Insulation paper polymerization degree prediction method and system

Abstract

The invention discloses an insulation paper polymerization degree prediction method and system. The method comprises the steps of obtaining first data and second data of a target sample, and taking the first data and the second data as a first training set; establishing a first prediction model, and training the first prediction model by using the first training set; generating third data according to the first data, and inputting the third data into a first prediction model to obtain fourth data; and forming a second training set by the first data, the second data, the third data and the fourth data, and retraining the first prediction model by using the second training set. A reliable data basis is provided for prediction of the polymerization degree and the aging degree of the insulation paper based on near infrared spectroscopy analysis, and a stable and credible data support is provided for prediction of the aging state of the insulation paper.

Description

Technical Field

[0001] The present invention relates to the technical field of near infrared spectroscopy analysis, and in particular to a method and system for predicting the polymerization degree of insulating paper. Background Art

[0002] Near-infrared spectroscopy is a fast and non-destructive in-situ detection method, which is widely used in many industries such as agriculture, food, petrochemicals, and pharmaceuticals. Near-infrared spectroscopy is generated by the absorption of light by material molecules during the transition of frequency-doubled and frequency-combined energy levels. It reflects the microscopic characteristics of the material at the molecular level and is a commonly used method for material identification, component analysis, and process monitoring. In terms of transformer insulation paper aging detection, near-infrared spectroscopy analysis technology has shown certain advantages. By analyzing the near-infrared spectrum reflected or transmitted by the insulation paper, important information about the chemical composition and structural changes of the material can be obtained, thereby indirectly evaluating the degree of aging of the insulation paper.

[0003] In near-infrared spectroscopy analysis technology, insufficient sample quantity, insufficient sample information and poor data quality will affect the accuracy and generalization ability of the evaluation model. In the field of near-infrared spectroscopy analysis, the current strategies for improving the predictive ability of the model can be divided into two categories: improving data quality and optimizing modeling methods. The former is mainly achieved through the following methods: normalization normalizes spectral data of different scales to a unified scale to reduce the impact of dimension; smoothing, using wavelet transform, smoothing filter and other methods to remove noise in spectral data; baseline correction removes baseline drift; standard normal transformation eliminates the influence of particle size and thickness on the spectrum; by calculating the first-order or second-order derivative spectrum, the chemical information in the spectrum is enhanced and the influence of background absorption and light scattering is reduced; by extracting useful features or screening some important bands to reduce redundant information and improve the model prediction ability. The latter is to improve the prediction accuracy and generalization ability of the model through ensemble learning or local modeling methods. However, the model is based on data, and optimizing the modeling method has a relatively limited effect on improving the model prediction ability.

[0004] In summary, it is difficult to accumulate a large number of insulating paper samples with different aging times and test their polymerization degrees one by one. Existing strategies to improve the prediction ability of the model can improve data quality or deeply mine effective information in the data to a certain extent, but these methods are ultimately difficult to solve the problems of lack of spatial information in the sample set itself and uneven information distribution, resulting in difficulty in effectively improving the accuracy and generalization ability of the evaluation model. Summary of the invention

[0005] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0006] In view of the above existing problems, the present invention is proposed.

[0007] Therefore, the present invention provides a method and system for predicting the degree of polymerization of insulating paper, which can solve the problems mentioned in the background technology.

[0008] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0009] In a first aspect, the present invention provides a method for predicting the degree of polymerization of insulating paper, comprising:

[0010] Acquire first data and second data of a target sample, and use the first data and the second data as a first training set;

[0011] Establishing a first prediction model, and training the first prediction model using the first training set;

[0012] Generate third data according to the first data, and input the third data into the first prediction model to obtain fourth data;

[0013] The first data, the second data, the third data and the fourth data are combined into a second training set, and the first prediction model is retrained using the second training set.

[0014] As a preferred solution of the insulation paper polymerization degree prediction method of the present invention, wherein: the establishing of the first prediction model and the training of the first prediction model using the first training set include:

[0015] The first prediction model is any model whose input is first data and whose output is second data.

[0016] As a preferred solution of the insulation paper polymerization degree prediction method of the present invention, wherein: generating the third data according to the first data comprises:

[0017] Determine the left boundary and the right boundary of the first data;

[0018] The third data is generated within the range of the left boundary and the right boundary.

[0019] As a preferred embodiment of the method for predicting the polymerization degree of insulating paper described in the present invention, the target sample includes: drying the insulating paper and insulating oil, carrying out accelerated thermal aging test of oil-paper insulation under vacuum conditions, and preparing insulating paper samples with different aging states.

[0020] As a preferred solution of the insulation paper polymerization degree prediction method of the present invention, wherein: the taking the first data and the second data as the first training set comprises:

[0021] The first data is preprocessed, wherein the preprocessing includes a standardization process and a smoothing process based on a Savitzky-Golay convolution smoothing method.

[0022] As a preferred solution of the method for predicting the polymerization degree of insulating paper described in the present invention, the second data includes: the polymerization degree of each insulating paper detected by a viscosity method.

[0023] As a preferred solution of the insulation paper polymerization degree prediction method described in the present invention, wherein: the retraining of the first prediction model using the second training set includes: using the root mean square error RMSE to judge the performance of the first prediction model, and training the first prediction model.

[0024] In a second aspect, the present invention provides an insulation paper polymerization degree prediction system, comprising:

[0025] A first training set acquisition module, used to acquire first data and second data of a target sample, and use the first data and the second data as a first training set;

[0026] A model building module, used to build a first prediction model, and train the first prediction model using the first training set;

[0027] a data acquisition module, configured to generate third data according to the first data, and input the third data into the first prediction model to acquire fourth data;

[0028] The model retraining module is used to combine the first data, the second data, the third data and the fourth data into a second training set, and use the second training set to retrain the first prediction model.

[0029] In a third aspect, the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned method when executing the computer program.

[0030] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the method described above when executed by a processor.

[0031] Compared with the prior art, the invention has the following beneficial effects: the invention proposes a method and system for predicting the polymerization degree of insulating paper, obtains the first data and the second data of the target sample, and uses the first data and the second data as the first training set; establishes a first prediction model, and uses the first training set to train the first prediction model; generates third data according to the first data, and inputs the third data into the first prediction model to obtain fourth data; the first data, the second data, the third data and the fourth data form a second training set, and uses the second training set to retrain the first prediction model. The method uses the overall trend diffusion technology to expand the spectral data set, improve the diversity of the sample spectrum, and thus enhance the representativeness of the data set. The method combines the partial least squares regression model to realize the prediction of the polymerization degree of insulating paper in the virtual sample, and then constructs a spectral data set containing richer information and more balanced information distribution, providing a reliable data basis for the prediction of the polymerization degree and aging degree of insulating paper based on near-infrared spectral analysis. Compared with other methods for improving the prediction performance of the aging assessment model, this method starts with the data itself and generates virtual samples to enhance the diversity and representativeness of the spectral data set. Compared with the existing virtual sample generation method, this method integrates the overall trend diffusion technology and the partial least squares regression model, fully utilizing the advantages of the partial least squares regression model in dealing with high-dimensional data and collinearity problems, and realizing accurate prediction of the polymerization degree value corresponding to the virtual spectrum. It is suitable for near-infrared spectral high-dimensional data and can provide solid and reliable data support for the prediction of the aging state of insulating paper. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0033] Figure 1 A method flow chart of a method and system for predicting the polymerization degree of insulating paper provided by one embodiment of the present invention;

[0034] Figure 2 A schematic diagram of prediction results of a PLS regression model of a method and system for predicting the polymerization degree of insulating paper provided by an embodiment of the present invention;

[0035] Figure 3 A schematic diagram of MTD of a method and system for predicting the polymerization degree of insulating paper provided by an embodiment of the present invention;

[0036] Figure 4 A schematic diagram of an original sample spectrum curve of a method and system for predicting the degree of polymerization of insulating paper provided by an embodiment of the present invention;

[0037] Figure 5 A schematic diagram of a virtual sample spectrum curve of a method and system for predicting the polymerization degree of insulating paper provided by an embodiment of the present invention;

[0038] Figure 6 A schematic diagram of 100 performance test results of a method and system for predicting the degree of polymerization of insulating paper provided by an embodiment of the present invention;

[0039] Figure 7 An internal structural diagram of a computer device of a method and system for predicting the polymerization degree of insulating paper provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0041] Example 1

[0042] Reference Figure 1-Figure 7 , which is the first embodiment of the present invention, provides a method and system for predicting the polymerization degree of insulating paper, including:

[0043] In the existing related technologies, there are problems such as low accuracy in predicting the polymerization degree of insulation paper and limited performance of aging assessment models.

[0044] The present application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to explain in detail how to implement the insulation paper polymerization degree prediction method;

[0045] Figure 1 A method flow chart of a method and system for predicting the polymerization degree of insulating paper is shown, including:

[0046] S101, obtaining first data and second data of a target sample, and using the first data and the second data as a first training set;

[0047] In the embodiment of the present application, the target sample includes drying the insulating paper and the insulating oil, carrying out an accelerated thermal aging test of the oil-paper insulation under vacuum conditions, and preparing insulating paper samples with different aging states.

[0048] In the embodiment of the present application, the insulating paper and the insulating oil are dried to make the water content of the insulating paper less than 1%, and the water content of the insulating oil less than 10 mg / L, and the insulating paper model is B125. The accelerated thermal aging test of oil-paper insulation is carried out under vacuum conditions, and the accelerated thermal aging temperature is set to prepare insulating paper samples with different aging states of 2-100 days.

[0049] In an optional embodiment, a NIR Quest near-infrared spectrometer produced by Ocean Optics can be used to collect near-infrared spectra of the insulating paper. The wavelength range collected by the near-infrared spectrometer is 900-2200nm, and a WS-1 diffuse reflectance standard plate is used as a standard white plate reference. Each spectrum is the average result obtained by scanning 12 times, and the near-infrared spectra of the insulating paper samples in different aging states are obtained. At the same time, the viscosity method is used to detect the degree of polymerization of each insulating paper.

[0050] In the embodiment of the present application, taking the first data and the second data as the first training set includes:

[0051] The first data is preprocessed, and the preprocessing includes standardization processing and smoothing processing based on Savitzky-Golay convolution smoothing method.

[0052] In the embodiment of the present application, the first data is the near infrared spectrum data of the insulating paper samples with different aging times;

[0053] In an optional embodiment, the first data is not limited to near-infrared spectral data, and any data whose relationship with the second data (i.e., degree of polymerization) can be obtained through the first data can be collected as the first data, for example, Fourier transform infrared spectral data, Raman spectral data, etc.

[0054] In the embodiment of the present application, the second data includes the degree of polymerization of each insulating paper detected by a viscosity method.

[0055] In an optional embodiment, the second data may also be obtained by other methods, such as infrared spectroscopy analysis, nuclear magnetic resonance (NMR) analysis, etc. In addition, the collection of the first data and the second data may be continuous or discrete, depending on the requirements and conditions of the actual application.

[0056] In the embodiment of the present application, the near infrared spectrum data of the insulating paper sample is standardized and smoothed to reduce the interference of the noise signal on the spectrum data. The standardization method is as follows:

[0057]

[0058] Wherein, x is the original spectral data, μ is the average value of all the original spectral data, σ is the standard deviation of all the original spectral data, and x* is the processed spectral data.

[0059] It should be noted that the function of smoothing is to filter out high-frequency noise, using the Savitzky-Golay convolution smoothing method (SG smoothing). The principle of the SG smoothing method is to slide a window on the original data, fit the data in the window to a polynomial using the least squares method, and use the fitted value as the value of the data in the window. The specific processing method is as follows:

[0060]

[0061] in, is the smoothed data set; x j+i is the collected data set; a i , a0 is the smoothing coefficient, that is, the original data x of the i-th period j+i The weight in the smoothing window; n is the number of data in the sliding window; m is the window width, where n=2m+1.

[0062] It should be noted that the benefit of obtaining the first data and the second data of the target sample and using the first data and the second data as the first training set is that the accuracy of the prediction of the degree of polymerization of the insulating paper can be improved by establishing a correlation model between the data. In this way, the aging state of the insulating paper can be more accurately evaluated, thereby providing a more reliable guarantee for the safe operation of the power equipment.

[0063] S102, establishing a first prediction model, and training the first prediction model using a first training set;

[0064] In an embodiment of the present application, a first prediction model is established, and training the first prediction model using a first training set includes: the first prediction model is any model whose input is first data and whose output is second data.

[0065] In an optional embodiment, the first prediction model can be constructed by a machine learning algorithm, such as a support vector machine (SVM), a random forest, a neural network, etc. Through these algorithms, the relationship between the first data and the second data of the insulating paper sample can be learned and modeled. It is also possible to construct a partial least squares regression model, use the first data as input, and the second data as output, and train the partial least squares regression model through a training set. It is also possible to use methods other than machine learning and building a partial least squares regression model to establish it. For example, statistical methods such as multivariate linear regression analysis and principal component analysis can be used to establish a prediction model. When selecting a suitable prediction model, factors such as the complexity of the model, prediction accuracy, and computational efficiency need to be considered. However, no matter what kind of model is established, as long as the input of the model is the first data and the output is the second data, the model should be within the scope of protection of this application.

[0066] In the embodiment of the present application, a partial least squares regression model is constructed, near-infrared spectral data (first data) is used as input, and the degree of polymerization (second data) is used as output, and the partial least squares regression model is trained using a training set.

[0067] It should be noted that the partial least squares (PLS) method reduces the dimensionality of spectral data by screening and sorting out the effective information in the spectrum and selecting several new comprehensive variables (i.e., principal components) for regression modeling. At the same time, the partial least squares method introduces the information in the polymerization degree data when decomposing the spectral data, so that the two types of data are correlated with each other, thereby effectively extracting spectral feature information and improving the accuracy of system modeling.

[0068] In the embodiment of the present application, the pre-processed near-infrared spectral data of the insulating paper is randomly divided into a training set and a test set in a ratio of 7:3. Based on the training set data, the spectral data is input, and the polymerization degree value is output, a preliminary prediction model of the polymerization degree of the insulating paper is constructed based on partial least squares regression. The partial least squares regression modeling method is as follows:

[0069] In the embodiment of the present application, the spectrum matrix X and the polymerization degree matrix Y of the near infrared spectrum are decomposed:

[0070]

[0071]

[0072] Among them, t k is the score of the kth principal factor of the spectral matrix X; p k is the load of the kth principal factor of the spectral matrix; u k is the score of the kth principal factor of the aggregation matrix Y; q kis the load of the kth principal factor of the aggregation matrix Y; f is the number of principal factors; T and U are the score matrices of the X and Y matrices, Р and Q are the load matrices of the X and Y matrices, E X and E Y are the partial least squares fitting residual matrices of X and Y respectively;

[0073] In the embodiment of the present application, linear regression is performed on T and U. When predicting, the score matrix T of the unknown sample spectrum matrix X is obtained according to Р, and then the predicted value Y of the degree of polymerization is obtained by the following formula:

[0074] U=TB

[0075] B=(T T T) -1 T T Y

[0076] Y 预测 =T·BQ

[0077] Among them, Y prediction is the predicted value of aggregation degree.

[0078] In an optional embodiment, the performance of the initially established PLS regression model can be tested using a test set, and the root mean square error RMSE is used to judge the model performance. The calculation method is as follows:

[0079]

[0080] It should be noted that Figure 2 The prediction results of the PLS regression model are as follows. The test results show that the root mean square error of the insulation paper polymerization degree prediction of the PLS regression model established through the original samples is 64.372. The prediction accuracy has reached a relatively high level and can be used as a method basis for predicting the insulation paper polymerization degree to a certain extent.

[0081] It should be noted that the first prediction model can effectively extract characteristic information related to the degree of polymerization from the near-infrared spectrum data of the insulating paper, thereby improving the accuracy and reliability of the prediction model. In addition, by continuously optimizing and adjusting the model parameters, the accuracy of the prediction can be further improved to ensure that the prediction results are more in line with actual application needs.

[0082] S103, generating third data according to the first data, and inputting the third data into the first prediction model to obtain fourth data;

[0083] In the embodiment of the present application, generating the third data according to the first data includes: determining the left boundary and the right boundary of the first data, and generating the third data within the range of the left boundary and the right boundary.

[0084] It should be noted that determining the left and right boundaries of the data is a relatively important point in this application, because the selection of the left and right boundaries directly affects the generation of the third data, and thus affects the accuracy and reliability of the prediction model. The selection of the left and right boundaries can be based on the key characteristic points in the aging process of the insulating paper. For example, it can be determined based on the range of changes in certain chemical components during the aging process of the insulating paper. In this way, it can be ensured that the generated third data can fully reflect the changing trend of the aging state of the insulating paper, thereby improving the sensitivity and prediction accuracy of the prediction model to changes in the degree of polymerization of the insulating paper.

[0085] In an optional embodiment, generating the third data within the left boundary and the right boundary can generate the third data (in the present application, a virtual spectrum sample) that obeys the normal distribution within the two boundaries, and can also generate the third data that obeys the uniform distribution. In this way, the spectral characteristics of the insulating paper at different aging stages can be simulated, thereby enhancing the model's ability to predict changes in the degree of polymerization of the insulating paper. In addition, the errors that may occur in the actual measurement can be simulated by introducing noise data, further improving the robustness of the model. Regardless of the third data generation method used, the third data should be guaranteed to be within the left boundary and the right boundary.

[0086] In an embodiment of the present application, third data that obeys a normal distribution are generated within two boundaries. In order to further improve the accuracy and stability of the PLS regression model in field applications, the overall trend diffusion technology is used to expand the training set samples, and a virtual spectrum corresponding to the near-infrared spectrum training data set is generated. By introducing virtual samples, the amount of model training data is increased, thereby improving the model prediction performance.

[0087] It should be noted that the Mega Trend Diffusion (MTD) technique generates new virtual samples based on the analysis of the main trends of existing data. The generation of virtual samples depends on identifying the main trends of the data and their variability, and then generating new data points within this framework. Figure 3 As shown, Figure 3 Schematic diagram of MTD.

[0088] In the embodiment of the present application, the key to the overall trend diffusion technology is to define the acceptable boundary of the generated virtual sample. The method for determining the acceptable boundary of the virtual sample of the present invention is to calculate the left boundary L and the right boundary U of the corresponding virtual sample information through the membership function, so as to generate the virtual sample information within the range. The specific boundary calculation method is as follows:

[0089]

[0090]

[0091] in:

[0092] CL=(max+min) / 2

[0093]

[0094]

[0095]

[0096] Among them, n represents the size of the sample set; L is the left boundary of the generated virtual data; U is the right boundary of the generated virtual data; CL is the center of the data; is the variance of the sample set; SkewL represents the left skewness that describes the data diffusion characteristics; SkewU represents the right skewness that describes the data diffusion characteristics.

[0097] The generation method of virtual spectrum samples is as follows:

[0098]

[0099] in, is the virtual spectrum sample information; n is the number of generated virtual spectrum samples; s is a normal distribution random number; L and U are the left and right boundaries. The original sample and virtual sample spectrum curves are shown in Figure 4 and Figure 5 As shown. The above method can generate virtual sample information that obeys the normal distribution within the two boundaries of the original data. By performing trend simulation and asymmetric diffusion on the original data, the amount and diversity of data can be effectively increased, thereby improving the prediction accuracy and robustness of the model during model training.

[0100] It should be noted that the MTD technology can only generate virtual spectrum data, and it is impossible to effectively simulate the degree of polymerization value corresponding to the virtual spectrum. Therefore, the PLS regression model is introduced to predict the degree of polymerization value corresponding to the virtual spectrum. The specific method is to input the generated virtual spectrum into the PLS regression model trained with the original sample, and use the output as the degree of polymerization value corresponding to the virtual spectrum, so as to obtain the complete virtual spectrum sample data. The PLS model can achieve a high level of polymerization prediction accuracy, so its output can be used as the basis for determining the degree of polymerization of the virtual sample.

[0101] It should also be noted that the training set divided when the PLS regression model is constructed is the data basis, and the MTD technology is used to obtain the virtual samples corresponding to the training set. The original training set samples and the virtual samples are used as new training sets to reconstruct the PLS regression model. The model performance is tested using the test set, and the root mean square error RMSE is used to judge the model performance.

[0102] It should be noted that the benefit of generating the third data based on the first data and inputting the third data into the first prediction model to obtain the fourth data is that the generalization ability of the prediction model can be significantly improved. By generating the third data within the left and right boundaries, the spectral characteristics of the insulating paper at different aging stages can be simulated, which helps the model learn more comprehensive feature information. The fourth data obtained in this way, that is, the predicted value of the degree of polymerization, will be closer to the actual value, providing strong technical support for the quality control and life prediction of the insulating paper.

[0103] S104, the first data, the second data, the third data and the fourth data are combined into a second training set, and the first prediction model is retrained using the second training set.

[0104] In an embodiment of the present application, retraining the first prediction model using the second training set includes: using a root mean square error (RMSE) to judge the performance of the first prediction model, and training the first prediction model.

[0105] In an optional embodiment, in the process of retraining the first prediction model, the parameters of the model need to be optimized first. This can be achieved through a variety of optimization algorithms, such as gradient descent, genetic algorithm or particle swarm optimization. The goal of optimization is to reduce the error between the model prediction value and the actual value, that is, to reduce the root mean square error RMSE. Through iterative optimization, a set of optimal model parameters can be found so that the model performs best on the new training set.

[0106] In an optional embodiment, after the optimization is completed, the first prediction model can be updated with new parameters. The updated model will contain richer feature information and can better capture the complex changes in the aging process of insulating paper. In addition, in order to further verify the generalization ability of the model, the model can be tested using an independent test set that is not involved in the training. By comparing the predicted results on the test set with the actual values, the accuracy and reliability of the model can be evaluated.

[0107] In an optional embodiment, in order to ensure the stability and accuracy of the prediction model in actual application, the model can be retrained and updated regularly with new data. This ensures that the model can adapt to the long-term changes in the aging characteristics of the insulation paper, thereby providing continuous technical support for the quality control and life prediction of the insulation paper. In this way, the insulation paper polymerization degree prediction system can provide an efficient and accurate prediction tool for related industries, helping related companies to better perform equipment maintenance and aging management.

[0108] In summary, the present invention proposes a method for predicting the polymerization degree of insulating paper, obtaining the first data and the second data of the target sample, and using the first data and the second data as the first training set; establishing a first prediction model, and using the first training set to train the first prediction model; generating third data according to the first data, and inputting the third data into the first prediction model to obtain fourth data; forming the first data, the second data, the third data and the fourth data into a second training set, and retraining the first prediction model using the second training set. The method uses the overall trend diffusion technology to expand the spectral data set, improve the diversity of the sample spectrum, and thus enhance the representativeness of the data set. The method combines the partial least squares regression model to realize the prediction of the polymerization degree of insulating paper in the virtual sample, and then constructs a spectral data set containing richer information and more balanced information distribution, providing a reliable data basis for the prediction of the polymerization degree and aging degree of insulating paper based on near-infrared spectral analysis. Compared with other methods for improving the prediction performance of the aging assessment model, this method starts with the data itself and generates virtual samples to enhance the diversity and representativeness of the spectral data set. Compared with the existing virtual sample generation method, this method integrates the overall trend diffusion technology and the partial least squares regression model, fully utilizing the advantages of the partial least squares regression model in dealing with high-dimensional data and collinearity problems, and realizing accurate prediction of the polymerization degree value corresponding to the virtual spectrum. It is suitable for near-infrared spectral high-dimensional data and can provide solid and reliable data support for the prediction of the aging state of insulating paper.

[0109] This embodiment also provides an insulation paper polymerization degree prediction system, including:

[0110] A first training set acquisition module, used to acquire first data and second data of a target sample, and use the first data and the second data as a first training set;

[0111] A model building module, used to build a first prediction model and train the first prediction model using a first training set;

[0112] A data acquisition module, used to generate third data according to the first data, and input the third data into the first prediction model to obtain fourth data;

[0113] The model retraining module is used to combine the first data, the second data, the third data and the fourth data into a second training set, and use the second training set to retrain the first prediction model.

[0114] The above-mentioned unit modules may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above-mentioned modules.

[0115] This embodiment also provides a computer device, which may be a terminal, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for predicting the degree of polymerization of insulating paper is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, a trackball or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0116] This embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0117] Acquire first data and second data of a target sample, and use the first data and the second data as a first training set;

[0118] Establishing a first prediction model, and training the first prediction model using a first training set;

[0119] Generate third data according to the first data, and input the third data into the first prediction model to obtain fourth data;

[0120] The first data, the second data, the third data and the fourth data are combined into a second training set, and the first prediction model is retrained using the second training set.

[0121] Example 2

[0122] Reference Figure 6 , which is an embodiment of the present invention, provides a method and system for predicting the polymerization degree of insulating paper. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through experiments.

[0123] The training set and test set were randomly selected, and the PLS regression model obtained by original sample modeling and expanded sample modeling was tested 100 times. The results are as follows: Figure 6 And as shown in Table 1.

[0124] Table 1 Performance test statistics

[0125] Training sample set RMSE mean RMSE median RMSE standard deviation RMSE is very poor Raw data 70.639 70.136 11.299 49.536 Augmenting Data 51.671 49.837 8.697 45.069

[0126] The test results show that by introducing virtual samples generated by combining overall trend diffusion with PLS regression method, the accuracy and stability of transformer insulation value prediction can be effectively improved, which has great practical value for the field application of the model.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0128] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0129] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0130] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0131] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0132] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0133] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for predicting the degree of polymerization of insulating paper, characterized in that: include: Acquire first data and second data of a target sample, and use the first data and the second data as a first training set; Establishing a first prediction model, and training the first prediction model using the first training set; Generate third data according to the first data, and input the third data into the first prediction model to obtain fourth data; The first data, the second data, the third data and the fourth data are combined into a second training set, and the first prediction model is retrained using the second training set.

2. The method for predicting the degree of polymerization of insulating paper according to claim 1, characterized in that: The establishing of the first prediction model and training the first prediction model using the first training set comprises: The first prediction model is any model that takes first data as input and outputs second data.

3. The method for predicting the degree of polymerization of insulating paper according to claim 2, characterized in that: Generating third data according to the first data comprises: Determine the left boundary and the right boundary of the first data; The third data is generated within the range of the left boundary and the right boundary.

4. The method for predicting the degree of polymerization of insulating paper according to claim 3, characterized in that: The target samples include: drying the insulating paper and insulating oil, carrying out accelerated thermal aging test of oil-paper insulation under vacuum conditions, and preparing insulating paper samples with different aging states.

5. The method for predicting the degree of polymerization of insulating paper according to claim 4, characterized in that: The taking the first data and the second data as a first training set comprises: The first data is preprocessed, wherein the preprocessing includes a standardization process and a smoothing process based on a Savitzky-Golay convolution smoothing method.

6. The method for predicting the degree of polymerization of insulating paper according to claim 5, characterized in that: The second data includes: the degree of polymerization of each insulating paper detected by a viscosity method.

7. The method for predicting the degree of polymerization of insulating paper according to claim 6, characterized in that: The retraining of the first prediction model using the second training set includes: using a root mean square error (RMSE) to judge the performance of the first prediction model, and training the first prediction model.

8. An insulation paper polymerization degree prediction system, characterized in that: include: A first training set acquisition module, used to acquire first data and second data of a target sample, and use the first data and the second data as a first training set; A model building module, used to build a first prediction model, and train the first prediction model using the first training set; a data acquisition module, configured to generate third data according to the first data, and input the third data into the first prediction model to acquire fourth data; The model retraining module is used to combine the first data, the second data, the third data and the fourth data into a second training set, and use the second training set to retrain the first prediction model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images