Method and system for predicting elongation at break of insulating material of silicone rubber cable accessory

Abstract

The invention belongs to the technical field of electrical insulation detection of electrical equipment, and particularly relates to a method and system for predicting the elongation at break of a silicone rubber cable accessory insulating material, and the method comprises the steps: obtaining a target sample of the silicone rubber cable accessory insulating material, testing the near infrared spectrum of the target sample, and carrying out the preprocessing to obtain near infrared spectrum data; performing a tensile test on the target sample to obtain the elongation at break of the target sample; obtaining a near infrared spectrum data point basic database based on the near infrared spectrum data and the elongation at break, and establishing a principal component analysis regression model based on the near infrared spectrum data point basic database; and based on the principal component analysis regression model, predicting the elongation at break of the unknown sample of the silicone rubber cable accessory insulating material. The problem that in the prior art, the breaking elongation of the insulating material cannot be rapidly and accurately detected in a full-coverage mode is solved.

Description

Technical Field

[0001] This invention belongs to the field of electrical insulation testing technology for power equipment, and specifically relates to a method and system for predicting the elongation at break of insulating materials for silicone rubber cable accessories. Background Technology

[0002] With the development of the national economy and the advancement of urbanization, electricity load is constantly increasing, and the importance of power cables in urban power grids is becoming increasingly prominent. At the same time, with the rapid growth of cable networks, the number of cable and accessory suppliers is enormous, and the quality of their products varies greatly. Some suppliers have prominent problems such as production not meeting national standards, equipment not meeting technical specifications, and unstable product quality, seriously threatening the safe and stable operation of urban power grids. Furthermore, during long-term operation, cable equipment is affected by factors such as light, heat, electricity, and mechanical stress, which accelerates its aging process, causing overheating, breakage, and other faults, leading to large-scale power grid outages.

[0003] Currently, the random inspection items for wire and cable products include conductor resistance testing at 20℃, insulation thickness testing, sheath thickness testing, tensile strength and elongation at break testing before and after insulation aging, and changes in elongation at break testing before and after sheath aging. Among these, the mechanical performance testing of insulation materials generally involves testing dumbbell-shaped samples before and after 7 days of aging. Existing cable random inspection methods suffer from long testing cycles, low efficiency, high equipment wear and tear, and difficulty for grassroots units to implement independently, making it impossible to achieve full coverage inspection of delivered cables and creating blind spots in quality control.

[0004] Currently, the identification of aging and deterioration status of silicone rubber cable accessory insulation materials is mainly based on qualitative analysis. There is no systematic standard for accurate testing and quality assessment of cable insulation materials. In the field of power cables, relevant research at home and abroad mainly uses Fourier mid-infrared spectroscopy to study the aging degree of cable insulation materials. However, due to limitations in light source and the size of spectral detection components, portable testing suitable for field use cannot be achieved, keeping the relevant technology at the stage of laboratory testing and analysis verification.

[0005] Therefore, establishing a portable, non-destructive rapid testing method for insulation material performance, accelerating testing speed, improving testing efficiency, reducing manpower, and achieving real-time, rapid, and comprehensive testing of the core component materials of delivered cables and accessories is a core and challenging issue that urgently needs to be addressed to effectively ensure the quality of cables entering the network and the safe and stable operation of the power grid. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a method and system for predicting the elongation at break of silicone rubber cable accessory insulation materials, aiming to solve the problem that the prior art cannot quickly and accurately detect the elongation at break of insulation materials with full coverage.

[0007] To achieve the above objectives, the present invention provides the following solution: A method for predicting the elongation at break of silicone rubber cable accessory insulation material, the method comprising: Obtain target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; Tensile tests are performed on the target specimen to obtain the elongation at break of the target specimen. Based on near-infrared spectral data and elongation at break, a basic database of near-infrared spectral data points was obtained, and a principal component analysis regression model was established based on the basic database of near-infrared spectral data points. Based on principal component analysis regression model, the elongation at break of unknown samples of silicone rubber cable accessory insulation material is predicted.

[0008] Preferably, the method for obtaining target samples of silicone rubber cable accessory insulation material, testing the near-infrared spectrum of the target samples, and preprocessing them to obtain near-infrared spectral data includes: The silicone rubber cable accessory insulation material sample was cut into uniform and flat sheet-like samples with a thickness of 1 mm, which were used as target samples and divided into modeling set and prediction set according to the proportion. Near-infrared spectroscopy was performed on the target sample to obtain initial near-infrared spectral data; The initial near-infrared spectral data were preprocessed using the SG smoothing method and the second derivative to obtain the near-infrared spectral data.

[0009] Preferably, the method for obtaining a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break includes: Based on the preprocessed near-infrared spectral data and the fracture elongation obtained from the test in the modeling set, a competitive adaptive reweighting algorithm is used to select wavelength points with high correlation to the fracture elongation to obtain a basic database of near-infrared spectral data points.

[0010] Preferred methods for establishing principal component analysis regression models based on a near-infrared spectral data point database include: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed. Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed; Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

[0011] Preferably, methods for predicting the elongation at break of unknown samples of silicone rubber cable accessory insulation material based on principal component analysis regression models include: Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

[0012] The present invention also provides a prediction system for the elongation at break of silicone rubber cable accessory insulation material. The system is used to implement the aforementioned method and includes: a first test module, a second test module, a model building module, and a prediction module. The first testing module is used to acquire target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; The second testing module is used to perform tensile tests on the target specimen to obtain the elongation at break of the target specimen. The model building module is used to obtain a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break, and to build a principal component analysis regression model based on the basic database of near-infrared spectral data points. The prediction module is used to predict the elongation at break of unknown samples of silicone rubber cable accessory insulation material based on a principal component analysis regression model.

[0013] Preferably, the first test module includes: a sample preparation unit, a test unit, and a pretreatment unit; The sample preparation unit is used to cut the silicone rubber cable accessory insulation material sample into uniform and flat sheet-like samples with a thickness of 1 mm, which are used as target samples, and divide them into modeling set and prediction set according to the proportion. The testing unit is used to perform near-infrared spectroscopy on the target sample to obtain initial near-infrared spectral data. The preprocessing unit is used to preprocess the initial near-infrared spectral data using the SG smoothing method and the second derivative to obtain near-infrared spectral data.

[0014] Preferably, the method for obtaining a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break includes: Based on the preprocessed near-infrared spectral data and the fracture elongation obtained from the test in the modeling set, a competitive adaptive reweighting algorithm is used to select wavelength points with high correlation to the fracture elongation to obtain a basic database of near-infrared spectral data points.

[0015] Preferred methods for establishing principal component analysis regression models based on a near-infrared spectral data point database include: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed. Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed; Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

[0016] Preferably, methods for predicting the elongation at break of unknown samples of silicone rubber cable accessory insulation material based on principal component analysis regression models include: Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method and system for predicting the elongation at break of silicone rubber cable accessory insulation materials. A sheet-like sample with uniform thickness and a smooth surface is used as the target sample. The near-infrared spectrum of the target sample is measured using a UV-3600 near-infrared / visible / ultraviolet spectrophotometer. Combining the variation of elongation at break with the peak height of various characteristic functional groups, a database of near-infrared spectral characteristics and macroscopic properties of samples with different structural parameters is constructed. The correlation and interaction coefficient between the principal component spatial parameters of the near-infrared characteristic spectra in the database and the principal component spatial parameters of the macroscopic properties are studied. A matrix relationship between near-infrared spectra and elongation at break is established, enabling the prediction of elongation at break based on near-infrared spectroscopy. This invention achieves rapid and automatic scanning detection of cable insulation material cross-sections and evaluation of their mechanical properties by constructing a correlation between the near-infrared spectral characteristics of cable accessory insulation materials and the mechanical properties of the insulation materials. A portable, non-destructive rapid detection method for insulation material performance is established, accelerating detection speed, improving detection efficiency, reducing manpower, and realizing real-time, rapid, and comprehensive detection of the core component material quality of delivered cables and accessories. Attached Figure Description

[0018] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of the method for predicting the elongation at break of the insulating material of silicone rubber cable accessories according to the present invention. Figure 2 The near-infrared spectrum of the target sample model set in the wavelength range of 1100~2500 nm is shown in the embodiment of the present invention. Figure 3 This is a result image of the near-infrared spectrum of the target sample after preprocessing in an embodiment of the present invention; Figure 4 The following is a diagram showing the results of wavelength selection for the near-infrared spectrum of the target sample after pretreatment in an embodiment of the present invention. (a) is a cumulative selected frequency diagram of each wavelength of CARS-500, and (b) is a schematic diagram of the distribution of the 12 wavelength points finally selected by CARS on the near-infrared second derivative spectrum. Figure 5 This is a graph showing the correlation between the elongation at break and the near-infrared spectrum of different silicone rubber samples in the embodiments of the present invention. Figure 6 This is a comparison chart of the predicted and actual values ​​of the predicted elongation at break of the sample in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] Example 1 This invention provides a method for predicting the elongation at break of silicone rubber cable accessory insulation materials, comprising: Obtain target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; Tensile tests are performed on the target specimen to obtain the elongation at break of the target specimen. Based on near-infrared spectral data and elongation at break, a basic database of near-infrared spectral data points was obtained, and a principal component analysis regression model was established based on the basic database of near-infrared spectral data points. Based on principal component analysis regression model, the elongation at break of unknown samples of silicone rubber cable accessory insulation material is predicted.

[0023] like Figure 1 As shown, the specific implementation process of the present invention is as follows: Sample preparation: Cut the silicone rubber cable accessory insulation material sample into uniform and flat sheet-like samples with a thickness of 1 mm as target samples, and divide them into modeling set and prediction set according to the proportion.

[0024] Near-infrared spectroscopy test: The near-infrared spectrum of the target sample was tested using a UV-3600 near-infrared / visible / ultraviolet spectrophotometer. First, the background spectrum without the target sample was tested. Then, the target sample was placed in the UV-3600 near-infrared / visible / ultraviolet spectrophotometer to test its near-infrared spectrum in the wavelength range of 1100~2500nm, and the initial near-infrared spectral data were obtained. Initial near-infrared spectral preprocessing: The initial near-infrared spectrum of the target sample is smoothed and subjected to second-order derivative processing to eliminate the effects of noise, clutter peaks, and baseline drift, resulting in near-infrared spectral data. Specifically, the existing Savitzky-Golay (SG) convolution smoothing method is used. Based on the assumption that the mean of random noise within the processing window is zero, the optimal estimate is obtained by smoothing or fitting several points before and after the smoothing point. This method removes high-frequency noise and improves the signal-to-noise ratio. When smoothing spectral data using the SG smoothing method, a data window containing an odd number of spectral variables is first defined. Then, the spectral variables within the data window are fitted using the polynomial least squares method. The fitted value of the spectral variable located at the center point of the data window is obtained using the fitted polynomial. The data window is moved one by one, and the above calculation process is repeated to obtain the fitted value of the center point of the new data window. After the data window has moved through the entire spectrum, the fitted values ​​of the center points of all data windows are arranged sequentially to obtain the spectrum after SG smoothing. The data window width and the polynomial fitting order are key parameters in the SG smoothing method, which directly affect the smoothing effect. This invention uses a trial-and-error method to determine the optimal window width and polynomial fitting order.

[0025] Simultaneously, differentiating the spectral data can eliminate baseline shifts and differences between overlapping spectra, increasing visual resolution. Since high derivative orders introduce significant random errors, the second derivative method (D2) is used for baseline correction. The second derivative can eliminate drift caused by linear correlations within the same wavelength. While derivative processing improves sensitivity, it also introduces noise; therefore, it is used in conjunction with the SG smoothing method.

[0026] Furthermore, a tensile test is performed on the target specimen to obtain its elongation at break: First, a dumbbell-shaped specimen is prepared based on the target specimen. Then, a tensile test is performed using a 5kNCMT-4503 universal electronic tensile testing machine from the Mester Industrial System. The two ends of the dumbbell-shaped specimen are fixed using the upper and lower clamps of the tensile testing machine. The dumbbell-shaped specimen is stretched at a constant tensile rate. During the process, the sensor on the instrument records the force applied to the dumbbell-shaped specimen and the corresponding length of the dumbbell-shaped specimen. The recorded data is then converted into stress and strain, and the stress and strain data are transmitted to a computer and plotted on an XY coordinate graph to obtain the stress-strain curve of the dumbbell-shaped specimen. The elongation at break of the target specimen is then obtained from the curve.

[0027] Furthermore, based on near-infrared spectral data and fracture elongation, a basic database of near-infrared spectral data points was obtained: Based on the near-infrared spectral data preprocessed using Savitzky-Golay (SG) convolution smoothing and second-order derivative in the modeling set, and the fracture elongation obtained from testing, a competitive adaptive reweighting algorithm (CARS algorithm) was used to select wavelength points with high correlation to fracture elongation as the basic database of near-infrared spectral data points for subsequent establishment of the prediction model (principal component analysis regression model). Figure 4 (b) It can be seen that the wavelengths selected by the CARS algorithm are concentrated in the range of 2000~2500 nm, of which 2200~2400 nm corresponds to the CH vibration. Since the crosslinking of silicone rubber mainly relies on the free radical addition of side chain vinyl or methyl groups, the -CH2-CH2- increases. When the -CH2-CH2- increases, the degree of crosslinking of silicone rubber increases. The degree of crosslinking affects the elongation at break by influencing the molecular weight of the mobile chain segments. Therefore, it can be seen that the selected wavelengths are highly correlated with the elongation at break of silicone rubber.

[0028] Furthermore, a principal component analysis regression model was established based on a near-infrared spectral data point database: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed as shown in Equation (1). Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed as shown in equation (2); Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

[0029] MERGEFORMAT (1) MERGEFORMAT (2) Where X is the near-infrared spectral feature matrix and Y is the attribute matrix.

[0030] Let X be a data table with n sample points and p variables. Where xp is the p-th variable. For ease of derivation, assume the data table is standardized and assume a composite variable F1, where F1 is... A linear combination, i.e. , These are eigenvectors; the variance of F1 is calculated as follows: , It is the covariance matrix of X in the data table. Denotes the transpose of X. This represents the transpose of a. When all variables in X are standardized, V is the new correlation coefficient matrix of X.

[0031] Rewrite the above problem as a mathematical expression, that is, solve the optimization problem. .

[0032] The Lagrange algorithm is used to solve this problem. Let L be the Lagrange function, and let... .

[0033] Find the answer for L with respect to L. and The partial derivative of , and set it to zero, has ,at this time , It is a standardized eigenvector of V, and its corresponding eigenvalue is .according to and ,calculate , , The corresponding eigenvalues It should reach its maximum value. In other words, the calculated value should be the maximum. It is the largest eigenvalue of matrix V. The corresponding standardized feature vector. Known as the first main axis, It is called the first principal component.

[0034] Next, calculate the second principal axis. , and Orthogonal, and second only to the first principal component. Second principal component It is the second largest component carrying mutation information. The variance is Rewritten as an optimization problem, that is... .

[0035] Similar to asking The process of defining the Langron function Please ask L about With Taking the partial derivative and setting it to 0, we get... , It is the standardized eigenvector of matrix V, and its corresponding eigenvalues ​​are And so on, to obtain the h-th principal axis of the X data table. It is the h-th eigenvalue of the covariance matrix V. The corresponding standardized feature vector.

[0036] Based on the magnitude of data variation to reflect the information in the data, this invention selects m principal components. The total information carried by these m principal components is... .

[0037] Based on the above definitions and analysis steps, the calculation steps of principal component analysis can be summarized as follows: ① Standardize the data: ,in, This is the data before standardization. It is standardized data. It is a sample The sample mean; yes The standard deviation of the samples. The purpose of standardization is to make the center of the sample point set coincide with the origin of the coordinate system, while compression can eliminate spurious variation caused by different dimensions, making the analysis results more reasonable. For convenience, the standardization matrix is ​​denoted as X; ② Calculate the covariance matrix V of the standardized data matrix X.

[0038] ③ Calculate the first m eigenvalues ​​of V and the corresponding feature vectors They are required to be orthogonal; ④ Calculate the h-th principal component Fh, then we have ; In the formula: It is the spindle The j-th component. Therefore, the principal component Fh is the original variable. A linear combination of the components, with coefficients exactly equal to 1. .

[0039] Principal component transformation does not reduce the information content of the original data; the total sample variance of the spectral matrix remains unchanged before and after the transformation. .

[0040] Based on the eigenvalues ​​of the sample covariance matrix, the contribution of each principal component to the total information content is calculated. Specific statistics are calculated as follows: (1) Variance contribution rate of principal components The larger this value, the stronger the ability of the i-th principal component to synthesize information.

[0041] (2) Cumulative contribution rate of principal components This indicates the percentage of information contained in all the measurement indicators basically included by the i-th principal component.

[0042] (3) Cross-validation. In the analysis of near-infrared spectra in this invention, establishing only the principal component regression model is not the ultimate goal; it is also necessary to use the spectral measurements to estimate the performance of interest. Therefore, it is necessary to reduce the independent variables, that is, to reduce the regression equation to a model represented by the spectrum. Since the transformation from spectral data to principal components is a linear transformation, this reduction is effective.

[0043] If the principal component matrix is ​​Z, then principal component regression can be described as follows: ,in: This represents the predicted value of elongation at break; Indicates the first i One principal component; m is the number of principal components selected; These are the regression coefficients. Since principal component transformation is generally a linear transformation, Z can also be expressed as a linear function of the spectral data x, and the principal component regression equation can also be transformed into... When a regression model is built using all m principal components and then restored, the resulting multiple linear model will be completely equivalent to the multiple linear regression model.

[0044] When building a principal component analysis regression model, the most critical issue is determining the number of principal components. In this invention, interactive validation is employed, using the data matrices X and Y of the calibration set itself to simulate data from the unknown sample set, thereby evaluating the actual predictive ability of the built model. The "calibration set" originates from the aforementioned "modeling set." Specifically, the modeling set is further randomly divided into... n A subset, of which n One subset is used as the calibration set, and the remaining subset is used as the validation set for verification. This partitioning aims to internally evaluate model performance and avoid overfitting.

[0045] The interactive verification method divides a known standard measurement matrix Y and its corresponding material property matrix X into n sample subsets. One sample is selected from these n subsets as the test sample, and the remaining (n-1) samples are used as the calibration set. Therefore, a calibration model is established using the (n-1) calibration set samples through calibration calculations. During the modeling process, samples are selected from each subset... The model is used to predict the y-value of the retained test samples, with y representing the actual measured value of elongation at break in the cross-validation. These are the model predictions. By taking n subsets in turn as test samples and performing n correction prediction processes, we can obtain the sum of squared prediction errors (PRESS) for different values. In the formula, Indicates the first i The sample at the th j The predicted spectral values ​​for each wavelength point are obtained through model inversion.

[0046] When building a prediction model, the number of principal components typically starts from 1. As the number of principal components increases, the fitting error decreases while improving the model's prediction accuracy. At this stage, the PRESS value decreases, indicating that the model is not fully fitted. When the PRESS value reaches its lowest point and then begins to rise again, it means that after this point, the added principal components are noise components unrelated to the measured component, leading to increased prediction error, overfitting, and reduced predictive ability. The principal components corresponding to the lowest prediction error are the optimal principal components.

[0047] Furthermore, based on the principal component analysis regression model, the elongation at break of unknown samples of silicone rubber cable accessory insulation material is predicted.

[0048] Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

[0049] In summary, this invention provides a method for predicting the elongation at break of silicone rubber cable accessory insulation materials. A sheet-like sample with uniform thickness and a smooth surface is used as the target sample. The near-infrared spectrum of the target sample is measured using a UV-3600 near-infrared / visible / ultraviolet spectrophotometer. Combining the variation of elongation at break with the peak height of each characteristic functional group, a database of near-infrared spectral characteristics and macroscopic properties of samples with different structural parameters is constructed. The correlation and interaction coefficient between the principal component spatial parameters of the near-infrared characteristic spectra in the database and the principal component spatial parameters of the macroscopic properties are studied. A matrix relationship between near-infrared spectra and elongation at break is established, enabling the prediction of elongation at break based on near-infrared spectroscopy. This invention achieves rapid and automatic scanning detection of cable insulation material cross-sections and evaluation of their mechanical properties by constructing a correlation between the near-infrared spectral characteristics of cable accessory insulation materials and the mechanical properties of the insulation materials. A portable, non-destructive rapid detection method for insulation material performance is established, accelerating detection speed, improving detection efficiency, reducing manpower, and realizing real-time, rapid, and comprehensive detection of the core component material quality of delivered cables and accessories.

[0050] Example 2 This embodiment uses specific experimental data to illustrate the specific implementation process of the method described in the foregoing embodiments.

[0051] This embodiment takes 20 non-operational 220kV domestic silicone rubber cable accessories as an example, of which 16 are used as the modeling set and 4 are used as the prediction set. Their near-infrared spectra are tested respectively. In this embodiment, a UV-3600 near-infrared / visible / ultraviolet spectrophotometer is used. The target sample is placed in the sample chamber and the transmission mode is adopted with a resolution of 0.1nm.

[0052] Specifically, the following steps are included: 1) Prepare the sample: Cut a flat sheet with a size of 70×30mm and a thickness of 1mm from the cable accessory insulation as the target sample.

[0053] 2) Near-infrared spectrum detection: First, the background spectrum without the target sample was detected using a UV-3600 near-infrared / visible / ultraviolet spectrophotometer. Then, the near-infrared spectrum of the target sample in the wavelength range of 1100nm~2500nm was measured using the same UV-3600 spectrophotometer. The near-infrared spectrum of the model set is shown below. Figure 2 As shown.

[0054] 3) Preprocessing of near-infrared spectra: The measured near-infrared spectra are smoothed and subjected to second-order derivative preprocessing to eliminate the effects of noise, cluttered peaks, and baseline drift. The preprocessed results are as follows: Figure 3 As shown. Figure 3 It contains the near-infrared spectral curves of 16 silicone rubber samples after pretreatment.

[0055] 4) Elongation at break test: Dumbbell-shaped silicone rubber specimens were prepared and subjected to tensile tests at room temperature. The test was set to tensile direction and large deformation. The stress-strain curves were obtained, and the elongation at break of the specimens was obtained. The elongation at break test results of the model set and the prediction set are shown in Table 1 and Table 2, respectively.

[0056] Table 1. Results of Elongation at Break Test for the Modeling Set Table 2. Results of Elongation at Break Test for Prediction Set 5) Wavelength Optimization: Wavelength optimization based on the CARS algorithm is performed on the preprocessed near-infrared spectral data to obtain the wavelength points with the highest correlation to the elongation at break of silicone rubber, thus obtaining a basic database of near-infrared spectral data points. The results are as follows: Figure 4 As shown.

[0057] 6) Establishing a principal component analysis regression model: Based on the near-infrared spectral data points database after wavelength optimization and the actual measured elongation at break, a principal component analysis regression model is established to obtain the correlation between near-infrared spectroscopy and elongation at break. The correlation curve is shown below. Figure 5 As shown, the prediction model has a goodness of fit R² of 0.9726 and a root mean square cross-validation error (RMSECV) of 10.1722, indicating that the model has high prediction accuracy.

[0058] 7) Predicting the elongation at break of unknown samples: The initial near-infrared spectra of the prediction set (unknown samples) in the target specimen are collected, preprocessed, and then input into the established principal component analysis regression model to obtain the predicted values ​​of the elongation at break of the prediction set. The comparison results between the predicted values ​​and the actual values ​​are as follows: Figure 6 As shown in Table 3, the prediction errors for the elongation at break of the four silicone rubber samples were all within 4%, indicating that the prediction model has high accuracy.

[0059] Table 3. Prediction results of elongation at break of the prediction set samples. Example 3 Based on the same inventive concept, the present invention also provides a prediction system for the breaking elongation of silicone rubber cable accessory insulation material, used to implement the method described in the foregoing embodiments. The system includes: a first test module, a second test module, a model building module, and a prediction module. The first testing module is used to acquire target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; The second testing module is used to perform tensile tests on the target specimen to obtain the elongation at break of the target specimen. The model building module is used to obtain a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break, and to build a principal component analysis regression model based on the basic database of near-infrared spectral data points. The prediction module is used to predict the elongation at break of unknown samples of silicone rubber cable accessory insulation material based on a principal component analysis regression model.

[0060] Furthermore, the first test module includes: a sample preparation unit, a test unit, and a pretreatment unit; The sample preparation unit is used to cut the silicone rubber cable accessory insulation material sample into uniform and flat sheet-like samples with a thickness of 1 mm, which are used as target samples, and divide them into modeling set and prediction set according to the proportion. The testing unit is used to perform near-infrared spectroscopy on the target sample to obtain initial near-infrared spectral data. The preprocessing unit is used to preprocess the initial near-infrared spectral data using the SG smoothing method and the second derivative to obtain near-infrared spectral data.

[0061] Furthermore, methods for obtaining a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break include: Based on the preprocessed near-infrared spectral data and the fracture elongation obtained from the test in the modeling set, a competitive adaptive reweighting algorithm is used to select wavelength points with high correlation to the fracture elongation to obtain a basic database of near-infrared spectral data points.

[0062] Furthermore, methods for establishing principal component analysis regression models based on a near-infrared spectral data point database include: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed. Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed; Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

[0063] Furthermore, based on principal component analysis regression models, methods for predicting the elongation at break of unknown samples of silicone rubber cable accessory insulation materials include: Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

[0064] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for predicting the elongation at break of silicone rubber cable accessory insulation material, characterized in that, The method includes: Obtain target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; Tensile tests are performed on the target specimen to obtain the elongation at break of the target specimen. Based on near-infrared spectral data and elongation at break, a basic database of near-infrared spectral data points was obtained, and a principal component analysis regression model was established based on the basic database of near-infrared spectral data points. Based on principal component analysis regression model, the elongation at break of unknown samples of silicone rubber cable accessory insulation material is predicted.

2. The method according to claim 1, characterized in that, Methods for obtaining target samples of silicone rubber cable accessory insulation material, testing the near-infrared spectrum of the target samples, and preprocessing the spectrum to obtain near-infrared spectral data include: The silicone rubber cable accessory insulation material sample was cut into uniform and flat sheet-like samples with a thickness of 1 mm, which were used as target samples and divided into modeling set and prediction set according to the proportion. Near-infrared spectroscopy was performed on the target sample to obtain initial near-infrared spectral data; The initial near-infrared spectral data were preprocessed using the SG smoothing method and the second derivative to obtain the near-infrared spectral data.

3. The method according to claim 2, characterized in that, Methods for obtaining a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break include: Based on the preprocessed near-infrared spectral data and the fracture elongation obtained from the test in the modeling set, a competitive adaptive reweighting algorithm is used to select wavelength points with high correlation to the fracture elongation to obtain a basic database of near-infrared spectral data points.

4. The method according to claim 3, characterized in that, Methods for establishing principal component analysis regression models based on a near-infrared spectral data point database include: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed. Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed; Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

5. The method according to claim 4, characterized in that, Methods for predicting the elongation at break of unknown samples of silicone rubber cable accessory insulation materials based on principal component analysis regression models include: Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

6. A system for predicting the elongation at break of silicone rubber cable accessory insulation material, said system being used to implement the method according to any one of claims 1-5, characterized in that, The system includes: a first testing module, a second testing module, a model building module, and a prediction module; The first testing module is used to acquire target samples of silicone rubber cable accessory insulation material, test the near-infrared spectrum of the target samples and perform preprocessing to obtain near-infrared spectral data; The second testing module is used to perform tensile tests on the target specimen to obtain the elongation at break of the target specimen. The model building module is used to obtain a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break, and to build a principal component analysis regression model based on the basic database of near-infrared spectral data points. The prediction module is used to predict the elongation at break of unknown samples of silicone rubber cable accessory insulation material based on a principal component analysis regression model.

7. The system according to claim 6, characterized in that, The first testing module includes: a sample preparation unit, a testing unit, and a pretreatment unit; The sample preparation unit is used to cut the silicone rubber cable accessory insulation material sample into uniform and flat sheet-like samples with a thickness of 1 mm, which are used as target samples, and divide them into modeling set and prediction set according to the proportion. The testing unit is used to perform near-infrared spectroscopy on the target sample to obtain initial near-infrared spectral data. The preprocessing unit is used to preprocess the initial near-infrared spectral data using the SG smoothing method and the second derivative to obtain near-infrared spectral data.

8. The system according to claim 7, characterized in that, Methods for obtaining a basic database of near-infrared spectral data points based on near-infrared spectral data and elongation at break include: Based on the preprocessed near-infrared spectral data and the fracture elongation obtained from the test in the modeling set, a competitive adaptive reweighting algorithm is used to select wavelength points with high correlation to the fracture elongation to obtain a basic database of near-infrared spectral data points.

9. The system according to claim 8, characterized in that, Methods for establishing principal component analysis regression models based on a near-infrared spectral data point database include: Based on the near-infrared spectral data of the target sample after pretreatment and the basic database of near-infrared spectral data points, a near-infrared spectral feature matrix is ​​constructed. Based on the elongation at break of the target specimen, an attribute matrix is ​​constructed; Based on the infrared spectral feature matrix and attribute matrix, a mathematical model of the relationship between the elongation at break and the spectral data is established using principal component analysis regression method, thus obtaining the principal component analysis regression model.

10. The system according to claim 9, characterized in that, Methods for predicting the elongation at break of unknown samples of silicone rubber cable accessory insulation materials based on principal component analysis regression models include: Near-infrared spectroscopy was performed on the predicted set of the target sample and preprocessed to obtain near-infrared spectral data. Near-infrared spectral data are input into the established principal component analysis regression model to obtain the predicted values ​​of the target attributes of the prediction set, which are the elongation at break of the prediction set.

Images