Angelica dahurica medicinal material quality evaluation method based on near infrared spectrum technology

Abstract

The invention discloses a method for evaluating the quality of an angelica dahurica medicinal material based on a near infrared spectrum technology, which realizes rapid determination of seven active components of the angelica dahurica medicinal material and identification of stoichiometric angelica dahurica by utilizing the near infrared spectrum technology and combining Python with chemometrics. According to the method, the quality of the angelica dahurica medicinal material is evaluated by utilizing the near infrared spectrum technology, and the method has the characteristics of rapidness, accuracy and low cost, and is simple, convenient, easy to implement and good in stability. The method can provide reference for the quality evaluation of the radix angelicae medicinal material, so that the stability and controllability of the quality of the radix angelicae medicinal material and the clinical medication safety are better ensured.

Description

Technical Field

[0001] The invention relates to a method for evaluating the quality of Chinese medicinal materials, and in particular to a method for evaluating the quality of angelica dahurica medicinal materials based on near infrared spectroscopy technology. Background Art

[0002] Angelica dahurica is the dried root of Angelica dahurica (Fisch.ex Hoffm.) Benth.et Hook.f. or Angelica daurica (Fisch.ex Hoffm.) Benth.et Hook.f.var.formosana (Boiss.) Shan et Yuan, a plant of the Umbelliferae family. It is a commonly used Chinese herbal medicine and can also be used as food. It has the effects of dispelling cold, dispelling wind and relieving pain, clearing the nasal passages, and drying dampness and stopping leukorrhea. The main chemical components are coumarin compounds, which have multiple pharmacological activities such as analgesia, anti-inflammatory, and anti-tumor.

[0003] Near infrared spectroscopy is a rapid analytical technology that has developed rapidly in recent years. It has the advantages of fast analysis speed, no damage to samples, and no chemical pollution. It has been widely used in the qualitative and quantitative analysis of traditional Chinese medicines, providing a basis for the quality evaluation of medicinal materials. A large number of studies have found that sulfur fumigation can cause changes in the chemical composition and pharmacological activity of medicinal materials, posing a safety hazard. Rapid and easy identification of sulfur-fumigated Angelica dahurica is crucial to ensuring the quality of Angelica dahurica medicinal materials. Currently, the content determination of Angelica dahurica active ingredients reported in the literature mostly uses liquid phase or liquid chromatography-mass spectrometry technology, which is expensive and relatively slow. The use of near infrared spectroscopy technology to determine the content of Angelica dahurica, on the basis of completing the modeling, only a small amount of medicinal powder is needed for analysis, without the need for expensive chemical reference substances and cumbersome medicinal material extraction and testing processes, can achieve low-cost, rapid and accurate quality evaluation of Angelica dahurica medicinal materials, especially suitable for multi-batch and large sample size detection and analysis in the production and research process, and has good application prospects in any link of the Angelica dahurica market circulation. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a method for evaluating the quality of angelica dahurica based on near infrared spectroscopy technology, which has low cost, high analysis efficiency and high analysis speed and can be used to achieve qualitative identification and quantitative determination of sulfur-fumigated angelica dahurica and non-sulfur-fumigated angelica dahurica. The present invention can provide a reference for the quality evaluation of angelica dahurica, thereby better ensuring the stability and safety of the quality of angelica dahurica.

[0005] Technical solution: In order to achieve the above objectives, the technical solution adopted by the present invention is:

[0006] A method for evaluating the quality of Angelica dahurica medicinal material, characterized by comprising the following steps:

[0007] S1, near infrared spectrum collection: collecting near infrared spectrum of Angelica dahurica samples;

[0008] S2, qualitative identification of sulfur-fumigated Angelica dahurica: standardization (SS) + Norris derivative filtering (ND) were selected as the spectral preprocessing method; support vector machine (SVM) was used as the modeling method to achieve qualitative identification of sulfur-fumigated Angelica dahurica.

[0009] Preferably, the quality evaluation method further comprises S3: quantitative analysis of active ingredients in the Angelica dahurica medicinal material, wherein S3 is analyzed by the following method:

[0010] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + Norris derivative filter (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of hydrated oxidized praeruptosine; and / or

[0011] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + SG convolution smoothing (SG) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of white angelica; and / or

[0012] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of bergamot lactone; and / or

[0013] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of isoanisole; and / or

[0014] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of oxidized praeruptosine; and / or

[0015] The spectrum obtained by S1 is preprocessed by the Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of imperatorin; and / or

[0016] The SG convolution smoothing (SG) method was used to preprocess the spectrum obtained by S1, and the classification boosting tree (CatBoost) was used for modeling to analyze the content of isoimperatorin.

[0017] Preferably, the scanning band range of step S1 is 12500-4000 cm -1 .

[0018] Preferably, the accuracy, precision, recall and F1 score of the qualitative identification of Angelica dahurica sulfur fumigation are all greater than 0.99.

[0019] The present invention also provides a method for establishing a quality evaluation model for Radix Angelicae Dahuricae, which is characterized by comprising the following steps:

[0020] S1, near infrared spectrum collection: collecting near infrared spectrum of Angelica dahurica samples;

[0021] S2, spectral preprocessing: The near-infrared spectrum obtained in S1 was preprocessed using various preprocessing methods and their combinations, including original spectrum (Constant), multivariate scatter correction (MSC), standard normal transformation (SNV), standardization (SS), first-order derivative (1st Der), second-order derivative (2nd Der), SG convolution smoothing (SG), and Norris derivative filtering (ND) smoothing;

[0022] S3, analysis model screening: the spectral data processed by S2 were modeled using algorithms including support vector machine (SVM), random forest (RF), naive Bayes (NB), logistic regression (LR), K nearest neighbor (KNN) or partial least squares discriminant analysis (PLS-DA);

[0023] S4, Model determination:

[0024] Identification of sulfur-fumigated Angelica dahurica: Standardization (SS) + Norris derivative filtering (ND) were selected as the spectral preprocessing method; support vector machine (SVM) was used as the modeling method to achieve qualitative identification of sulfur-fumigated Angelica dahurica.

[0025] Preferably, the method further comprises: quantitatively analyzing the active ingredients in the Angelica dahurica medicinal material by the following method:

[0026] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + Norris derivative filter (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of hydrated oxidized praeruptosine; and / or

[0027] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + SG convolution smoothing (SG) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of white angelica; and / or

[0028] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of bergamot lactone; and / or

[0029] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of isoanisole; and / or

[0030] The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of oxidized praeruptosine; and / or

[0031] The spectrum obtained by S1 is preprocessed by the Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of imperatorin; and / or

[0032] The SG convolution smoothing (SG) method was used to preprocess the spectrum obtained from S1, and the classification boosting tree (CatBoost) was used for modeling to analyze the content of isoimperatorin.

[0033] Preferably, S3 also includes an algorithm optimization step: the model obtained by S3 is subjected to a comprehensive search and tuning of the model parameters and the number of principal components of principal component analysis (PCA) through a grid search method, the candidate value range of the parameters is set, a 5-fold cross validation is adopted, and the root mean square error of the internal cross validation (RMSECV) is used as the performance evaluation standard to determine the optimal parameter combination; the model parameters are adaptively adjusted according to the characteristics of different components.

[0034] Beneficial effects:

[0035] The near infrared spectroscopy technology obtained by the present invention through a large number of experimental screening and optimization can be efficiently and accurately used for the qualitative identification of the Angelica dahurica medicinal material (sulfur-fumigated Angelica dahurica and non-sulfur-fumigated Angelica dahurica), and the quantitative analysis model established can be efficiently and accurately used to detect the content of the effective ingredients of hydrated oxidized peucedanum, white angelica, bergamot lactone, isoanisole lactone, oxidized peucedanum, imperatorin and isoimperatorin in the Angelica dahurica, and has the advantages of fast detection speed and low detection cost compared with high-performance liquid phase. The present invention can provide a reference for the quality evaluation of the Angelica dahurica medicinal material, so as to better ensure the stability and safety of the medicinal material quality, and has important application value. Using algorithms such as support vector machine and category boosting tree, compared with the traditional partial least squares algorithm, the effect of the model is further improved in both the qualitative discrimination of the Angelica dahurica medicinal material and the quantitative analysis of 7 active ingredients in the Angelica dahurica medicinal material. In addition to selecting a suitable algorithm, the present invention also performs parameter adjustment to further optimize the model performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is the original near-infrared overlay image of the Angelica dahurica sample of the present invention.

[0037] Figure 2 This is a qualitative model diagram of sulfur fumigation of dried ginger in the present invention.

[0038] Figure 3 This is a near-infrared content determination diagram of hydrated oxidized praeruptoside, a component of Angelica dahurica of the present invention.

[0039] Figure 4 This is a near-infrared content determination diagram of the white angelica component of the present invention.

[0040] Figure 5 This is a near-infrared content determination diagram of bergamot lactone, a component of Angelica dahurica of the present invention.

[0041] Figure 6 This is a near-infrared content determination diagram of isoanisole lactone, a component of Angelica dahurica of the present invention.

[0042] Figure 7 This is a near-infrared content determination diagram of oxidized praeruptosin, a component of Angelica dahurica of the present invention.

[0043] Figure 8 This is a near-infrared content determination diagram of imperatorin, a component of Angelica dahurica of the present invention.

[0044] Fig. 9 This is a near-infrared content determination diagram of isoimperatorin, a component of Angelica dahurica of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further explained below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0046] Example 1

[0047] 1. Instruments and Materials

[0048] 1.1 Experimental materials

[0049] Three batches of Angelica dahurica medicinal materials were collected in Bozhou, Anhui Province. All samples were identified by Associate Professor Hou Fangjie of Hebei University of Traditional Chinese Medicine as dried roots of Angelica dahurica (Fisch.ex Hoffm.) Benth.et Hook.f. or Angelica dahurica (Fisch.ex Hoffm.) Benth.et Hook.f.var.formosana (Boiss.) Shanet Yuan of the Umbelliferae family. The specific sample information is shown in Table 5.

[0050] Table 5: Information of Radix Angelicae Dahuricae

[0051] serial number Origin Harvest period Whether sulfur fumigation S1 Anhui Bozhou 2021.06 Sulfur fumigation S2 Anhui Bozhou 2021.06 Sulfur fumigation S4 Anhui Bozhou 2021.06 Sulfur-free

[0052] 2 Test methods

[0053] 2.1 Collection of near-infrared spectra

[0054] Take an appropriate amount of Angelica dahurica powder, put it into the quartz sample cup to two-thirds, spread it evenly, and dry it at 45℃ to constant weight. Use near-infrared spectrometer for data acquisition. Preheat the instrument for 30 minutes at room temperature. Collect the spectrum with air as reference to subtract the background. Use integrating sphere diffuse reflectance to collect the spectrum. The scanning conditions are: resolution 8cm-1, scanning band range: 12500~4000cm-1, sample background and sample scanning time: 32s. Keep the sample in a compacted, uniform and flat state before each scan. Repeat the scan 3 times for each batch of samples, and take the average value as the sample near-infrared spectrum to establish the model. The original near-infrared superposition of the Angelica dahurica sample is shown in Figure 1 .

[0055] 2.2 Identification of sulfur-fumigated Angelica dahurica

[0056] 2.2.1 Spectral preprocessing

[0057] Depend on Figure 1It can be seen that the original near-infrared spectra of each sample are basically the same, and the spectral bands are complex and overlapped. It is difficult to see the difference in spectral information of the medicinal materials based on this figure. Therefore, it is necessary to perform corresponding pre-processing on the original spectrum data to make the spectrum information more intuitive and clear.

[0058] Selection of spectral preprocessing method During the near-infrared spectrum acquisition process, the near-infrared spectrum may undergo subtle changes due to the difference in the state of the instrument and the measurement conditions, such as light scattering, stray light and instrument response, which may cause baseline drift and non-repetition of the near-infrared spectrum. The spectrum can be corrected by preprocessing it. Therefore, in order to establish a stable and reliable correction model, the spectrum should be reasonably processed before data analysis to reduce or even eliminate the impact of various non-target factors on the spectral information. The present invention uses a combination of original spectrum (Constant), multivariate scatter correction (MSC), standard normal transformation (SNV), standardization (SS), first-order derivative (1st Der), second-order derivative (2nd Der), SG convolution smoothing (SG) and Norris derivative filtering (ND) smoothing methods and various preprocessing methods for processing.

[0059] 2.2.2 Sulfur fumigation model establishment and screening:

[0060] Using the discriminant analysis model established in Python software, the optimization content includes the selection of spectral preprocessing methods and modeling methods. The algorithms involved include support vector machine (SVM), random forest (RF), naive Bayes (NB), logistic regression (LR), K nearest neighbor (KNN), partial least squares discriminant analysis (PLS-DA), and the accuracy, precision, recall and F1 score are used as comprehensive evaluation indicators to select the best qualitative analysis model. The higher the accuracy, precision, recall and F1 score, the better the model is. The modeling algorithm parameters are shown in Table 1, and the modeling results are shown in Table 2.

[0061] Table 1 Classification algorithm parameter settings

[0062]

[0063]

[0064] Table 2 Near infrared discriminant analysis model and its performance

[0065]

[0066]

[0067]

[0068]

[0069] From the data in Table 2, we can see that when the spectral preprocessing step adopts standardization combined with Norris derivative filtering smoothing method (SS+ND) and matches the support vector machine (SVM) when the sulfur fumigation model is established, the classification accuracy reaches 100%, which is significantly better than other preprocessing methods. This method can effectively improve the model performance, has high accuracy and stability, and is suitable for the classification task of sulfur-fumigated Angelica dahurica and non-sulfur Angelica dahurica. The qualitative model diagram is shown in the figure. Figure 2 shown.

[0070] 2.3 Determination of seven active components in Angelica dahurica

[0071] 2.3.1 Analysis model screening

[0072] Preprocessing methods: The original spectrum (Constant), multivariate scatter correction (MSC), standard normal transformation (SNV), standardization (SS), first-order derivative (1st Der), second-order derivative (2nd Der), SG convolution smoothing (SG) and Norris derivative filtering (ND) smoothing method as well as a combination of various preprocessing methods were used for processing.

[0073] Modeling methods and model parameter selection: The algorithms involved mainly include partial least squares regression (PLSR), random forest (RF), gradient boosted decision tree (GBDT), classification boosted tree (CatBoost), and the optimization content is the selection of modeling algorithms and parameter optimization. Prediction set determination coefficient Mean square error (MSE) and mean absolute error (MAE) and relative percentage deviation (RPD) of the prediction set p ) is a comprehensive evaluation index, and the best quantitative analysis model is selected. and RPD p The higher the MSE and MAE, the better the model's fitting effect and prediction ability. The stronger the applicability of the constructed model, the better the prediction effect.

[0074] The results of different algorithms and preprocessing combinations are shown in Table 3.

[0075] Table 3 Modeling effects of different algorithm combinations

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] Quantitative analysis model conditions of hydrated oxidized praeruptoside: modeling band 12500~4000cm -1 , the spectral preprocessing method is Constant+1st Der+ND;

[0082] Quantitative analysis model conditions of white angelica: modeling band 12500~4000cm -1 ,The spectral preprocessing method is Constant+1st Der+SG;

[0083] Quantitative analysis model conditions of bergamot lactone: modeling band 12500~4000cm -1 , the spectral preprocessing method is Constant+ND;

[0084] Quantitative analysis model conditions of isoanisole: Modeling band 12500~4000cm -1 , the spectrum preprocessing method is Constant;

[0085] Quantitative analysis model conditions of oxidized praeruptoside: modeling band 12500~4000cm -1 , the spectrum preprocessing method is Constant;

[0086] Quantitative analysis model conditions of imperatorin: Modeling band 12500~4000cm -1 , the spectral preprocessing method is ND;

[0087] Quantitative analysis model conditions of isoimperatorin: Modeling band 12500~4000cm -1 , the spectral preprocessing method is SG.

[0088] The quantitative model was established using the CatBoost method in Python software. CatBoost is a decision tree algorithm based on gradient boosting, which is optimized for the processing of categorical features. It processes categorical features through efficient algorithms without the need for cumbersome feature encoding (such as one-hot encoding), thereby reducing the complexity of preprocessing. The model map is shown in Figures 3 to 9 , it can be seen from the figure that the predicted value of the near infrared detection provided by the present invention is close to the reference value, indicating that the established quantitative model can be used for the quantitative analysis of the effective components of Angelica dahurica.

[0089] Table 4 CatBoost parameter settings

[0090]

[0091] 2.4 Determination of the contents of seven active components in Angelica dahurica by HPLC

[0092] 2.4.1 HPLC chromatographic conditions

[0093] An H-Class ultra-high performance liquid chromatograph (Waters, USA) was used; a Phenomenon Titank C18 column (2.1 mm) was used for chromatographic separation. The mobile phase was acetonitrile (A)-0.2% formic acid solution (B) with gradient elution (0-6 min, 13%-14% A; 6-13.5 min, 14%-15.5% A; 13.5-18 min, 15.5%-28% A; 18-24 min, 28%-28.4% A; 24-31 min, 28.4%-37% A; 31-35.5 min, 37%-45% A; 35.5-37.5 min, 45%-47.3% A; 37.5-40 min, 47.3%-53% A; 40-50 min, 53%-60% A; 50-52 min, 60%-13% A), with a flow rate of 0.3 mL min -1 , column temperature 35°C, injection volume 1.2μL.

[0094] 2.4.2 Preparation of mixed reference solution

[0095] Take appropriate amounts of hydrated oxidized peucedanum, white angelica, bergamot lactone, isoanisole, oxidized peucedanum, imperatorin, and isoimperatorin, accurately weigh them, and dilute them with methanol to obtain a mixed reference solution (108.29 μg mL -1 , 44.15 μg·mL -1 , 17.56 μg·mL -1 , 5.75 μg·mL -1 , 72.68 μg·mL -1 , 1155.12 μg·mL -1 , 76.0 μg·mL -1 ) and you’ll get it.

[0096] 2.4.3 Preparation of test solution

[0097] 2.0 g of Angelica dahurica powder was accurately weighed (passed through a 60-mesh sieve) and placed in a 50-mL conical flask. 20 mL of 70% methanol was accurately added and weighed. The mixture was ultrasonically treated (power 250 W, frequency 40 kHz) for 45 min. The mixture was cooled to room temperature and weighed again. The lost mass was supplemented with 70% methanol. The mixture was shaken and the filtrate was filtered through a microporous filter membrane (0.22 μm).

[0098] 2.4.4 Content determination

[0099] The standard curve method was used to determine the contents of hydrated oxypeucedanin, white angelicae, bergamotolide, isoanisoleide, oxypeucedanin, imperatorin and isoimperatorin in Angelica dahurica.

[0100] 3 Results

[0101] 3.1 The results of qualitative analysis are shown in Table 6:

[0102] Table 6 Near infrared qualitative analysis results

[0103] serial number actual Discrimination S1 Sulfur fumigation Sulfur fumigation S2 Sulfur fumigation Sulfur fumigation S4 Sulfur-free Sulfur-free

[0104] It can be seen from the results in Table 6 that the sulfur-fumigation of Angelica dahurica identified by near infrared is consistent with the actual result, so the qualitative method provided by the present invention can realize the rapid identification of sulfur-fumigated Angelica dahurica by near infrared modeling.

[0105] 3.2 Quantitative analysis results, the results of the near infrared spectroscopy (NIR) and HPLC established by the present invention are shown in Table 7 below:

[0106] Table 7 Near infrared quantitative analysis results

[0107]

[0108]

[0109] It can be seen from the data in Table 7 that the hydrated oxidized peucedanum, white angelicae, bergamotolide, isoanisole, oxidized peucedanum, imperatorin and isoimperatorin in Angelica dahurica determined by near infrared in the present invention are basically consistent with the results determined by HPLC, and their p values ​​are all greater than 0.05, indicating that there is no significant difference, which proves that the near infrared technology established by the present invention can be used for the content determination of Angelica dahurica medicinal material.

[0110] The present invention utilizes near-infrared spectroscopy results chemometrics to rapidly identify whether the Angelica dahurica medicinal material has been sulfur-fumigated and to determine its effective ingredients of hydrated oxidized peucedanum, white angelicae, bergamotolide, isoanisole, oxidized peucedanum, imperatorin and isoimperatorin. The method has stable and reliable results, is simple and easy to operate, has low detection cost, and is conducive to promotion.

[0111] Although specific embodiments have been described above with reference to disclosed embodiments and examples, such embodiments are merely illustrative and do not limit the scope of the invention. Changes and modifications may be made according to those of ordinary skill in the art without departing from the broader aspects of the invention as defined in the appended claims. All publications, patents, and patent documents are incorporated herein by reference as if incorporated herein by reference alone. Limitations inconsistent with the present disclosure should not be understood thereby. The present invention has been described with reference to various specific and preferred embodiments and techniques. However, it should be understood that many changes and modifications may be made while remaining within the spirit and scope of the invention.

Claims

1. A method for evaluating the quality of Angelica dahurica medicinal material, characterized in that: The steps include: S1, near infrared spectrum collection: collecting near infrared spectrum of Angelica dahurica samples; S2, qualitative identification of Baizhi sulfur fumigation: standardization (SS) + Norris derivative filtering (ND) was selected as the spectral preprocessing method; Support vector machine (SVM) was used as the modeling method to achieve qualitative identification of sulfur-fumigated Angelica dahurica.

2. The quality evaluation method according to claim 1, characterized in that: The invention also includes S3: quantitative analysis of active ingredients in the medicinal material of Angelica dahurica, wherein S3 is analyzed by the following method: The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + Norris derivative filter (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of hydrated oxidized praeruptosine; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + SG convolution smoothing (SG) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of white angelica; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of bergamot lactone; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of isoanisole; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of oxidized praeruptosine; and / or The spectrum obtained by S1 is preprocessed by the Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of imperatorin; and / or The SG convolution smoothing (SG) method was used to preprocess the spectrum obtained by S1, and the classification boosting tree (CatBoost) was used for modeling to analyze the content of isoimperatorin.

3. The quality evaluation method according to claim 1 or 2, characterized in that: The scanning band range of step S1 is 12500~4000cm -1 .

4. The quality evaluation method according to claim 1, characterized in that: The accuracy, precision, recall and F1 score of the qualitative identification of Angelica dahurica by sulfur fumigation are all greater than 0.

99.

5. A method for establishing a quality evaluation model for Angelica dahurica, characterized in that: The steps include: S1, near infrared spectrum collection: collecting near infrared spectrum of Angelica dahurica samples; S2, spectral preprocessing: The near-infrared spectrum obtained in S1 was preprocessed using various preprocessing methods and their combinations, including original spectrum (Constant), multivariate scatter correction (MSC), standard normal transformation (SNV), standardization (SS), first-order derivative (1st Der), second-order derivative (2nd Der), SG convolution smoothing (SG), and Norris derivative filtering (ND) smoothing; S3, analysis model screening: the spectral data processed by S2 were modeled using algorithms including support vector machine (SVM), random forest (RF), naive Bayes (NB), logistic regression (LR), K nearest neighbor (KNN) or partial least squares discriminant analysis (PLS-DA); S4, Model determination: Identification of sulfur-fumigated Angelica dahurica: Standardization (SS) + Norris derivative filtering (ND) was selected as the spectral preprocessing method; Support vector machine (SVM) was used as the modeling method to achieve qualitative identification of sulfur-fumigated Angelica dahurica.

6. The model building method according to claim 5, characterized in that: It also includes: quantitative analysis of active ingredients in Angelica dahurica by the following method: The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + Norris derivative filter (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of hydrated oxidized praeruptosine; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + first-order derivative (1st Der) + SG convolution smoothing (SG) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of white angelica; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) + Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of bergamot lactone; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of isoanisole; and / or The spectrum obtained by S1 is preprocessed by the original spectrum (Constant) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of oxidized praeruptosine; and / or The spectrum obtained by S1 is preprocessed by the Norris derivative filtering (ND) method, and the classification boosting tree (CatBoost) is used for modeling to analyze and obtain the content of imperatorin; and / or The SG convolution smoothing (SG) method was used to preprocess the spectrum obtained by S1, and the classification boosting tree (CatBoost) was used for modeling to analyze the content of isoimperatorin.

7. The model building method according to claim 5 or 6, characterized in that: The S3 also includes an algorithm optimization step: the model obtained by S3 is comprehensively searched and tuned for model parameters and the number of principal components of principal component analysis (PCA) through a grid search method, the candidate value range of the parameters is set, 5-fold cross validation is adopted, and the root mean square error of internal cross validation (RMSECV) is used as a performance evaluation standard to determine the optimal parameter combination; the model parameters are adaptively adjusted according to the characteristics of different components.

8. The model building method according to claim 5 or 6, characterized in that: It also includes validating the accuracy of the model using a reference method.

9. The model building method according to claim 8, characterized in that: The reference method is the HPLC method.

Images