Method, system and device for predicting breakdown field strength of nano-modified cellulose insulation paper
By using Morishita index, skewness value, and minimum neighborhood distance as indicators of nanoparticle dispersion, and combining them with a multiple linear regression model, the problem of difficulty in distinguishing the degree of nanoparticle dispersion was solved, enabling accurate prediction of breakdown field strength and improving the performance optimization of cellulose insulating paper.
Patent Information
- Application Number
- CN202511192362.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-25
- Publication Date
- 2025-11-07
AI Technical Summary
Existing technologies cannot effectively distinguish the degree of dispersion of nanoparticles in cellulose insulating paper, resulting in insufficient accuracy in predicting breakdown field strength.
The Morishita index, skewness value, and minimum neighborhood distance were used as indicators of nanoparticle dispersion. Combined with a multiple linear regression model, the relationship between nanoparticle dispersion and breakdown field strength was established through image processing and classification recognition technology.
It enables quantitative assessment of nanoparticle dispersion and accurate prediction of breakdown field strength, thereby improving the data-driven decision-making capability for optimizing the performance of cellulose insulating paper.
Smart Images
Figure CN120908232A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of nanocomposite characterization and modification of cellulose insulating paper for power equipment, and particularly relates to a breakdown field strength prediction method, system and device for nanomodified cellulose insulating paper. BACKGROUND
[0002] Oil-immersed power transformers are one of the key equipment in the power system, and their reliability is crucial to the safe and stable operation of the power system. Cellulose insulating paper, as the main solid insulating material inside the transformer, cannot be replaced during use, and its electrical properties directly determine the service life of the transformer. At the same time, the rapid development of ultra-high voltage technology puts forward higher requirements for insulating materials. Therefore, developing high-performance modified cellulose insulating paper is of great significance for prolonging the service life of the transformer and ensuring the stable operation of power equipment.
[0003] In recent years, the use of inorganic nanoparticle doping technology to improve the electrical properties of cellulose insulating paper has become a research hotspot. However, due to the large specific surface area and high surface energy of nanoparticles, they tend to agglomerate, leading to deterioration of the performance of modified insulating paper. Therefore, it is necessary to characterize the dispersion of nanoparticles in nanomodified cellulose insulating paper.
[0004] Currently, a simple statistical method based on image binarization is commonly used to perform threshold segmentation on images taken under scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The dispersion degree of nanoparticles is determined based on the segmentation results. However, this method cannot effectively distinguish nanoparticles with similar dispersion degrees, leading to unclear relationships between nanoparticle dispersion and the breakdown field strength of modified insulating paper, thereby reducing the prediction accuracy of the breakdown field strength of nanomodified cellulose insulating paper. SUMMARY
[0005] To address the problem that existing technologies cannot effectively distinguish nanoparticles with similar dispersion degrees, leading to unclear relationships between nanoparticle dispersion and the breakdown field strength of modified insulating paper, thereby reducing the prediction accuracy of the breakdown field strength of insulating paper, the present application proposes a breakdown field strength prediction method, system and device for nanomodified cellulose insulating paper, which uses the Morishita index, skewness and minimum neighborhood distance as nanoparticle dispersion indicators to effectively distinguish the random dispersion and agglomeration state of nanoparticles, thereby solving the problems existing in the prior art.
[0006] A breakdown field strength prediction method for nanomodified cellulose insulating paper, comprising the following steps: Obtaining an electron microscope image of a nanomodified cellulose insulating paper sample; The nanoparticles and the cellulose matrix in the electron microscope image are classified and recognized to generate a classification result image, the classification result image is binarized, the adhered nanoparticles are segmented, and the total number and coordinates of all the nanoparticles in the electron microscope image are counted; the Moriwaki index, skewness and minimum neighborhood distance for representing the dispersion degree of the nanoparticles are calculated as the nanoparticle dispersion index according to the total number and coordinates of the nanoparticles; According to the test value of the breakdown field strength of the nanomodified cellulose insulation paper sample, a multiple linear regression model between the nanoparticle dispersion index of the sample and the breakdown field strength of the insulation paper is established; The nanoparticle dispersion index of the nanomodified cellulose insulation paper sample to be tested is input into the multiple linear regression model to predict the breakdown field strength of the nanomodified cellulose insulation paper to be tested.
[0007] Further, the classification and recognition of the nanoparticles and the cellulose matrix in the electron microscope image to generate a classification result image specifically includes the following steps: A Trainable Weka Segmentation plug-in in the Fiji software is used to create two types of labels; The brush tool is used to label the nanoparticle regions and the texture regions of the cellulose matrix in the nanomodified cellulose insulation paper sample, which are classified into the two types of labels respectively, and the two types of labels after classification are used to train the random forest classifier, and the classification of the nanoparticles and the cellulose matrix is realized through multi-scale texture feature extraction and edge detection feature; The trained random forest classifier is used to classify and recognize the nanoparticles and the cellulose matrix in the electron microscope image to generate a classification result image.
[0008] Further, the classification result image after binarization is processed by using a watershed algorithm to segment the adhered nanoparticles.
[0009] Further, it further includes that in the process of segmenting the adhered nanoparticles, the internal cavities of the nanoparticle regions in the classification result image after binarization are closed by a morphological filling processing method.
[0010] Further, the Moriwaki index, skewness and minimum neighborhood distance for representing the dispersion degree of the nanoparticles are calculated as the nanoparticle dispersion index, and the calculation processes thereof are respectively represented as: The Moriwaki index is represented as: ; Wherein, I is the Moriwaki dispersion index, p represents the number of electron microscope images processed and analyzed, n i is the number of nanoparticles in the first i electron microscope image, MThe total number of nanoparticles in all processed electron microscopy images; Skewness value is expressed as: ; in, This is the skewness value. N The number of electron microscopy images analyzed. x i For the first i Number of nanoparticles in electron microscopy images This represents the average number of nanoparticles in all electron microscopy images. for x i The standard deviation; The minimum neighborhood distance is expressed as: , , ; in, R The minimum neighborhood distance W i For particles i Its nearest neighbor particle j Euclidean distance, ( x i , y i )and( x j , y j ) are particles i and nearest neighbor particles j coordinates For particle number density, n This represents the total number of nanoparticles. S The area of the electron microscope image is denoted as .
[0011] Furthermore, after acquiring the electron microscope image of the nano-modified cellulose insulating paper to be tested, the method includes preprocessing the electron microscope image, specifically including the following steps: Import the electron microscope image of nano-modified cellulose insulating paper into Fiji software, draw the scale line of the electron microscope image using the straight line tool, set the electron microscope image scale, and complete the scale calibration of the electron microscope image. Use the rectangular selection tool to select the main area of the electron microscope image after ruler calibration, excluding the black background at the bottom and the white ruler part, to obtain the cropped image; Use the Brightness or Contrast function to drag the slider to enhance the contrast of the cropped image; Gaussian blur denoising was applied to the optimized image to remove image noise, resulting in the processed electron microscope image.
[0012] Further, the statistical process of the total number and coordinates of all nanoparticles in the electron microscope image further includes setting the particle size range, roundness value and aspect ratio parameters.
[0013] The application also includes a breakdown field strength prediction system for nanomodified cellulose insulation paper, comprising: An acquisition module is configured to acquire an electron microscope image of a nanomodified cellulose insulation paper sample. An index determination module is configured to classify and identify the nanoparticles and cellulose matrix in the electron microscope image, generate a classification result image, segment the adhered nanoparticles after binarization processing of the classification result image, and statistically obtain the total number and coordinates of all nanoparticles in the electron microscope image. A model establishment module is configured to establish a multiple linear regression model between the nanomodified cellulose insulation paper sample breakdown field strength test value and the nanomodified cellulose insulation paper sample nanometer particle dispersity index. A prediction module is configured to input the nanomodified cellulose insulation paper sample nanometer particle dispersity index into the multiple linear regression model to predict the breakdown field strength of the nanomodified cellulose insulation paper sample.
[0014] The application also includes a nanomodified cellulose insulation paper breakdown field strength prediction computer device, which comprises a memory, a processor and a computer program stored in the memory.
[0015] The application also includes a readable storage medium, which stores a computer program, and the computer program comprises program instructions.
[0016] The application provides a nanomodified cellulose insulation paper breakdown field strength prediction method, which has the following advantages: This invention enables quantitative assessment of the dispersibility of nanoparticles in cellulose insulating paper. It utilizes the Morishita index, skewness value, and minimum neighborhood distance—indicators of nanoparticle dispersion—to effectively distinguish between random dispersion and aggregation of nanoparticles. Furthermore, by establishing the relationship between the breakdown field strength of the insulating paper and the nanoparticle dispersibility, it achieves quantitative characterization of nanoparticle dispersibility and prediction of the breakdown field strength of the insulating paper, improving the accuracy of breakdown field strength prediction. This invention overcomes the long-standing technical bottleneck of relying on subjective evaluation using electron microscopy images in the field of cellulose insulating paper modification, providing data-driven decision-making basis for performance optimization and process improvement of cellulose insulating paper. Attached Figure Description
[0017] Figure 1 This is a schematic flowchart of the SEM image analysis method for nano-modified cellulose insulating paper in an embodiment of the present invention; Figure 2 This is a schematic diagram of SEM images of nano-modified cellulose insulating paper in an embodiment of the present invention; Figure 3 This is a classification result image after classification and recognition in an embodiment of the present invention; Figure 4 This is a binary image after binarization processing in an embodiment of the present invention; Figure 5 This is a statistical chart of nanoparticle recognition in an embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0019] This invention proposes a method for predicting the breakdown field strength of nano-modified cellulose insulating paper. This method processes and analyzes SEM or TEM images of nano-modified cellulose insulating paper to obtain characteristic parameters such as the total number of nanoparticles in a batch of images, and calculates the Morishita index, skewness value and minimum neighborhood distance, thereby quantitatively characterizing the dispersion of nanoparticles in cellulose insulating paper.
[0020] like Figure 1 As shown, the method specifically includes the following steps: S1. The microstructure of the nano-modified cellulose insulating paper is photographed under a scanning electron microscope (SEM) or a transmission electron microscope (TEM) to obtain electron microscope images.
[0021] S2. Image Preprocessing: The electron microscope images of the nano-modified cellulose insulating paper are preprocessed, including scale calibration, cropping, contrast optimization, and noise reduction. Fiji software can be used for batch processing of images to improve analysis efficiency.
[0022] Ruler calibration: Import the SEM image of nanomodified cellulose insulation paper in Fiji software, draw the ruler line of the image with the straight line tool, and set the image ruler. Since the fiber size of the insulation paper is usually in microns and the nanoparticles are in nanometers, the ruler unit is generally selected as “pm”.
[0023] Crop: Use the rectangular selection tool to frame the main area of the SEM image (excluding the black background at the bottom and the white ruler part), and make sure that the selected area completely covers the nanoparticles and the cellulose matrix to be analyzed.
[0024] Contrast optimization: Use the Brightness / Contrast function to enhance the light and dark contrast of the image by dragging the slider to make the boundary between the nanoparticles and the fibers clearer.
[0025] Gaussian blur noise reduction: Use Gaussian blur noise reduction to remove image noise with a radius of 0.5 pixels.
[0026] S3, enable the Trainable Weka Segmentation plug-in in Fiji software, create two labels, class1 (label 1) and class2 (label 2), and use the brush tool to label typical areas of nanoparticles and cellulose matrix, respectively, and classify them into class1 and class2. Select enough typical areas to distinguish cellulose matrix and nanoparticles, enable multi-scale texture feature extraction (including Gabor filter and Haralick texture) and edge detection feature. Set the number of random forest classifier trees to 150, and dynamically adjust the labeling area through real-time preview to verify the classification effect until the nanoparticles and cellulose matrix can be clearly classified and recognized.
[0027] S4, nanoparticle recognition and statistics. After generating the classification result image, extract the nanoparticle channel and convert it to a binary image. Apply the watershed algorithm to segment the adherent particles, and close the internal holes by morphological filling. In the particle analysis function, set the recognition parameters, including particle size range and circularity value, and set the aspect ratio to 1.0-2.0 to exclude fibrous impurities. Record the total number and coordinates of all nanoparticles in the processed SEM image.
[0028] S5, take at least 6 SEM images of different parts of the same nanomodified cellulose insulation paper and repeat the above steps S2-S4 to obtain the total number of nanoparticles in each image, and calculate the Senoue index, skewness and minimum neighborhood distance as the dispersion index according to the following formula: Senoue index: ; In the formula, I is the Senoue index, pThis represents the number of SEM images processed and analyzed. n i For the first i The number of nanoparticles in the image. M Total number of nanoparticles in all processed SEM images. Morishita Dispersion Index. I >1 indicates that nanoparticles have aggregated. I The higher the value, the more severe the family reunion; I <1 indicates that the nanoparticles are uniformly dispersed. I The closer the value is to 1, the more uniformly the nanoparticles are dispersed.
[0029] Skewness value: ; In the formula Here, represents the skewness value, and N represents the number of SEM images analyzed. x i For the first i Number of nanoparticles in a single SEM image This represents the average number of nanoparticles across all images. for x i The standard deviation of the skewness value. When the value approaches 0, it indicates that the nanoparticles are uniformly dispersed in the insulating paper; when... When <0 and far from zero, it indicates that the particles are mainly small particles with few aggregates; when When the value is greater than 0 and far from zero, it indicates that there are large agglomerates of particles in the insulating paper. The larger the value, the more severe the agglomeration.
[0030] Minimum neighborhood distance: ; ; ; In the formula R The minimum neighborhood distance W i For particles i Its nearest neighbor particle j Euclidean distance, ( x i , y i )and( x j , y j ) are particles i and nearest neighbor particles j coordinates For particle number density, n This represents the total number of nanoparticles. S The area of the SEM image. When... RA value greater than 1 indicates good particle dispersion in the insulating paper; the larger the deviation from 1, the better the dispersion. R When the value is less than 1, it indicates that the nanoparticles are agglomerated, and the greater the deviation from 1, the more severe the agglomeration phenomenon.
[0031] S6. Based on the measured value E of the breakdown field strength of the nano-modified insulating paper, the correlation between the breakdown field strength of the insulating paper and the nano-dispersion index parameters is established using the multiple linear regression method: ; In the formula, E The breakdown field strength of the nano-modified insulating paper, , , , These are the regression coefficients, calculated using the least squares method.
[0032] Based on the above methods, the present invention proposes an embodiment, which specifically includes the following steps: S1. Take samples of the prepared nano-SiO2 modified cellulose insulating paper, take SEM images at an appropriate magnification, and save them. Figure 2 As shown.
[0033] S2. Import the SEM image of nano-modified cellulose insulating paper into Fiji software for preprocessing. Use the line tool to calibrate the scale in the electron microscope image. Set the image scale. Since the fiber size of cellulose insulating paper is usually in the micrometer range and the nanoparticles are in the nanometer range, the scale unit is set to the μm level.
[0034] To avoid errors in subsequent image recognition processing, the main body of the image was selected using the rectangular box tool in the software, ensuring that the selection area completely covered the nanoparticles and cellulose matrix to be analyzed. The scale bar at the bottom of the SEM image was then cropped and deleted. The brightness and contrast of the cropped image were adjusted to make the interface between the nanoparticles and fibers clearer, while also removing image noise.
[0035] S3. Using the Trainable Weka Segmentation plugin, classify and identify the cellulose matrix and nanoparticles in the insulating paper. Select multiple typical regions of each and classify them into class1 and class2 respectively. Enable multi-scale texture feature extraction and edge detection features. Set the number of trees in the random forest classifier to 150. Verify the classification effect by dynamically adjusting the labeled areas in real-time preview until the nanoparticles and cellulose matrix are completely distinguishable. Figure 3 As shown.
[0036] S4. Extract the nanoparticle channels from the image processed in step S3 and convert it into a binary image, such as... Figure 4The watershed algorithm in the application software was used to segment the adhered particles and fill the closed internal holes. The recognition parameters were set in the particle analysis function, including the particle size range and the circularity value, and the aspect ratio was set to 1.0-2.0 to exclude fibrous impurities. The total number of nanoparticles in the processed image was counted.
[0037] S5, in this example, 9 SEM images of different parts of the nanomodified cellulose insulation paper treated in different ways were taken, and the above steps S2-S4 were repeated to obtain the total number of nanoparticles in each image. The results are shown in Figure 1 According to the Moriwaki index, skewness value and minimum neighborhood distance as the nanoparticle dispersion index, the calculation formulas of the three parameters are as follows: Moriwaki index: ; In the formula, I Moriwaki index, p represents the number of SEM images analyzed, n i represents the number of nanoparticles in the i-th image, M represents the total number of nanoparticles in all SEM images. Moriwaki dispersion index I > 1, indicating that the nanoparticles are aggregated, I the greater the aggregation; I < 1, indicating that the nanoparticles are uniformly dispersed, I the closer the value is to 1, the more uniform the dispersion of nanoparticles.
[0038] Skewness value: ; In the formula skewness value, N is the number of SEM images analyzed, x i is the number of nanoparticles in the i-th SEM image, i is the average number of nanoparticles in all images, is x i standard deviation. When the skewness value tends to 0, it indicates that the nanoparticles are uniformly dispersed in the insulation paper; when < 0 and far from zero, it indicates that the particles are mainly small particles with fewer aggregates; when > 0 and far from zero, it indicates that there are larger aggregates of particles in the insulation paper, and the greater the value, the more serious the aggregation.
[0039] Minimum neighborhood distance: , , ; In the formulaR The minimum neighborhood distance W i For particles i Its nearest neighbor particle j Euclidean distance, ( x i , y i )and( x j , y j ) are particles i and nearest neighbor particles j coordinates For particle number density, n Let S be the total number of nanoparticles, and S be the area of the SEM image. When... R A value greater than 1 indicates good particle dispersion in the insulating paper; the larger the deviation from 1, the better the dispersion. R When the value is less than 1, it indicates that the nanoparticles are agglomerated, and the greater the deviation from 1, the more severe the agglomeration phenomenon.
[0040] S4. Based on the measured value E of the breakdown field strength of the nano-modified insulating paper, the correlation between the breakdown field strength of the insulating paper and the nano-dispersion index parameters is established using the multiple linear regression method: ; In the formula, E is the breakdown field strength of the nano-modified insulating paper. , , , These are the regression coefficients, calculated using the least squares method.
[0041] By testing and calculating the breakdown field strength and nanoparticle dispersion parameters of multiple nano-modified insulating papers, and fitting the above formula, the relationship between the breakdown field strength of the insulating paper and the nanoparticle dispersion can be established, realizing the quantitative characterization of nanoparticle dispersion and its prediction of the breakdown field strength of the insulating paper.
[0042] Based on the same inventive concept, this invention also proposes a breakdown field strength prediction system for nano-modified cellulose insulating paper, comprising: The acquisition module is used to acquire electron microscopic images of nano-modified cellulose insulating paper samples.
[0043] The index determination module is used for classifying and identifying the nanoparticles and the cellulose matrix in the electron microscope image, generating a classification result image, segmenting the adhered nanoparticles after binarizing the classification result image, and counting the total number and coordinates of all the nanoparticles in the electron microscope image; and the Morishita index, skewness value and minimum neighborhood distance used for representing the dispersion degree of the nanoparticles are calculated as the nanoparticle dispersion index according to the total number and coordinates of the nanoparticles.
[0044] The model establishment module is used for establishing a multiple linear regression model between the nanoparticle dispersion index of the nanomodified cellulose insulation paper sample and the breakdown field strength of the insulation paper according to the test value of the breakdown field strength of the nanomodified cellulose insulation paper sample.
[0045] The prediction module is used for inputting the nanoparticle dispersion index of the nanomodified cellulose insulation paper sample to be tested into the multiple linear regression model, and predicting the breakdown field strength of the nanomodified cellulose insulation paper to be tested.
[0046] The present application also provides a computer device for predicting the breakdown field strength of nanomodified cellulose insulation paper, which comprises a memory, a processor and a computer program stored in the memory, and the processor implements the steps of the method for predicting the breakdown field strength of nanomodified cellulose insulation paper when executing the computer program.
[0047] The present application also provides a readable storage medium, which stores a computer program, and the computer program comprises program instructions, and the program instructions are executed by a processor to implement the steps of the method for predicting the breakdown field strength of nanomodified cellulose insulation paper.
[0048] The above description is only the preferred embodiment of the present application, but the protection scope of the present application is not limited to this, and any person skilled in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.
Claims
1. A method of predicting the breakdown field strength of a nanomodified cellulose insulation paper, characterized by, The method comprises the following steps: An electron microscope image of the nanomodified cellulose insulation paper sample is obtained; The nanometer particles and the cellulose matrix in the electron microscope image are classified and identified to generate a classification result image, and after the classification result image is binarized, the adhered nanometer particles are segmented, and the total number and coordinates of all nanometer particles in the electron microscope image are counted; the Harashima index, skewness value and minimum neighborhood distance for representing the dispersion degree of the nanometer particles are calculated as the nanometer particle dispersion index according to the total number and coordinates of the nanometer particles; A multiple linear regression model between the nanometer particle dispersion index of the nanomodified cellulose insulation paper sample and the breakdown field strength of the insulation paper is established according to the test value of the breakdown field strength of the nanomodified cellulose insulation paper sample; The nanometer particle dispersion index of the nanomodified cellulose insulation paper sample to be tested is input into the multiple linear regression model to predict the breakdown field strength of the nanomodified cellulose insulation paper to be tested.
2. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of 20 kV / mm or more. The nanometer particles and the cellulose matrix in the electron microscope image are classified and identified to generate a classification result image, and the classification result image is binarized, and the adhered nanometer particles are segmented, and the total number and coordinates of all nanometer particles in the electron microscope image are counted; the Harashima index, skewness value and minimum neighborhood distance for representing the dispersion degree of the nanometer particles are calculated as the nanometer particle dispersion index according to the total number and coordinates of the nanometer particles; A classification result image is generated by using a brush tool to mark the nanometer particle region and the texture region of the cellulose matrix in the nanomodified cellulose insulation paper sample and classifying them into two types of labels, and training a random forest classifier through the two types of labels after classification, and realizing the classification of nanometer particles and cellulose matrix through multi-scale texture feature extraction and edge detection feature; The adhered nanometer particles are segmented by using a watershed algorithm on the binarized classification result image. In the process of segmenting the adhered nanometer particles, the internal cavities of the nanometer particle region in the binarized classification result image are closed by a morphological filling processing method.
3. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of 20 kV / mm or more. The Harashima index, skewness value and minimum neighborhood distance for representing the dispersion degree of the nanometer particles are calculated as the nanometer particle dispersion index, and the calculation process is respectively represented as:
4. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of 20 kV / mm or more. After obtaining the electron microscope image of the nanomodified cellulose insulation paper to be tested, the electron microscope image is preprocessed, and the preprocessing specifically comprises the following steps:
5. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of at least 20 kV / mm. The electron microscope image of the nanomodified cellulose insulation paper is imported into the Fiji software, a scale line of the electron microscope image is drawn by using a straight line tool, a scale of the electron microscope image is set, and the scale calibration of the electron microscope image is completed; Morishita's index is expressed as: ; wherein, I as Morinaga's dispersion index, p represents the number of electron microscope images analyzed, n i as the first i nanoparticles in the electron microscope image, M as the total number of nanoparticles in the electron microscope images of all treatments; Skewness values are expressed as: ; wherein, is the skewness value, N is the number of analyzed electron microscope images, x i is the first i is the number of nanoparticles in the z-th electron microscope image, is the average number of nanoparticles in all electron microscope images, is the x i is the standard deviation; The minimum neighborhood distance is represented as: , , ; where, R is the minimum neighborhood distance, W i is the particle i is the Euclidean distance between the particle j and its nearest neighbor particle, x i , y i and x j , y j are the coordinates of the particle i and the nearest neighbor particle j respectively, is the particle number density, n is the total number of nanoparticles, S is the electron microscopy image area.
6. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of at least 20 kV / mm. The main body region of the electron microscope image after scale calibration is framed by using a rectangular selection tool, the black background at the bottom and the white scale part are excluded, and a cropped image is obtained; The Brightness or Contrast function is used to enhance the light and dark contrast of the cropped image by dragging the slider; The image noise points are removed by using Gaussian blur denoising on the contrast-optimized image to obtain a processed electron microscope image. In the counting process of the total number and coordinates of all nanometer particles in the electron microscope image, the particle size range, circularity value and aspect ratio parameters are set. The method comprises the following steps:
7. The method of claim 1, wherein the nanoreformed cellulose insulation paper has a breakdown field strength of at least 20 kV / mm. An electron microscope image of the nanomodified cellulose insulation paper sample is obtained by using the obtaining module; 8. A system for predicting the breakdown field strength of a nanomodified cellulose insulation paper, characterized by, The index determination module is used for classifying and identifying the nanoparticles and the cellulose matrix in the electron microscope image, generating a classification result image, segmenting the adhered nanoparticles after the binary processing of the classification result image, and counting the total number and coordinates of all the nanoparticles in the electron microscope image; according to the total number and coordinates of the nanoparticles, the Morishita index, the skewness value and the minimum neighborhood distance used for representing the dispersion degree of the nanoparticles are calculated as the nanoparticle dispersion index; The model establishment module is used for establishing a multiple linear regression model between the nanoparticle dispersion index of the nanomodified cellulose insulation paper sample and the breakdown field strength of the insulation paper according to the test value of the breakdown field strength of the nanomodified cellulose insulation paper sample; The prediction module is used for inputting the nanoparticle dispersion index of the nanomodified cellulose insulation paper sample to be tested into the multiple linear regression model, and predicting the breakdown field strength of the nanomodified cellulose insulation paper to be tested.
9. A computer device for predicting the breakdown field strength of a nanomodified cellulose insulation paper, characterized by It comprises: A memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to realize the steps of the breakdown field strength prediction method of the nanomodified cellulose insulation paper according to any one of claims 1-7.
10. A readable storage medium, characterized by, The readable storage medium stores a computer program, and the computer program comprises program instructions, wherein the program instructions are executed by the processor to realize the steps of the breakdown field strength prediction method of the nanomodified cellulose insulation paper according to any one of claims 1-7.