1, characterized in that: S5 uses the maximum inter-class variance method, combined with the segmentation of gray values, to find the appropriate binarization thresholds for the central burning area and the arc radiation area of ​​the electric arc image, and then performs image binarization processing.

5. The fractal quantitative characterization method for electric arc morphology according to claim 1, characterized in that: S6 uses the Sobel operator to extract edge features from a binarized image.

6. The application of the fractal quantitative characterization method of electric arc morphology as described in claim 1 in railway pantograph-catenary system.

Images