Quantitative detection method of Brassica rapa polysaccharide based on near infrared spectroscopy and chemometrics

Abstract

The invention discloses a quick quantitative detection method for Brassica rapa polysaccharide based on near infrared spectroscopy and chemometrics, and belongs to the technical field of polysaccharide detection of agricultural products. According to the method, a near-infrared spectrometer is used for collecting a sample to prepare spectral information, PyCharm software is used for optimizing an original spectrum, a PLS quantitative prediction model between the polysaccharide content and the spectral information is established, and the performance and predictive capacity of the model are evaluated, so that the accurate and rapid detection of the Brassica rapa polysaccharide content by the near-infrared spectroscopy is realized. According to the quantitative regression model established by the method, R2c and R2p are respectively 0.879 and 0.836, the corrected root mean square error RMSEC and the predicted root mean square error RMSEP are respectively 2.560 and 2.974, the predicted relative analysis error RPD is 2.474, each index is ideal, the model fitting degree is good, the prediction effect is better, and the precision and the stability both meet the expected use requirements.

Description

Technical Field

[0001] The present invention belongs to the technical field of polysaccharide detection in agricultural products, and specifically relates to a rapid quantitative detection method for chamagu polysaccharide based on near-infrared spectroscopy and chemometrics. Background Art

[0002] Chamagu, a Xinjiang specialty plant with both medicinal and edible properties, is rich in various active ingredients. Chamagu polysaccharides, among them, exhibit significant potential in functional food development and pharmaceutical applications due to their immunomodulatory, antioxidant, and anti-tumor activities. However, the content and structural differences of polysaccharides directly influence their efficacy, making accurate measurement of chamagu polysaccharide content crucial for product quality control and efficacy evaluation.

[0003] Currently, regulatory authorities are placing increasing emphasis on the detection of plant-derived polysaccharides. While high-performance liquid chromatography and spectrophotometry are widely used in current polysaccharide detection standards, they have significant limitations. HPLC requires complex sample derivatization, resulting in cumbersome procedures, long testing cycles, and high costs. Spectrophotometry relies on numerous chemical reagents, stringent color development conditions, and is susceptible to interference from the sample matrix, resulting in inaccurate test results. These traditional methods struggle to meet the demands of rapid, high-throughput testing, necessitating the development of more efficient detection technologies.

[0004] Near-infrared spectroscopy, with its advantages of rapidity, non-destructiveness, the absence of chemical reagents, and in-situ detection, has achieved remarkable results in agricultural product quality analysis. While this technology has been initially explored in polysaccharide detection, a dedicated detection method for chamagu polysaccharides remains unavailable. The development of a rapid chamagu polysaccharide detection technology based on near-infrared spectroscopy not only overcomes the drawbacks of traditional methods but also provides a fast and accurate detection method for raw material screening, product development, and quality control. This is of great significance for promoting the standardization of the chamagu industry. Summary of the Invention

[0005] This research was funded by the Xinjiang Uygur Autonomous Region Science and Technology Department's key R&D project sub-project: Research and development of key technologies for efficient preparation of Chamagu functional factors (2022B02037-3) and the Young Elite Talents Project - Young Science and Technology Innovation Talents: Quality Evaluation and Processing and Utilization Innovation Research of Medicinal and Edible Resources (2023TSYCCX0074).

[0006] The present invention discloses a method for rapid quantitative detection of polysaccharides in Chamagu based on near-infrared spectroscopy and chemometrics. The method first determines the polysaccharide content in the Chamagu sample by referring to the phenol-sulfuric acid method, then collects the spectral information of the sample to be tested by a near-infrared spectrometer, optimizes the original spectrum using PyCharm (2024.1.3) software, establishes a PLS quantitative prediction model between the polysaccharide content and the spectral information, and evaluates and verifies the performance and predictive ability of the model. The present invention predicts the polysaccharide content in the Chamagu sample through the near-infrared model, and deduces the polysaccharide content in the sample based on the sample quality, thereby achieving accurate and rapid detection of the polysaccharide content in Chamagu by near-infrared spectroscopy.

[0007] The technical solution used in the present invention is: a rapid quantitative detection method for chamagu polysaccharide based on near-infrared spectroscopy and chemometrics, comprising the following steps:

[0008] (1) Cut different batches of chamagu root tubers into pieces, dry them, crush them, and make them into powder; add distilled water for ultrasonic extraction, add ethanol for precipitation and separation, and then dissolve them in distilled water to obtain chamagu polysaccharide test solution;

[0009] (2) Determine the polysaccharide content of the Qiamagu polysaccharide test solution by spectrophotometry;

[0010] (3) Using a near-infrared spectrometer to collect spectral information of the chamagu polysaccharide test solution; removing abnormal samples and preprocessing them; and using an algorithm to perform feature processing to obtain optimized spectral data;

[0011] (4) using the polysaccharide content obtained by spectrophotometry in step (2) as the Y value and the spectral data obtained in step (3) of the corresponding sample as the X value, dividing the data into a calibration set for establishing the model and a prediction set for verifying the predictive ability of the model, and establishing a quantitative regression model using the PLS method;

[0012] (5) Prepare the unknown Chamagu sample into a test sample, perform near-infrared spectral scanning on it, and import the spectral data into the quantitative regression model in step (4) to obtain the polysaccharide content in the unknown Chamagu sample.

[0013] Furthermore, step (1) includes the following sub-steps:

[0014] S1.1 Cut and crush different batches of Chamagu root tubers into 0.1-0.2 cm particles, dry and crush using a freeze dryer, and pass through an 80-mesh sieve to obtain powder;

[0015] S1.2 Accurately weigh different batches of chamagu sample powder in a centrifuge tube, add distilled water, shake well, perform ultrasonic extraction, shake well, filter, and then adjust the volume; add ethanol to the filtrate, shake well, let it stand, centrifuge, discard the supernatant, and obtain a precipitate at the bottom;

[0016] S1.3 Add distilled water to the precipitate and shake well to dissolve it to obtain different batches of Chamagu polysaccharide test solutions.

[0017] Furthermore, in step (3), the wavelength range of the near-infrared spectrometer is 1000~2400 nm, and the collection method is plane diffuse reflection.

[0018] Furthermore, in step (3), the near-infrared spectrometer is equipped with a measuring device; each sample is scanned twice, and the scanning parameters are set as follows: spectrum set range 1000~2400 nm, scan number 32 times, resolution 8 , one data point was collected every 4 cm, a total of 1500 data points were collected, and the scanning temperature was 22 ℃.

[0019] Furthermore, in step (3), abnormal samples are eliminated and preprocessed: the ODXY algorithm is used to eliminate abnormal samples through Pycharm software, the first-order derivative is used for preprocessing, and the 1D-2DCars algorithm is used for feature processing.

[0020] Furthermore, the number of abnormal samples does not exceed 20% of the number of calibration sets.

[0021] Furthermore, the 1D-2D CARS algorithm processing process is as follows:

[0022] The one-dimensional competitive adaptive reweighted sampling method constructs a statistical mathematical model through the Monte Carlo sampling method, and uses PLS to calculate the regression coefficient of each characteristic wavelength variable and calculate the absolute value weight. The formula for calculating the weight value is as follows

[0023] (1)

[0024] in, : The absolute value weight of the regression coefficient of the i-th variable; : The absolute value of the regression coefficient of the i-th variable; : The number of variables remaining in each sampling;

[0025] Sort all wavelength variables by absolute value from large to small, and calculate the retention ratio using the exponential decay function ; At each sampling, select the variables with the highest absolute value of regression coefficient from the n variables left over from the previous sampling. The wavelength variables are then PLS modeled and the RMSECV value is calculated to retain the wavelength ratio. The calculation formula is as follows:

[0026] (2)

[0027] After N samplings are completed, the CARS algorithm obtains N sets of candidate characteristic wavelength subsets and their corresponding RMSECV values. The wavelength variable subset corresponding to the minimum RMSECV value is selected as the characteristic wavelength, thus obtaining the features after 1D-CARS screening.

[0028] The two-dimensional competitive adaptive reweighted sampling method first uses the PCA algorithm to reduce the dimension, and then uses element-by-element difference to perform two-dimensional conversion. The element-by-element difference formula is as follows:

[0029] (3)

[0030] in, and represent the spectral characteristic values ​​of the i-th and j-th components respectively;

[0031] Then, the 1D-CARS core algorithm is used to perform feature screening to obtain the characteristic wavelength, that is, the feature after 2D-CARS screening is obtained. Finally, the features obtained after 1D-CARS and 2D-CARS screening are fused to obtain the feature after 1D-2DCARS screening.

[0032] Furthermore, in step (4), a quantitative regression model was established using PLSR with five main factors.

[0033] Furthermore, in step (4), the quantitative regression model is evaluated and verified, which means that the spectral data of the prediction set is input into the quantitative regression model established in step (4) to obtain the predicted value. When the RPD index is greater than 2.0, the performance of the model meets the requirements.

[0034] Application of the above method in the determination of polysaccharide content in Chamagu samples.

[0035] Compared with the prior art, the present invention has the following improvements:

[0036] (1) The quantitative regression model established by the method of the present invention, and The corrected root mean square error RMSEC and the predicted root mean square error RMSEP were 2.560 and 2.974 respectively, and the predicted relative analysis error RPD was 2.474, indicating that the detection model established by the method of the present invention has high predictive ability and generalization ability. The higher RPD value proves that the model has good accuracy and stability.

[0037] (2) The present invention utilizes near-infrared spectroscopy technology, combined with the results of chemical determination, to construct a mathematical model for the content of Chamagu polysaccharides, and establishes a stable and efficient Chamagu polysaccharide quantification method based on near-infrared spectroscopy technology. The polysaccharide content can be obtained by simply importing the spectral data of the Chamagu sample to be tested into the model. The operation is simple and fast, and is of great significance to Chamagu quality monitoring and food safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a technical flow chart of the present invention.

[0039] Figure 2 This is a diagram of a matching measuring device for the near-infrared spectrometer of the present invention.

[0040] Figure 3 This is the original spectrum diagram in Example 1 of the present invention.

[0041] Figure 4 This is a distribution diagram of the physicochemical values ​​of the polysaccharides in Example 1 of the present invention.

[0042] Figure 5 This is the first-order derivative preprocessing optimization spectrum in Example 1 of the present invention.

[0043] Figure 6 This is the cross-validation diagram of the optimal number of principal components of PLS ​​in Example 1 of the present invention.

[0044] Figure 7 This is a diagram of the quantitative regression (PLS prediction) model of the polysaccharide near-infrared spectroscopy calibration set in Example 1 of the present invention. DETAILED DESCRIPTION

[0045] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process. However, the protection scope of the present invention is not limited to the following embodiment.

[0046] A method for rapid quantitative detection of polysaccharides in Qiamagu based on near-infrared spectroscopy and chemometrics comprises the following steps:

[0047] (1) Different batches of Chamagu root tubers were cut into pieces and crushed into 0.1-0.2 cm particles, dried and crushed using a freeze dryer, and sieved through an 80-mesh sieve to make powder. The powder was divided into a calibration set for establishing the model and a prediction set for verifying the model's predictive ability;

[0048] (2) using spectrophotometry to determine the polysaccharide content of the samples in step (1); using a near-infrared spectrometer to collect spectral information of the samples in step (1);

[0049] The wavelength range of the collection is 1000~2400 nm, and the collection method is plane diffuse reflection;

[0050] (3) preprocessing the spectral information obtained in step (2) to obtain optimized spectral data;

[0051] (4) using the polysaccharide content of the calibration set obtained by spectrophotometry in step (2) as the Y value and the spectral data of the corresponding sample obtained in step (3) as the X value, and establishing a quantitative regression model using the PLS method;

[0052] (5) Prepare the unknown Chamagu sample into a test sample, perform near-infrared spectral scanning on it, and import the spectral data into the quantitative regression model in step (4) to obtain the polysaccharide content in the unknown Chamagu sample.

[0053] Preferably, the sample preparation method in step (1) is:

[0054] S1. Accurately weigh 0.1 g of chamagu sample powder from each batch into a 10 mL centrifuge tube, add 5 mL of distilled water, shake well, and ultrasonically extract for 0.5 h. Shake well, filter, and dilute to 5 mL with distilled water. Add ethanol to the filtrate to a volume fraction of approximately 80%, shake well, let stand, centrifuge for 5 min, and discard the supernatant.

[0055] S2. Add 15 mL of distilled water to the precipitate and shake well to dissolve it to obtain different batches of chamagu polysaccharide test solutions.

[0056] Preferably, the near-infrared spectrometer in step (2) is equipped with a measuring device; each sample is scanned twice, and the scanning parameters are set as follows: spectrum set range 1000~2400 nm, scan number 32 times, resolution 8 , one data point was collected every 4 cm, a total of 1500 data points were collected, and the scanning temperature was 22 ℃.

[0057] Preferably, in step (3), 11 abnormal samples are eliminated using the odxy algorithm using Pycharm software, preprocessed using the first-order derivative, and feature processed using the 1D-2DCars algorithm to obtain the best processing result. The spxy algorithm is used to divide the samples, and the ratio of the number of samples in the divided prediction set and calibration set is 2:8.

[0058] Preferably, the number of abnormal samples does not exceed 20% of the number of calibration samples.

[0059] Preferably, in step (4), a quantitative regression model is established using PLSR with five main factors.

[0060] Preferably, step (4) evaluates and verifies the quantitative regression model, which means inputting the spectral data of the prediction set into the quantitative regression model established in step (4) to obtain the predicted value. When the RPD index is greater than 2.0, the performance of the model meets the requirements.

[0061] The present invention discloses an application of a rapid quantitative detection method for chamagu polysaccharide based on near-infrared spectroscopy and chemometrics.

[0062] Example 1

[0063] The method for quantitatively detecting the polysaccharide content in the Qiamagu sample by near-infrared spectroscopy of this embodiment comprises the following steps:

[0064] (1) Sample preparation

[0065] Different batches of Chamagu root tubers were cut into pieces and crushed into 0.1-0.2 cm particles, dried and crushed using a freeze dryer, and passed through an 80-mesh sieve to form a powder for later use;

[0066] S1. Accurately weigh 0.1 g of chamagu sample powder from each batch into a 10 mL centrifuge tube, add 5 mL of distilled water, shake well, and ultrasonically extract for 0.5 h. Shake well, filter, and dilute to 5 mL with distilled water. Add ethanol to the filtrate to a volume fraction of approximately 80%, shake well, let stand, centrifuge for 5 min, and discard the supernatant.

[0067] Add 15 mL of distilled water to the precipitate and shake well to dissolve it to obtain different batches of chamagu polysaccharide test solutions.

[0068] (2) Sample spectrophotometry

[0069] According to the phenol-sulfuric acid method, the polysaccharide content in the different batches of Chamagu samples was independently determined.

[0070] A. Sample extraction

[0071] Accurately draw 100 To the polysaccharide test solution, add 0.32 mL of 5% phenol solution and 1.00 mL of concentrated sulfuric acid, shake thoroughly, react in a 100°C water bath for 25 min, cool to room temperature with ice water, and detect the absorbance at 480 nm. Each sample should be repeated three times.

[0072] B determination

[0073] Accurately measure 0.0, 0.1, 0.2, 0.4, 0.6, 0.8, and 1.0 mL of the Astragalus polysaccharide standard solution into seven 10 mL stoppered test tubes, and distilled water is added to 1 mL. Add 0.32 mL of 5% phenol reagent and 1 mL of concentrated sulfuric acid, respectively, and mix thoroughly. Incubate the mixture in a 100°C water bath for 25 minutes. After cooling to room temperature with ice water, measure the absorbance of the standard solutions and the sample solution at 480 nm.

[0074] Table 1 Standard series and sample polysaccharide preparation system

[0075]

[0076] Preparation of 5% phenol reagent: Weigh 5.00 g of phenol reagent into a 100 mL beaker, add 95 mL of secondary water to completely dissolve it, shake well, protect from light, and place at room temperature. Prepare and use immediately.

[0077] C result calculation

[0078] Calculation formula:

[0079]

[0080] in:

[0081] X——polysaccharide content in the sample, g / 100g;

[0082] c——polysaccharide concentration in the test solution, mg / mL;

[0083] f——dilution factor of test solution;

[0084] V——extraction volume, mL;

[0085] M——sample mass, g;

[0086] 1000, 100——Conversion factor.

[0087] Note: In this example, the polysaccharide content (g / 100g) in the sample reaction system test solution is used for modeling. The polysaccharide content X (g / 100g) in the sample can also be converted according to the above formula for modeling.

[0088] D Result Statistics

[0089] The samples were tested by the aforementioned spectrophotometric method to obtain the statistical results of the polysaccharide content in the test solution shown in Table 2.

[0090] Table 2 Statistics of polysaccharide content in test solution

[0091]

[0092] *Only the 100 samples ultimately used for modeling are counted.

[0093] (3) Sample near-infrared spectroscopy

[0094] The remaining samples with corresponding numbers were measured independently using a near-infrared spectrometer simultaneously to collect the near-infrared absorption spectrum data of the polysaccharide of the 100 samples.

[0095] Example 2

[0096] The method for building a near-infrared spectroscopy quantitative analysis model in this embodiment includes the following steps:

[0097] A. Spectral data acquisition

[0098] Sample near-infrared spectrum scanning and sample set establishment: A near-infrared spectrometer (SUPNIR 2700, a near-infrared spectrometer produced by China Juguan Technology) was used to perform near-infrared spectrum scanning on the calibration samples of Chamagu. Each time, the Chamagu sample powder was taken and placed in the rotating diffuse reflectance sampling disk of the near-infrared spectrum. The built-in background of the instrument was used as a reference. The near-infrared spectrum of the Chamagu sample was collected in uniform rotation mode. Each sample was scanned 3 times. The scanning parameter settings were: spectral set range 1000~2400nm, number of scans 32 times, resolution 8 , one data point was collected every 4 cm, a total of 1500 data points were collected, the scanning temperature was 22℃, and the near-infrared spectrum of the Qiama ancient sample is as follows Figure 3 As shown, the horizontal axis is wavelength (nm) and the vertical axis is absorbance (A).

[0099] B. Spectral data processing

[0100] The 100 spectral data and physicochemical value information were imported into PyCharm (2024.1.3). Eleven outliers were removed using the ODXY algorithm, and the samples were partitioned into a 2:8 ratio using the SPXY algorithm. The 89 spectra were then smoothed and denoised using seven preprocessing methods: SG0 (sg_derivative), SG1 (sg_derivative_1st, first-order derivative), SG2 (sg_derivative_2nd, second-order derivative), MSC (multiplicative_scatter_correction), SG0+MSC, SG1+MSC, and SG2+MSC. The preprocessing parameter settings are shown in Table 3.

[0101] Table 3 Spectral preprocessing method parameters

[0102]

[0103] After preprocessing, 1D-2DCARS is used for feature extraction to achieve feature screening and dimensionality reduction to retain the effective information of the features and improve the generalization ability of the model. The 1D-2DCARS algorithm is implemented as follows:

[0104] The one-dimensional competitive adaptive reweighted sampling method (1D-CARS) constructs a statistical mathematical model through MC (Monte Carlo sampling method), and uses PLS to calculate the regression coefficient of each characteristic wavelength variable and calculate the absolute value weight. The weight value calculation formula is as follows

[0105] (1)

[0106] in, : The absolute value weight of the regression coefficient of the i-th variable; : The absolute value of the regression coefficient of the i-th variable; : The number of variables remaining in each sampling.

[0107] Sort all wavelength variables by absolute value from large to small, and calculate the retention ratio using the exponential decay function ; At each sampling, select the variables with the highest absolute value of regression coefficient from the n variables left over from the previous sampling. The wavelength variables are then PLS modeled and the RMSECV value is calculated to retain the wavelength ratio. The calculation formula is as follows:

[0108] (2)

[0109] After N samplings are completed, the CARS algorithm obtains N groups of candidate characteristic wavelength subsets and the corresponding RMSECV values. The wavelength variable subset corresponding to the minimum RMSECV value is selected as the characteristic wavelength, that is, the characteristics after 1D-CARS screening are obtained.

[0110] The two-dimensional competitive adaptive reweighted sampling method (2D-CARS) first uses the PCA algorithm to reduce the dimension, and then uses element-by-element difference to perform two-dimensional conversion. The element-by-element difference formula is as follows:

[0111] (3)

[0112] in, and represent the spectral characteristic values ​​of the i-th and j-th components respectively.

[0113] The 1D-CARS core algorithm is then used to filter the features to obtain the characteristic wavelengths, which are the features filtered by 2D-CARS. Finally, the features obtained by 1D-CARS and 2D-CARS filtering are fused to obtain the features filtered by 1D-2DCARS.

[0114] C quantitative regression model establishment

[0115] The polysaccharide content determined by spectrophotometry was taken as the Y value (reference value), and the spectral data after processing of the corresponding sample number was taken as the X value (predicted value). The model was then built after the optimal number of principal components was determined through 10-fold cross validation (see Figure 6 ), compare the model performance under different processing methods to determine the best modeling method.

[0116] Table 4 Performance of the model under different processing methods

[0117]

[0118] D model evaluation

[0119] Analyze the calibration set quantitative regression (PLS prediction) model diagram (see Figure 7 ), and evaluate the performance and stability of the model. Examine the change of the determination coefficient R 2 . The closer R 2 is to 1, the better the model effect. Generally, 0.66 ≤ R 2 < 0.80 indicates that the prediction effect is achieved, 0.81 ≤ R 2 < 0.90 indicates that the prediction effect is better, and R 2 ≥ 0.90 indicates that the prediction effect is the best. At the same time, use the ratio of the sample standard deviation of the prediction set to the root mean square error of prediction, that is, the relative analysis error of prediction RPD to evaluate the actual prediction effect of the model. Usually, RPD < 1.4 indicates that the established model is unreliable, 1.4 < RPD < 2.0 indicates that the established model is relatively reliable, and RPD > 2.0 indicates that the established model has high reliability and can be used for model analysis. The established model R 2 c and R 2 p are 0.879 and 0.836 respectively (the results are shown in Table 4), indicating a high degree of correlation between the predicted value and the measured value; the root mean square error of calibration RMSEC and the root mean square error of prediction RMSEP are 2.560 and 2.974 respectively (the results are shown in Table 4), and the difference between the two is small, indicating that both the calibration samples and the validation samples have good representativeness; at the same time, both values are small, indicating that the model has high prediction ability and generalization ability; the relative analysis error of prediction RPD is 2.474 (the results are shown in Table 4), indicating that the model has good accuracy and stability.

[0120] The above is only the preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A rapid quantitative detection method for chamagu polysaccharide based on near-infrared spectroscopy and chemometrics, characterized in that: The following steps are involved: (1) Cut different batches of chamagu root tubers into pieces, dry them, crush them, and make them into powder; add distilled water for ultrasonic extraction, add ethanol for precipitation and separation, and then dissolve them in distilled water to obtain chamagu polysaccharide test solution; (2) Determine the polysaccharide content of the Qiamagu polysaccharide test solution by spectrophotometry; (3) Using a near-infrared spectrometer to collect spectral information of the chamagu polysaccharide test solution; removing abnormal samples and preprocessing them; and using an algorithm to perform feature processing to obtain optimized spectral data; (4) using the polysaccharide content obtained by spectrophotometry in step (2) as the Y value and the spectral data obtained in step (3) of the corresponding sample as the X value, dividing the data into a calibration set for establishing the model and a prediction set for verifying the predictive ability of the model, and establishing a quantitative regression model using the PLS method; (5) Prepare the unknown Chamagu sample into a test sample, perform near-infrared spectral scanning on it, and import the spectral data into the quantitative regression model in step (4) to obtain the polysaccharide content in the unknown Chamagu sample.

2. The method according to claim 1, wherein: Step (1) includes the following sub-steps: S1.1 Cut and crush different batches of Chamagu root tubers into 0.1-0.2 cm particles, dry and crush using a freeze dryer, and pass through an 80-mesh sieve to obtain powder; S1.2 Accurately weigh different batches of chamagu sample powder in a centrifuge tube, add distilled water, shake well, perform ultrasonic extraction, shake well, filter, and then adjust the volume; add ethanol to the filtrate, shake well, let it stand, centrifuge, discard the supernatant, and obtain a precipitate at the bottom; S1.3 Add distilled water to the precipitate and shake well to dissolve it to obtain different batches of Chamagu polysaccharide test solutions.

3. The method according to claim 1, wherein: In step (3), the wavelength range of the near-infrared spectrometer is 1000~2400 nm, and the acquisition method is plane diffuse reflection.

4. The method according to claim 3, characterized in that In step (3), the near-infrared spectrometer is equipped with a measuring device; each sample is scanned twice, and the scanning parameters are set as follows: spectral range 1000~2400 nm, number of scans 32 times, resolution 8cm -1 , one data point was collected every 4 cm, a total of 1500 data points were collected, and the scanning temperature was 22 ℃.

5. The method according to claim 1, wherein In step (3), abnormal sample removal and preprocessing: using Pycharm software to use the ODXY algorithm to remove abnormal samples, using the first-order derivative for preprocessing and using the 1D-2DCars algorithm for feature processing.

6. The method according to claim 5, characterized in that The number of abnormal samples does not exceed 20% of the number of calibration samples.

7. The method according to claim 6, characterized in that: The 1D-2DCARS algorithm processing process is as follows: The one-dimensional competitive adaptive reweighted sampling method constructs a statistical mathematical model through the Monte Carlo sampling method, and uses PLS to calculate the regression coefficient of each feature and calculate the absolute value weight. The formula for calculating the weight value is as follows （1） in, : The absolute value weight of the regression coefficient of the i-th variable; : The absolute value of the regression coefficient of the i-th variable; : The number of variables remaining in each sampling; Sort all wavelength variables by absolute value from large to small, and calculate the retention ratio using the exponential decay function ; At each sampling, select the variables with the highest absolute value of regression coefficient from the n variables left over from the previous sampling. *n wavelength variables, then PLS modeling is performed to calculate the RMSECV value, retaining the wavelength ratio The calculation formula is as follows: （2） After N samplings are completed, the CARS algorithm obtains N sets of candidate characteristic wavelength subsets and their corresponding RMSECV values. The wavelength variable subset corresponding to the minimum RMSECV value is selected as the characteristic wavelength, thus obtaining the features after 1D-CARS screening. The two-dimensional competitive adaptive reweighted sampling method first uses the PCA algorithm to reduce the dimension, and then uses element-by-element difference to perform two-dimensional conversion. The element-by-element difference formula is as follows: （3） in, and represent the spectral characteristic values ​​of the i-th and j-th components respectively; Then, the 1D-CARS core algorithm is used to perform feature screening to obtain the characteristic wavelength, that is, the feature after 2D-CARS screening is obtained. Finally, the features obtained after 1D-CARS and 2D-CARS screening are fused to obtain the feature after 1D-2DCARS screening.

8. The method according to claim 1, characterized in that In step (4), a quantitative regression model was established using PLSR with five main factors.

9. The method according to claim 8, characterized in that In step (4), the quantitative regression model is evaluated and verified, which means that the spectral data of the prediction set is input into the quantitative regression model established in step (4) to obtain the predicted value. When the RPD index is greater than 2.0, the performance of the model meets the requirements.

10. Application of the method according to any one of claims 1 to 9 in determining the polysaccharide content of Qiamagu samples.

Images