Method for rapidly testing authenticity of degradable plastic on basis of near-infrared technology

Abstract

Disclosed in the present invention is a method for rapidly testing the authenticity of degradable plastic on the basis of near-infrared technology, comprising the following steps: step S1, using packaging bag samples of known materials to collect spectra, and using a kennard-Stone method to perform division to obtain a training set and a prediction set; step S2, by means of the spectra in the training set in combination with analysis based on a normalization algorithm, using a support vector machine algorithm to perform modeling, and using sample data in the training set to train a support vector machine model; and step S3, using the prediction set and inputting same into the trained support vector machine model to obtain a prediction result of the model, and performing analysis and adjustment to obtain an optimized support vector machine model. The method of the present invention features a simple testing process and high identification accuracy, and enables rapid determination of whether a material is a degradable material and identification of the specific type of the degradable material; additionally, the method saves a large amount of manpower, material resources and financial resources, achieving high economic benefits.

Description

A method for rapid detection of the authenticity of biodegradable plastics using near-infrared technology Technical Field

[0001] The present invention relates to the technical field of biodegradable plastic detection, and in particular to a method for rapidly detecting and identifying the authenticity of biodegradable plastic using near-infrared technology. Background Art

[0002] Biodegradable plastics can be broken down into smaller molecules under certain conditions, reducing their environmental impact. However, genuine and counterfeit biodegradable materials are mixed on the market, necessitating a rapid detection method to identify these materials.

[0003] Traditional testing methods for degradable plastics include physical property testing, chemical analysis, and thermal analysis. Physical property testing involves measuring the density, melting point, and melt fluidity of degradable plastics to determine their degree of degradation. These tests are usually destructive to the samples and cannot provide detailed molecular structure information. Chemical analysis involves decomposing degradable plastics using chemical methods and analyzing the products to determine the degree of degradation. For example, the degree of degradation can be assessed by measuring changes in solubility, gas release, or product generation. However, these methods take a long time and may cause environmental pollution when processing samples. Thermal analysis techniques such as thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) can evaluate changes in the properties of degradable plastics by monitoring the mass changes or thermal behavior of samples at different temperatures.

[0004] Near-infrared spectroscopy is a simple, rapid, and nondestructive detection technique that enables rapid, simultaneous quantitative analysis of multiple components in a very short time with exceptionally high accuracy. This technique detects sample information by measuring the transmittance or reflectance of a sample at different wavelengths. A complete near-infrared spectrum consists of a series of absorption bands whose intensities vary as specific functional groups in the sample absorb energy. Applying near-infrared detection technology for the rapid detection and identification of biodegradable plastics is not only highly efficient and non-destructive, but also capable of accurately distinguishing different types of biodegradable plastics, providing reliable support for sustainable material selection and environmentally friendly practices. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a method for quickly detecting the authenticity of degradable plastics using near-infrared technology. The present invention uses near-infrared spectroscopy technology combined with spectrum processing technology and support vector machine method to establish a model, thereby realizing the authenticity judgment of most degradable plastics on the market.

[0006] To achieve the above purpose, the present invention provides the following technical solutions

[0007] A method for rapidly detecting the authenticity of degradable plastics using near-infrared technology comprises the following steps:

[0008] Step S1: Spectra of a large number of plastic films with known materials such as PBAT, PP, PLA, PBS, LDPE, HDPE, LLDPE, and PET are collected, and the Kennard-Stone method is used to divide the training set and the prediction set;

[0009] Step S2: Extract features from the normalized spectra using the training set spectra and the normalization algorithm. The features include absorption peak positions and absorption intensities. The support vector machine algorithm is used to model the features, and the sample data in the training set is used to train the support vector machine model.

[0010] Step S3: Use the prediction set to input into the trained support vector machine model to obtain the model's prediction results, and analyze and adjust to obtain an optimized support vector machine model, which is used to quickly detect the authenticity of degradable plastics.

[0011] As a further solution of the present invention, step S1 specifically includes: extracting features from the spectral data of each sample, and selecting a spectral interval with significant information based on the characteristics of the spectral data of known materials, wherein PBAT has a characteristic peak at 1175nm and 1400nm in the plastic film of PP, PLA, PBS, LDPE, HDPE, LLDPE, and PET, PBAT has a characteristic peak at 1198nm and 1300nm, PP has a characteristic peak at 1163nm and 1387nm, PBS has a characteristic peak at 1170nm and 1398nm, and LDPE has a characteristic peak at 1203nm. m and 1412nm, HDPE has one characteristic peak each at 1204nm and 1407nm, LLDPE has one characteristic peak each at 1204nm and 1412nm, and PET has one characteristic peak each at 1151nm and 1405nm; thus, the near-infrared band of 1110-1500nm is selected for modeling to distinguish the materials of plastic films; the Euclidean distance is used to calculate the distance between each sample and other samples; all samples are sorted according to the calculated distance to create a distance matrix between samples; the first N samples are selected as the training set, and M samples are selected from the remaining samples as the prediction set.

[0012] As a further solution of the present invention, in step S2: before modeling, analysis is performed through normalization and SVM algorithm.

[0013] As a further solution of the present invention, the normalization process is: determining which features need to be normalized; for each selected feature, calculating its statistical information; and selecting an appropriate normalization method based on the requirements of the problem and the distribution of the data.

[0014] As a further solution of the present invention, the normalization method includes: minimum-maximum scaling: linearly scaling the data to a specified range, and for each feature x, normalizing using the following formula: ,in is the normalized eigenvalue, x is the original eigenvalue, is the minimum value of the feature, is the maximum value of the feature; Z-score normalization: transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1. For each feature x, use the following formula for standardization: , is the normalized eigenvalue, is the original eigenvalue, is the mean of the feature, is the standard deviation of the feature.

[0015] As a further aspect of the present invention, the normalization process further includes: using the normalized data for training and testing a machine learning algorithm.

[0016] As a further solution of the present invention, in step S3, the training set is used to train the SVM model. During the training process, the SVM will find an optimal hyperplane to minimize the classification error; the prediction set is used and input into the trained SVM model to obtain the prediction result of the model.

[0017] As a further solution of the present invention, a prediction error is calculated based on the prediction result and the actual value, and a performance index is calculated based on the prediction error and the actual value.

[0018] As a further solution of the present invention, the model is evaluated based on the calculated performance indicators to analyze the performance and accuracy of the model. If the performance of the model does not meet the requirements, the above training and evaluation process is repeated until an optimized support vector machine model is obtained.

[0019] As a further solution of the present invention, the model is optimized by a recursive feature elimination method, the sample data in the training set is used to train the support vector machine model to establish the prediction ability of the model, and finally the sample data in the prediction set is used to test the performance of the SVM model.

[0020] The present invention has the following beneficial effects:

[0021] This method collects spectra from packaging bags of known materials, normalizes them, and constructs a model using the Support Vector Machine (SVM) algorithm. This model is then optimized using Recursive Feature Elimination (RFE). Sample data from the training set is then used to train the SVM model to establish its predictive capabilities. Finally, sample data from the prediction set is used to test the SVM model's performance. This method not only simplifies the detection process and achieves high accuracy, but also allows for rapid analysis of whether a material is biodegradable and what type of material it is. It also saves significant manpower, material, and financial resources, resulting in high economic benefits.

[0022] In order to more clearly illustrate the structural features and effects of the present invention, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] BRIEF DESCRIPTION OF THE DRAWINGS

[0024] FIG1 is a flow chart of a modeling method for detecting degradable materials using near-infrared technology in the present invention.

[0025] DETAILED DESCRIPTION

[0026] The present invention will be further described below with reference to the accompanying drawings and related knowledge, and described clearly and completely. Obviously, the described applications are only part of the embodiments of the present invention, rather than all of the embodiments.

[0027] Biodegradable plastics can break down into smaller molecules under specific conditions, reducing their environmental impact. However, genuine and counterfeit biodegradable materials on the market are often mixed, necessitating a rapid detection method to identify these materials. This invention utilizes near-infrared spectroscopy combined with spectral processing and support vector machine methods to develop a model that can verify the authenticity of most biodegradable plastics on the market.

[0028] In order to solve the above problems, the present invention provides a method for quickly detecting the authenticity of degradable plastics using near-infrared technology, as shown in FIG1 , comprising the following steps:

[0029] Step S1: Spectra of packaging bag samples of known materials are collected and the Kennard-Stone method is used to divide the training set and the prediction set;

[0030] Specifically, spectra were collected from packaging bag samples of known materials (including both degradable and non-degradable materials), and the Kennard-Stone method was used to divide the training set into a prediction set. The advantage of the Kennard-Stone method is that it ensures that both the training set and the prediction set are representative, which helps reduce the risk of overfitting and improve the model's generalization performance.

[0031] Feature extraction: Extract features from the spectral data of each sample. Based on the known characteristics of the material's spectral data, select spectral ranges that contain significant information. In this implementation, a specific wavelength range is selected, which typically contains information about the material's characteristics. These features are used to train and test the machine learning model.

[0032] Distance metric: Use Euclidean distance to calculate the distance between each sample and other samples. Generally speaking, the smaller the Euclidean distance, the higher the similarity between the two samples. In two-dimensional space, Euclidean distance .

[0033] Sample sorting: Sort all samples according to the calculated distance to create a distance matrix between samples.

[0034] Training set selection: Select the first N (70%) samples as the training set. These N samples will be used to train the support vector machine (SVM) model.

[0035] Prediction set selection: Select M (30%) samples from the remaining samples as the prediction set. These M samples will be used to test and verify the performance of the SVM model.

[0036] Step S2: using the spectrum of the training set and the analysis of the normalization algorithm, a support vector machine algorithm is used to build a model, and the sample data in the training set is used to train the support vector machine model;

[0037] Specifically, the Support Vector Machine (SVM) algorithm was used to build a model based on the spectra of the training set, combined with a normalization algorithm. Practical experience shows that SVMs offer significant advantages in quantitative regression analysis, particularly in solving small sample, nonlinear, and high-dimensional pattern recognition problems. They can effectively process datasets with numerous features without being limited by dimensionality. To increase the contribution of each sample feature to the model, recursive feature elimination (RFE) was used for feature selection, filtering out unimportant features.

[0038] Step S3: Use the prediction set to input into the trained support vector machine model to obtain the prediction results of the model, and analyze and adjust the optimized support vector machine model.

[0039] Specifically, the support vector machine model is trained using the sample data in the training set to establish the model's predictive capabilities. Before modeling, normalization and SVM algorithm analysis are performed.

[0040] Data normalization is an important step in data preprocessing. It involves scaling the value ranges of different features to the same scale so that machine learning algorithms can better process the data. This involves selecting the features to be normalized, calculating their statistics (such as minimum and maximum values ​​or mean and standard deviation), and then using appropriate methods to map the feature values ​​to a specified range, such as [0, 1] or a standard normal distribution. This helps prevent features of different scales from having an unfair impact on the model, improving the performance and stability of machine learning models.

[0041] Normalization process:

[0042] Select features: First, determine which features need to be normalized. Typically, continuous numerical features need to be normalized, while categorical features usually do not.

[0043] Collect statistics: For each selected feature, calculate its statistics, including the minimum (Min) and maximum (Max), or the mean (Mean) and standard deviation (Standard Deviation).

[0044] Choose a normalization method: Choose an appropriate normalization method based on the problem requirements and the distribution of the data. Two common normalization methods are min-max scaling and standardization (Z-score normalization).

[0045] a. Minimum-maximum scaling: This method linearly scales the data to a specified range, usually [0, 1] or [-1, 1]. For each feature x, the following formula is used for normalization: ,here," " is the normalized eigenvalue, x is the original eigenvalue, is the minimum value of the feature, is the maximum value of the feature.

[0046] b. Standardization (Z-score normalization): This method transforms the data into a standard normal distribution with a mean of 0 and a standard deviation of 1. For each feature x, the following formula is used for standardization: ,here," ” is the standardized eigenvalue, is the original eigenvalue, is the mean of the feature, is the standard deviation of the feature.

[0047] Considering that the data needs to be input into the SVM model for processing later, this distance-based machine learning algorithm mostly requires the data set to obey the standard normal distribution (Gaussian distribution) with a mean of 0 and a variance of 1. Therefore, the standardization method is chosen here to preprocess the data set.

[0048] (4) Apply normalization: Use normalized data for training and testing machine learning algorithms.

[0049] Support vector machine (SVM) is a machine learning algorithm based on statistical learning theory. Its core idea is to find a decision boundary with the largest margin to separate data points of different categories to improve the generalization ability of the model. The support vector is the data point closest to the boundary. The kernel function can be used to map data to a high-dimensional space to make linearly inseparable data linearly separable. It is suitable for classification and regression problems, but the parameters and kernel function need to be carefully adjusted to obtain the best performance.

[0050] Given a set of training sample data T={(x1,y1),(x2,y2),...,(x l ,y l )}where x i ∈R n , is an n-dimensional vector, y i ∈R, i=1,2,...,l, l is the number of samples, find the objective function y(x) that best fits the input-output relationship of the training sample, and predict the y value corresponding to the input x. If the objective function is a linear function, such as y(x)=ω T x+b, corresponds to linear regression; if it is a nonlinear function, it corresponds to nonlinear regression. For linear regression, all data are within the relaxation factor ξ i ,ξ i * The function is used to fit under the requirements of and accuracy ε, that is, the problem is transformed into under the constraints:

[0051]

[0052] Find the minimum of the following function: , Where C is the penalty factor, a constant greater than zero, indicating the degree of penalty for samples exceeding the error ε. The smaller the C value, the smaller the penalty for the empirical error. Conversely, the larger the C value, the greater the penalty for the empirical error. The Lagrange optimization method can be used to obtain the dual optimization problem, that is, under the following constraints: ,

[0053] By maximizing W( , * ) to obtain the Lagrange factor :

[0054]

[0055] The regression function is obtained as:

[0056] For nonlinear problems, the main solution of SVM is to introduce nonlinear mapping through kernel function, convert the nonlinear problem of input space into a linear problem in a high-dimensional space, and realize nonlinear regression of sample space in high-dimensional feature space.

[0057] In this scheme, the SVM algorithm models and performs classification or regression tasks by finding an optimal hyperplane to maintain the maximum margin between spectral data points of different categories while minimizing the classification error.

[0058] Recursive Feature Elimination (RFE) is a feature selection method that recursively selects the best feature subset by considering smaller and smaller subsets of features. Specifically, a model is selected from the original feature set, and then the least important features are eliminated based on some criteria (such as coefficient, mean squared error, etc.). This process is then repeated on the remaining feature set until the desired number of features is reached.

[0059] The main steps are as follows:

[0060] Given a dataset D, each sample consists of a feature vector x and a label y, and the dimension of the feature vector is p. Use a model f(x;w) to fit the data, where w is the weight parameter of the model.

[0061] (1) Initialize the feature set S to all features, that is, S = {1, 2, …, p}.

[0062] (2) Use the current feature set S to train the model f(x;w) and calculate the importance score of each feature.

[0063] (3) Sort the feature set S by importance score from large to small, denoted as S={s1​,s2​,…,s p ​}.

[0064] (4) Select the feature s1 with the highest importance score and delete it from the feature set S to obtain a new feature set S′.

[0065] (5) If the number of features of S′ is 0, stop the recursion and output the current feature set S; otherwise, return to step S2.

[0066] In step S2, the method for calculating the importance score of each feature depends on the model used. For support vector machine (SVM) models, the weight coefficient of each feature is used. By recursively executing the above steps, RFE can gradually eliminate unimportant features, ultimately obtaining a smaller feature set. This feature set can be used to build a more efficient model.

[0067] In step S3, the optimized support vector machine model is analyzed and adjusted, specifically in the following manner:

[0068] Model performance evaluation

[0069] (1) Use the training set (containing N samples) to train the above SVM model. During the training process, SVM will find an optimal hyperplane to minimize the classification error.

[0070] (2) Test model: Use the prediction set (containing M samples) and input it into the trained SVM model to obtain the model's prediction results. The model will assign a class label output to each sample.

[0071] (3) Calculate the prediction error: Calculate the prediction error based on the prediction result and the actual value. Commonly used error indicators include mean square error (MSE) and root mean square error (RMSE).

[0072] (4) Calculate performance indicators: Calculate performance indicators such as accuracy based on the prediction error and actual value.

[0073] (5) Analyze model performance: Evaluate the model based on the calculated performance indicators and analyze the performance and accuracy of the model.

[0074] (6) Adjust the model: If the performance of the model does not meet the requirements, you can adjust the SVM model parameters to optimize the performance. Repeat the above steps until you get an SVM model with better performance.

[0075] This method collects spectra from packaging bags of known materials, normalizes them, and constructs a model using the Support Vector Machine (SVM) algorithm. This model is then optimized using Recursive Feature Elimination (RFE). Sample data from the training set is then used to train the SVM model to establish its predictive capabilities. Finally, sample data from the prediction set is used to test the SVM model's performance. This method not only simplifies the detection process and achieves high accuracy, but also allows for rapid analysis of whether a material is biodegradable and what type of material it is. It also saves significant manpower, material, and financial resources, resulting in high economic benefits.

[0076] The present invention provides the following specific embodiments Example 1

[0077] 1 , a method for rapidly detecting the authenticity of degradable plastics using near-infrared technology specifically includes the following steps:

[0078] Step S1: Spectra are collected using samples of packaging bags of known materials, and the Kennard-Stone method is used to divide the training set and the prediction set. For example, spectra are collected using a large number of plastic bags of known materials such as PE, PVC, PP, PET, EVA, PS, PBAT, and PLA, and the Kennard-Stone method is used to divide the training set and the prediction set;

[0079] Step S2: Through the spectrum of the training set, combined with the analysis of the normalization algorithm, and the extraction of features from the normalized spectrum, mainly including the absorption peak position and absorption intensity. The absorption peak position is the wavelength position corresponding to the strong absorption region in the spectral data, corresponding to the vibration frequency of a specific chemical bond or molecular structure. The absorption intensity represents the amplitude or depth of the absorption peak, reflecting the relative content or concentration of a specific chemical bond or molecular structure. PBAT has one characteristic peak each at 1175nm and 1400nm, and PP has one characteristic peak at 1198nm. 1100nm and 1500nm, respectively, there is a characteristic peak at 1110nm and 1500nm, PLA has a characteristic peak at 1163nm and 1387nm, PBS has a characteristic peak at 1170nm and 1398nm, LDPE has a characteristic peak at 1203nm and 1412nm, HDPE has a characteristic peak at 1204nm and 1407nm, LLDPE has a characteristic peak at 1204nm and 1412nm, and PET has a characteristic peak at 1151nm and 1405nm. Therefore, the near-infrared band of 1110-1500nm was selected for modeling to distinguish the above 8 plastics; the support vector machine algorithm was used for modeling, and the sample data in the training set was used to train the support vector machine model;

[0080] The training set and prediction set are divided by using the Kennard-Stone method. The training set trains and optimizes the model, while the prediction set is used to evaluate the model's predictive and generalization capabilities. The feature data of the training set is used to train the SVM classifier and optimize the feature subset, and then feature selection is performed based on the weight of each sample feature. The Recursive Feature Elimination (RFE) method is used to gradually select the most important k features (the number of features is determined by the experimental conditions) by repeatedly training the model and deleting the least important features. During the RFE iteration process, the feature data of the training set is used to train the classifier and evaluate the performance indicators of each iteration. During the iteration process, methods such as cross-validation can be used to evaluate the performance of the model and select the best feature subset;

[0081] Step S3: The prediction set is input into the trained support vector machine model to obtain the model's prediction results. The optimized support vector machine model is then analyzed and adjusted. The performance of the final feature subset and SVM classifier is evaluated using an independent prediction set. Model parameters are further adjusted and optimized by comparing metrics such as classification accuracy and recall on the prediction set. After feature extraction, RFE iterations, feature optimization, and validation, the final feature subset and SVM classifier model can be validated on both the training and prediction sets. These models can be used in subsequent biodegradable plastic identification tasks.

[0082] The optimized SVM model is trained based on the selected important features.

[0083] Model Performance Evaluation and Feedback: Sample data from the prediction set is used to test the performance of the SVM model, including evaluation of performance indicators such as model accuracy and precision. Based on the output of the SVM model, the material and biodegradability of the packaging bag being tested are determined. The method of this invention can quickly determine not only whether a material is biodegradable, but also what type of biodegradable material it is.

[0084] The technical principles of the present invention have been described above in conjunction with specific embodiments, which are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention fall within the scope of protection of the present invention. Those skilled in the art will be able to conceive of other specific embodiments of the present invention without inventive effort, and such methods will fall within the scope of protection of the present invention.

Claims

1. A method for rapidly detecting the authenticity of degradable plastics using near-infrared technology, characterized in that: The following steps are involved: Step S1: Spectra of a large number of plastic films with known materials such as PBAT, PP, PLA, PBS, LDPE, HDPE, LLDPE, and PET are collected, and the Kennard-Stone method is used to divide the training set and the prediction set; Step S2: Extract features from the normalized spectra using the training set spectra and the normalization algorithm. The features include absorption peak positions and absorption intensities. The support vector machine algorithm is used to model the features, and the sample data in the training set is used to train the support vector machine model. Step S3: Use the prediction set to input into the trained support vector machine model to obtain the model's prediction results, and analyze and adjust to obtain an optimized support vector machine model, which is used to quickly detect the authenticity of degradable plastics.

2. The method for rapidly detecting the authenticity of degradable plastics using near-infrared technology according to claim 1, characterized in that: The step S1 specifically comprises: extracting features from the spectral data of each sample, and selecting a spectral interval with significant information according to the characteristics of the spectral data of known materials, wherein PBAT has one characteristic peak at 1175nm and 1400nm, PP has one characteristic peak at 1198nm and 1300nm, PLA has one characteristic peak at 1163nm and 1387nm, PBS has one characteristic peak at 1170nm and 1398nm, LDPE has one characteristic peak at 1203nm and 1412nm, and PP has one characteristic peak at 1198nm and 1300nm. There is one characteristic peak at each nm, HDPE has one characteristic peak at each 1204nm and 1407nm, LLDPE has one characteristic peak at each 1204nm and 1412nm, and PET has one characteristic peak at each 1151nm and 1405nm; thus, the near-infrared band of 1110-1500nm is selected for modeling to distinguish the materials of plastic films; the Euclidean distance is used to calculate the distance between each sample and other samples; all samples are sorted according to the calculated distance to create a distance matrix between samples; the first N samples are selected as the training set, and M samples are selected from the remaining samples as the prediction set.

3. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 1, characterized in that: In the step S2: before modeling, normalization and SVM algorithm analysis are performed.

4. The method for rapidly detecting the authenticity of degradable plastics using near-infrared technology as claimed in claim 3, characterized in that: The normalization process is: determine which features need to be normalized; for each selected feature, calculate its statistical information; select an appropriate normalization method based on the requirements of the problem and the distribution of the data.

5. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 4, characterized in that: Normalization methods include: Minimum-maximum scaling: linearly scale the data to the specified range. For each feature x, normalize it using the following formula: ,in is the normalized eigenvalue, x is the original eigenvalue, is the minimum value of the feature, is the maximum value of the feature; Z-score normalization: transform the data into a standard normal distribution with a mean of 0 and a standard deviation of 1. For each feature x, use the following formula for standardization: , is the normalized eigenvalue, is the original eigenvalue, is the mean of the feature, is the standard deviation of the feature.

6. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 5, characterized in that: The normalization process also includes: using the normalized data for training and testing machine learning algorithms.

7. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 6, characterized in that: In step S3, the training set is used to train the SVM model. During the training process, the SVM will find an optimal hyperplane to minimize the classification error; the prediction set is used and input into the trained SVM model to obtain the prediction result of the model.

8. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 7, characterized in that: The prediction error is calculated based on the prediction result and the actual value, and the performance index is calculated based on the prediction error and the actual value.

9. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology as claimed in claim 8, characterized in that: The model is evaluated based on the calculated performance indicators to analyze the performance and accuracy of the model. If the performance of the model does not meet the requirements, the above training and evaluation process is repeated until the optimized support vector machine model is obtained.

10. The method for rapidly detecting the authenticity of degradable plastics using near infrared technology according to claim 6, characterized in that: The model is optimized by recursive feature elimination method, and the sample data in the training set is used to train the support vector machine model to establish the prediction ability of the model. Finally, the sample data in the prediction set is used to test the performance of the SVM model.

Images