A method for identifying safflower quality based on near-infrared spectroscopy

Abstract

The present application belongs to the technical field of safflower identification, and particularly relates to a safflower quality identification method based on near-infrared spectroscopy. The safflower quality identification method based on near-infrared spectroscopy collects near-infrared spectroscopy data of genuine safflower samples of different growth periods and different origins and various mixed or dyed counterfeit samples at 12500 cm ‑1 4000 cm ‑1 The present application collects near-infrared spectroscopy data of genuine safflower samples of different growth periods and different origins and various mixed or dyed counterfeit samples at 12500 cm ‑1 4000 cm ‑1 , performs multivariate scatter correction, and establishes a safflower period and origin identification model through partial least squares-discriminant analysis. The present application performs multivariate scatter correction, characteristic variable screening through combined interval partial least squares, and establishes a safflower mixed or dyed type identification model through partial least squares-discriminant analysis. The distribution of the near-infrared spectroscopy data of the safflower sample to be tested is used to determine the growth period, origin, authenticity, mixed or dyed type of the safflower sample to be tested. The safflower quality identification method of the present application has an identification accuracy of 100% for the period, origin and authenticity, and an identification accuracy of 90% for the mixed or dyed type of safflower.

Description

Technical Field

[0001] This invention belongs to the field of safflower identification technology, specifically relating to a method for identifying the quality of safflower based on near-infrared spectroscopy. Background Technology

[0002] Safflower, scientifically known as Carthamus tinctorius L Safflower, the dried flower of the safflower plant (Carthamus tinctorius), is also known as grass safflower, Huai safflower, and prickly safflower. It has the effects of promoting blood circulation, regulating menstruation, removing blood stasis, and relieving pain. It is a traditional Chinese medicine and an important raw material for health foods and cosmetics. Safflower is mainly produced in Xinjiang, Hunan, Zhejiang, Yunnan, and other places. The quality of safflower is affected by various factors such as the growth cycle and origin. The content of effective components in safflower varies significantly at different growth stages, and the quality of safflower from different producing areas also varies considerably due to differences in climate, soil, and other geographical conditions.

[0003] Currently, the main methods for identifying safflower include traditional morphological identification, microscopic identification, thin-layer chromatography identification, high-performance liquid chromatography identification, and near-infrared spectroscopy. Among these, near-infrared spectroscopy for safflower identification involves obtaining near-infrared spectra through Fourier transform near-infrared scanning, followed by analysis and comparison to identify the quality of safflower. However, the high accuracy of safflower identification is due to significant overlap in near-infrared spectra. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for identifying the quality of safflower based on near-infrared spectroscopy.

[0005] To facilitate understanding of this invention, the materials used in this invention and their abbreviations are listed below: Competitive adaptive reweighted sampling (CARS). Uninformative variable elimination (UVE). Variable importance projection (VIP). Combined interval partial least squares (SiPLS).

[0006] The purpose of this invention is to provide a method for identifying the quality of safflower based on near-infrared spectroscopy, comprising the following steps: Collect genuine safflower samples from different growth cycles and origins, and prepare adulterated or dyed samples.

[0007] Test samples of genuine and counterfeit safflower were placed at 12500cm. -1 ~4000cm -1 The raw spectral data is obtained from the near-infrared spectral data within the wavenumber range.

[0008] Multivariate scattering correction is used to preprocess the original spectral data to correct baseline shifts and intensity changes caused by scattering, resulting in preprocessed spectral data.

[0009] Using preprocessed spectral data as the dependent variable matrix and growth cycle, origin, and authenticity as the independent variable matrix, a regression model was developed using partial least squares-discriminant analysis to obtain a model for identifying the authenticity of safflower origin during its growth cycle.

[0010] Safflower samples were collected at 12500 cm. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range are used to import the near-infrared spectral data of the safflower sample to be tested into the safflower cycle origin authenticity identification model to determine the growth cycle, origin, and authenticity of the safflower sample.

[0011] Preferably, the method also includes a method for identifying the adulteration or staining type of counterfeit samples, comprising the following steps: Feature variables were selected from the preprocessed spectral data of counterfeit samples using the combined interval partial least squares method to obtain feature wavenumber variables. These feature wavenumber variables were used as the dependent variable matrix, and the categories of adulteration and staining were used as the independent variable matrix. Partial least squares-discriminant analysis was then used for regression modeling to obtain a model for identifying adulteration and staining categories of safflower.

[0012] The near-infrared spectral data of the counterfeit sample to be tested are imported into the safflower adulteration and staining category identification model to determine the category of adulteration or staining of the counterfeit sample to be tested.

[0013] To further identify the adulteration or staining type of counterfeit safflower, this invention uses a combined-interval partial least squares method to screen characteristic variables from the preprocessed spectral data of counterfeit samples. The safflower adulteration and staining type identification model established from the screened characteristic wavenumber variables has an accuracy of over 90% in identifying the adulteration or staining type of safflower, and can effectively distinguish the adulteration or staining type of counterfeit samples.

[0014] Preferably, the wavenumber in the characteristic wavenumber variable is 9933.814 cm⁻¹. -1 ~7391.648cm -1 .

[0015] Preferably, the different growth cycles are the early stage of initial flowering, the late stage of initial flowering, the early stage of full bloom, the middle stage of full bloom, the late stage of full bloom, the early stage of withering, and the late stage of withering.

[0016] Preferably, the different production areas are Henan, Sichuan, Yunnan and Xinjiang.

[0017] Preferably, the doping refers to safflower mixed with stone powder, sugar water, or alum water.

[0018] Preferably, the stone powder is barite powder or talc powder.

[0019] Preferably, the dyeing refers to safflower doped with synthetic dyes.

[0020] Preferably, the synthetic dye is any one of crimson, lemon yellow, sunset yellow, golden orange II, carmine, and annatto orange.

[0021] Preferably, the resolution of the near-infrared spectral data acquisition is 4 cm. -1 The number of scans was 32.

[0022] Compared with the prior art, the present invention has the following beneficial effects: 1. The near-infrared spectroscopy-based safflower quality identification method of this invention has a 100% accuracy rate in identifying safflower cycle, origin, and authenticity. This near-infrared spectroscopy-based safflower quality identification method can be used to identify safflower of different cycles, origins, and whether it is adulterated or dyed. It has a wide coverage of counterfeit products, high model specificity, strong adaptability of the modeling method, no overfitting, model stability, and high detection accuracy.

[0023] This invention relates to a near-infrared spectroscopy-based method for identifying the quality of safflower. It involves collecting genuine safflower samples from different growth stages and origins, and preparing adulterated or dyed samples. The genuine and adulterated safflower samples are tested at 12500 cm⁻¹. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range were used to obtain raw spectral data. Multivariate scattering correction was applied to preprocess the raw spectral data to correct baseline shifts and intensity variations caused by scattering, resulting in preprocessed spectral data. After multivariate scattering correction preprocessing, the PLS-DA model achieved 100% recognition rates for growth cycle, origin, and authenticity. Using the preprocessed spectral data as the dependent variable matrix and growth cycle, origin, and authenticity as the independent variable matrix, partial least squares-discriminant analysis (PLS-DA) regression model was used to obtain a safflower growth cycle origin authenticity identification model. The output of the safflower growth cycle origin authenticity identification model is the category label for safflower growth cycle, origin, and authenticity. Safflower samples were collected at 12500 cm⁻¹. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range are used to import the near-infrared spectral data of the safflower sample to be tested into the safflower cycle origin authenticity identification model to determine the growth cycle, origin, and authenticity of the safflower sample.

[0024] 2. This invention uses combined-interval partial least squares (PLLS) to screen characteristic variables in the preprocessed spectral data of counterfeit samples, obtaining characteristic wavenumber variables. Using the characteristic wavenumber variables as the dependent variable matrix and the adulteration and staining categories as the independent variable matrix, a regression model is performed using PLS-discriminant analysis to obtain a safflower adulteration and staining category identification model. The near-infrared spectral data of the counterfeit sample to be tested is imported into the safflower adulteration and staining category identification model to determine the adulteration or staining category of the counterfeit sample. To further identify the adulteration or staining type of counterfeit safflower, this invention uses combined-interval partial least squares to screen characteristic variables in the preprocessed spectral data of counterfeit samples. The safflower adulteration and staining category identification model established using the screened characteristic wavenumber variables has an accuracy of over 90% in identifying the adulteration or staining type of safflower, effectively distinguishing the adulteration or staining type of counterfeit samples.

[0025] The near-infrared spectroscopy-based method for identifying safflower quality of the present invention has a high recognition rate and good stability in distinguishing safflower of different growth periods, origins, and those that have been adulterated or dyed, providing a reliable means for the quality control and standardized application of safflower medicinal materials. Attached Figure Description

[0026] Figure 1 These are appearance diagrams of genuine safflower and safflower adulterated or dyed according to the present invention. In the diagrams, A represents safflower; B represents safflower adulterated with talc powder; C represents safflower adulterated with barite powder; D represents safflower adulterated with alum; E represents safflower adulterated with sugar water; F represents safflower dyed with bright red pigment; G represents safflower dyed with lemon yellow; H represents safflower dyed with sunset yellow; I represents safflower dyed with Acid Orange II; J represents safflower dyed with carmine; and K represents safflower dyed with annatto orange.

[0027] Figure 2 This is a diagram showing the distribution of main components of safflower from different growth periods, origins, authenticity, and adulterated or dyed safflowers according to the present invention. In the diagram, A represents the growth period, B represents the origin, C represents authenticity, and D represents adulteration or dyeing.

[0028] Figure 3 This is a near-infrared spectrum of safflower according to the present invention. A represents the average near-infrared spectrum over different periods. B represents the average near-infrared spectrum from different origins. C represents the average near-infrared spectrum of genuine safflower and adulterated or dyed products.

[0029] Figure 4 The result is a graph showing the characteristic wavenumber variables screened by SiPLS in this invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand and implement the technical solutions of the present invention, the following detailed description, in conjunction with preferred embodiments and accompanying drawings, provides a clear and complete account of the technical solutions of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0031] It should be noted that all technical terms used in this invention are for the purpose of describing specific embodiments only and are not intended to limit the scope of protection of this invention. Unless otherwise specified, all raw materials, reagents, instruments and equipment used in the following embodiments of this invention can be purchased from the market or prepared by existing methods.

[0032] 1. Preparation of safflower samples The safflower samples of this invention include those from different growth cycles and origins. Specifically: safflower samples from different growth cycles were all collected in Shihezi, Xinjiang (longitude 86.04°, latitude 44.31°). These safflower samples from different growth cycles cover the early flowering stage, late withering stage, and seven growth stages, totaling 105 samples. The collection time and cycle are shown in Table 1. This invention purchased safflower from Henan, Sichuan, Yunnan, and Xinjiang, and identified them according to the requirements of the 2025 edition of the Chinese Pharmacopoeia, which met the requirements. Specifically, the Henan safflower was purchased from Hehe Herbals, batch number 25080101, trade name: Safflower (Huai Safflower). The Sichuan safflower was purchased from Chengdu Xinfuyuan Traditional Chinese Medicine Beverage Co., Ltd., batch number C250250901, trade name: Safflower. The Yunnan safflower was purchased from Sichuan Yipianye Pharmaceutical Co., Ltd., batch number 250801, trade name: safflower. The Xinjiang safflower was purchased from Huizhou Fansheng Traditional Chinese Medicine Pieces Co., Ltd., batch number 250802, trade name: safflower.

[0033] Table 1. Safflower Collection Information The counterfeit samples of this invention were all prepared by adulterating or dyeing Xinjiang safflower. Specifically, they included safflower adulterated with talc powder, barite powder, alum, sugar water, bright red dye, lemon yellow dye, sunset yellow dye, Acid Orange II dye, carmine dye, and annatto orange dye, totaling 10 types of counterfeit samples, with 15 samples of each type, for a total of 150 samples. The specific preparation methods are as follows:

[0034] Xinjiang safflower, talc powder and mustard oil were mixed in a mass ratio of 100:25:2 and stirred until the powder was evenly coated with oil and there was no clumping. The mixture was then dried at 50°C for 20 minutes to obtain talc powder mixed with safflower.

[0035] Xinjiang safflower, barite powder, and rapeseed oil were mixed in a mass ratio of 40:10:1 and stirred for 8 minutes until the powder formed loose, non-clumping oil clumps, thus obtaining barite powder mixed with safflower.

[0036] Xinjiang safflower and 0.4 g / mL alum solution were mixed evenly at a mass ratio of 25:2 to obtain alum-adulterated safflower.

[0037] Xinjiang safflower and 0.6 g / mL sugar water were mixed evenly at a mass ratio of 25:3 to obtain safflower adulterated with sugar water.

[0038] Xinjiang safflower and 1 mg / mL bright red pigment were mixed evenly at a mass ratio of 1000:1 to obtain safflower dyed with bright red pigment.

[0039] Xinjiang safflower and 1 mg / mL lemon yellow were mixed evenly at a mass ratio of 1000:1 to obtain lemon yellow-stained safflower.

[0040] Xinjiang safflower and 1 mg / mL sunset yellow were mixed evenly at a mass ratio of 1000:1 to obtain sunset yellow-stained safflower.

[0041] Xinjiang safflower and 1 mg / mL Acid Orange II were mixed evenly at a mass ratio of 1000:1 to obtain safflower stained with Acid Orange II.

[0042] Xinjiang safflower and 1 mg / mL carmine were mixed evenly at a mass ratio of 1000:1 to obtain carmine-stained safflower.

[0043] Xinjiang safflower and 1 mg / mL annatto were mixed evenly at a mass ratio of 1000:1 to obtain annatto-stained safflower.

[0044] The appearance of genuine safflower and adulterated or dyed safflower of the present invention is as follows: Figure 1 As shown in the results, adulterated safflower containing talc has a whitish surface and a slightly loose texture. Adulterated safflower containing barite has a smooth surface, increased density, and a heavy feel. Adulterated safflower containing alum has a whitish surface with a crystalline luster. Adulterated safflower containing sugar water has a darker color, a sticky surface, and is prone to clumping. Safflower dyed with bright red pigment has an abnormally bright red color that looks unnatural. Safflower dyed with lemon yellow has an overall yellowish-green hue. Safflower dyed with sunset yellow has an orange-yellow color that is bright and uniform. Safflower dyed with acid orange II has an orange-red color with a relatively uniform pigment distribution. Safflower dyed with carmine has a bright red color with a slight fluorescent feel. Safflower dyed with annatto orange has an orange-red color that is uniform but somewhat dark. The appearance of adulterated or dyed safflower is similar to that of genuine safflower samples.

[0045] 2. PCA analysis of safflower samples Test samples of genuine and counterfeit safflower were placed at 12500cm. -1 ~4000cm-1 Near-infrared spectral data within the wavenumber range were used to obtain raw spectral data. PCA analysis was performed on the raw near-infrared spectral data of safflower to interpret the cumulative variance contribution rate and determine the PCA score. When the PCA score reached 2, the cumulative variance contribution rate exceeded 95%, so the first two PCA scores were selected for subsequent analysis. A score plot was generated using the scores from the first two PCA analyses. The score plot shows the distribution and clustering of different categories of samples in the sample space.

[0046] PCA was performed on safflower spectral data from 7 cycles, 4 origins, 10 adulteration methods, and the original spectra, and a two-dimensional plot was generated. The results are shown below. Figure 2 As shown in the figure. The results show that the contribution rate of the first PCA of safflower in different periods is 81.70%, the contribution rate of the second PCA is 17.34%, and the cumulative contribution rate of the first two PCA is 99.04%, which can reflect the information of the near-infrared spectrum in different periods. From Figure 2 As can be seen from A, safflower samples from different growth stages exhibit a relatively obvious clustering trend in PCA space. Samples from each stage are distributed in an orderly manner along the first PCA. There is some overlap between adjacent stage samples, but overall samples from different growth stages can be distinguished. This indicates that near-infrared spectroscopy combined with PCA can effectively reflect the continuous changes in chemical composition during the growth of safflower, providing a basis for stage identification modeling.

[0047] The contribution rate of the first PCA of safflower from different origins was 97.14%, the contribution rate of the second PCA was 2.65%, and the contribution rate of the first two PCA was 99.77%, which can reflect the information of the near-infrared spectra of different origins. From Figure 2 As shown in B, safflower samples from different origins exhibit significant clustering in the PCA space. Samples from Xinjiang are significantly separated from samples from other origins, while samples from Sichuan and Yunnan show some overlap. However, overall, the four types of samples show certain clustering areas, indicating that near-infrared spectroscopy combined with PCA can effectively distinguish safflower samples from different origins, providing feasibility for origin identification.

[0048] The contribution rate of the first PCA of adulterated safflower was 87.76%, the contribution rate of the second PCA was 9.57%, and the contribution rate of the first two PCA was 97.33%, which can reflect the information of the near-infrared spectrum of adulterated safflower. From Figure 2 As shown in C, the genuine, adulterated, and dyed safflower samples have clear boundaries in the PCA space, with the two types of samples showing obvious separation and no overlapping areas. Further subdividing them into 11 categories... Figure 2 As can be seen from D in the data, most samples that are adulterated or stained can also form independent or relatively concentrated clusters, indicating that near-infrared spectroscopy combined with PCA can not only effectively distinguish between genuine and counterfeit products, but also make preliminary distinctions between some types of adulteration, laying the foundation for the identification of genuine and counterfeit products and the identification of adulteration types.

[0049] Example 1 A method for identifying the quality of safflower based on near-infrared spectroscopy includes the following steps: 1. Construction of a model for identifying the authenticity of safflower origins during its life cycle (1) Collect genuine safflower samples from different growth cycles and origins, and prepare adulterated or dyed samples.

[0050] The different growth cycles are the early stage of initial flowering, the late stage of initial flowering, the early stage of full bloom, the middle stage of full bloom, the late stage of full bloom, the early stage of withering, or the late stage of withering.

[0051] The different production areas mentioned are Henan, Sichuan, Yunnan, and Xinjiang.

[0052] The doping refers to safflower adulterant samples doped with heavy talc powder, barite powder, sugar water, or alum water.

[0053] The staining was performed on a safflower adulterant sample containing synthetic dyes such as Crimson, Lemon Yellow, Sunset Yellow, Golden Orange II, Carmine, or Natto Orange.

[0054] Samples of genuine safflower and samples of adulterated or dyed safflower were prepared according to the method described in "Preparation of Safflower Samples".

[0055] (2) Test genuine and counterfeit safflower samples at 12500cm -1 ~4000cm -1 The raw spectral data is obtained from the near-infrared spectral data within the wavenumber range.

[0056] Near-infrared spectral data of genuine and counterfeit safflower samples were collected using a Bruker Tango-R near-infrared spectrometer, with scanning parameters set to a resolution of 4 cm⁻¹. -1 Against a blank background, the scan was performed 32 times, with a wavenumber range of 12500 cm⁻¹. -1 ~4000cm -1 Each sample was measured three times, and the average value was used as the raw spectral data. The Kennard-Stone algorithm was used to divide the raw spectral data of safflower samples from different growth stages, origins, and genuine / counterfeit products into a calibration set and a validation set. The number of safflower samples in the calibration and validation sets is shown in Table 2. The results show that the calibration set includes 70 genuine safflower samples from different growth stages, 50 genuine safflower samples from different origins, and 200 counterfeit samples. The validation set includes 35 genuine safflower samples from different growth stages, 25 genuine safflower samples from different origins, and 100 counterfeit samples.

[0057] Table 2. Safflower cycle, origin, number of genuine and counterfeit samples in the calibration and validation sets. The near-infrared spectral results of genuine and counterfeit safflower samples are as follows: Figure 3 As shown in the figure. The results show that in the near-infrared spectra of safflower samples from 7 growth cycles, 4 origins, and 10 adulterated or stained samples, the near-infrared spectral curves of safflower samples from different growth cycles, origins, and authenticitys showed high overlap and a generally consistent trend, with the main absorption peak concentrated at 4000 cm⁻¹. -1 ~10000cm -1 Within the specified range, some bands exhibit strong absorption signals, reflecting the spectral response characteristics of active components such as organic matter, flavonoids, and polysaccharides in the sample.

[0058] The near-infrared spectra of safflower at different growth stages are shown in the results. Figure 3 As shown in A in the figure. The results show that in the wavenumber range of 6900 cm⁻¹ -1 and 5150cm -1 A strong moisture absorption peak was observed at 8300 cm⁻¹. Simultaneously, a strong moisture absorption peak was observed at 8300 cm⁻¹. -1 6500cm -1 and 5800cm -1 Absorption features related to chemical bond vibrations such as CH and NH can be observed nearby, corresponding to carbohydrates, flavonoids, and phenolic substances. The spectral characteristics show a regular evolution as the growth cycle progresses. Samples from the early flowering stage show absorption characteristics at 6900 cm⁻¹. -1 and 5150cm -1 The highest absorption peak intensity is observed at 5800 cm⁻¹, indicating abundant water content. During peak flowering, the aforementioned water absorption peak weakens relatively, and the intensity shifts to 5800 cm⁻¹. -1 ~6200cm -1 Within the range, the CH absorption characteristics associated with flavonoids are significantly enhanced, reflecting the peak period of active ingredient accumulation. Towards the later stage of the decay period, the intensity of the water absorption peak continues to decline, and the absorption bands of some organic components show broadening and decreased intensity.

[0059] The near-infrared spectra of safflower from different origins are shown in the results. Figure 3 As shown in B in the figure. The results show that all spectra exhibit similar organic absorption patterns at a wavenumber of 6900 cm⁻¹. -1 and 5150cm -1 A significant moisture absorption peak exists at 8300 cm⁻¹. -1 5800cm -1 ~6200cm -1 The absorption ranges exhibited characteristic absorptions associated with active ingredients such as flavonoids and polysaccharides. However, there were systematic differences in the intensity of key absorption peaks among samples from different origins. Safflower from Xinjiang showed the strongest absorption, indicating a higher content of flavonoid components, such as hydroxysafflower yellow A. Other origins showed varying absorption intensities, with Sichuan and Yunnan samples exhibiting similar absorption strengths, reflecting differences in secondary metabolite accumulation due to geographical environmental variations.

[0060] The near-infrared spectra of genuine and counterfeit safflower are shown in the results. Figure 3 As shown in C in the figure. The results show that there are significant differences in spectral absorption between genuine safflower and adulterated and dyed samples. Adulterated products exhibit additional absorption peaks at specific wavenumbers, such as the talc-adulterated sample at 4600 cm⁻¹. -1 Abnormally increased absorption intensity, such as in samples stained with bright red pigment at 6000 cm⁻¹. -1 ~6500cm -1 Within the range, the spectral baseline of genuine samples is stable and the characteristic peaks are regular. The two major types of samples can be intuitively distinguished through spectral characteristics.

[0061] (3) The original spectral data of the calibration set is preprocessed by multivariate scattering correction to correct the baseline shift and intensity change caused by scattering in the spectral data, and to enhance the absorption information related to the chemical properties of the sample, so as to obtain preprocessed spectral data.

[0062] (4) Using the preprocessed spectral data as the dependent variable matrix and the growth cycle, origin, and authenticity as the independent variable matrix, a partial least squares-discriminant analysis algorithm is used for regression modeling to obtain the safflower cycle origin authenticity identification model. The output of the safflower cycle origin authenticity identification model is the category label of growth cycle, origin, and authenticity. Each genuine sample and each counterfeit sample corresponds to a set of category labels of growth cycle, origin, and authenticity, forming a database A for safflower quality identification. Database A contains the category labels of safflower cycle, origin, and authenticity as shown in Tables 3 to 5. The cycle identification results of the calibration set of the safflower cycle origin authenticity identification model are shown in Table 3. The origin identification results of the calibration set of the safflower cycle origin authenticity identification model are shown in Table 4. The sample authenticity identification results of the calibration set of the safflower cycle origin authenticity identification model are shown in Table 5.

[0063] 2. Construction of a model for identifying safflower adulteration through staining (1) The characteristic wavenumber variables were obtained by screening the preprocessed spectral data of the adulterated safflower in the calibration set using the combined interval partial least squares method. The characteristic wavenumber variables screened by SiPLS in this invention are as follows: Figure 4 As shown in the figure. The results show that SiPLS divides the full spectrum into multiple intervals, establishes multiple local PLS models by combining different intervals, and selects the optimal interval combination to improve the model's identification performance and interpretability. For the safflower sample, the full spectrum is divided into 10 intervals, and 3 interval combinations are used for modeling, resulting in a characteristic wavenumber variable of 9933.814 cm⁻¹. -1 ~7391.648cm -1 .

[0064] (2) Using the characteristic wavenumber variable as the dependent variable matrix and the categories of adulteration and staining as the independent variable matrix, a partial least squares-discriminant analysis algorithm is used for regression modeling to obtain the safflower adulteration and staining category identification model. The output of the safflower adulteration and staining category identification model is the category label of adulteration or staining of the counterfeit samples, where each genuine sample and counterfeit sample corresponds to a category label of authenticity, forming a database B for safflower quality identification. Database B contains the category labels of safflower authenticity as shown in Table 6. The authenticity category identification results of the calibration set of the safflower adulteration and staining category identification model are shown in Table 6.

[0065] 3. Identification of the safflower sample to be tested Safflower samples were collected at 12500 cm. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range are imported into the safflower cycle origin authenticity identification model to obtain cycle, origin, and authenticity category labels. The cycle, origin, and authenticity category labels are compared with database A to determine the growth cycle, origin, and authenticity of the safflower sample to be tested.

[0066] The near-infrared spectral data of the safflower sample to be tested is imported into the safflower adulteration and staining category identification model to obtain the adulteration or staining category label. The category label of the adulteration or staining category label is compared with the database B to determine the adulteration or staining category of the safflower sample to be tested.

[0067] The validation set data was used as the safflower samples to be tested for verification. The cycle identification results of the validation set for the safflower cycle origin authenticity identification model are shown in Table 7. The origin identification results of the validation set for the safflower cycle origin authenticity identification model are shown in Table 8. The sample authenticity identification results of the validation set for the safflower cycle origin authenticity identification model are shown in Table 9. The authenticity category identification results of the validation set for the safflower adulteration staining category identification model are shown in Table 10. The PLS-DA modeling results for different cycles, origins, and authenticity of safflower are shown in Table 11. The results show that after MSC preprocessing, when the latent variable factors for different origins, cycles, and authenticity categories of safflower are 6, 5, and 4 respectively, the identification accuracy of both the calibration and validation sets of the established models can reach 100%.

[0068] The results of SiPLS variable screening are as follows: Figure 4 As shown in the figure. The results show that the selected characteristic band is 9933.814 cm. -1 ~7391.648cm -1The obtained safflower cycle, origin, or authenticity identification model has 11 latent variable factors. The identification accuracy rate of the calibration set is 91%, and the identification accuracy rate of the validation set is 90%. The safflower cycle, origin, or authenticity identification model has an accuracy rate of over 90% in identifying the adulteration or dyeing types of genuine and counterfeit products, and can achieve relatively accurate identification of 11 different adulteration or dyeing categories and genuine safflower.

[0069] In Tables 3-10 and 13-15, the sample periods are 1 for early initial flowering, 2 for late initial flowering, 3 for early full bloom, 4 for mid-full bloom, 5 for late full bloom, 6 for early withering, and 7 for late withering. The sample origins are 1 for Henan, 2 for Sichuan, 3 for Yunnan, and 4 for Xinjiang. In the authenticity category, 1 indicates genuine products, and 2 indicates counterfeit products. The authenticity categories are 1 for bright red dye, 2 for adulterated with heavy talc, 3 for adulterated with heavy alum water, 4 for dyed with lemon yellow, 5 for dyed with sunset yellow, 6 for dyed with golden orange II, 7 for adulterated with heavy sugar water, 8 for dyed with carmine, 9 for dyed with annatto, 10 for adulterated with heavy barium sulfate, and 11 for genuine safflower. "-" indicates that this item is not present; the same applies below.

[0070] Table 3. Periodic identification results of the calibration set for the safflower periodic origin authenticity identification model. Table 4. Origin identification results of the calibration set of the safflower cycle origin authenticity identification model. Table 5. Authenticity and counterfeit identification results of the calibration set for the safflower cycle origin identification model. Table 6. Results of the authenticity classification of the calibration set for the safflower adulteration staining category identification model. Table 7. Periodic identification results of the validation set for the safflower periodic origin authenticity identification model. Table 8. Origin identification results of the validation set of the safflower cycle origin authenticity identification model. Table 9. Authenticity identification results of samples in the validation set of the safflower cycle origin authenticity identification model. Table 10. Validation results of the safflower adulteration staining category identification model in Example 1. Table 11. Accuracy of Safflower Cycle, Origin, and Authenticity Identification Comparative Example 1 A method for identifying the quality of safflower based on near-infrared spectroscopy includes the following steps: By replacing SiPLS with CARS in Example 1, while keeping all other conditions the same as in Example 1, a detailed identification model for genuine and counterfeit safflower was obtained. The obtained model for identifying genuine and counterfeit safflower had 12 latent variable factors, with an identification accuracy of 89.5% on the calibration set and 89% on the validation set.

[0071] Using 35 independent genuine and counterfeit safflower samples that were not involved in model training as an external validation set, we conducted identification of 11 categories of genuine and counterfeit products, with an accuracy rate of 82.86%.

[0072] Comparative Example 2 A method for identifying the quality of safflower based on near-infrared spectroscopy includes the following steps: By adjusting SiPLS to UVE in Example 1 while keeping all other conditions the same as in Example 1, a detailed identification model for genuine and counterfeit safflower was obtained. The obtained model for identifying genuine and counterfeit safflower had 12 latent variable factors, with an identification accuracy of 86.5% on the calibration set and 84% on the validation set.

[0073] Using 35 independent genuine and counterfeit safflower samples that were not involved in model training as an external validation set, we conducted identification of 11 categories of genuine and counterfeit products, achieving an accuracy rate of 80%.

[0074] Comparative Example 3 A method for identifying the quality of safflower based on near-infrared spectroscopy includes the following steps: By adjusting SiPLS to VIP in Example 1, while keeping all other conditions the same as in Example 1, a detailed identification model for genuine and counterfeit safflower was obtained. The obtained detailed identification model for genuine and counterfeit safflower has 12 latent variable factors, and the identification accuracy rate on the calibration set is 87%, while the identification accuracy rate on the validation set is also 87%.

[0075] Using 35 independent genuine and counterfeit safflower samples that were not involved in model training as an external validation set, the authenticity and counterfeit products were identified in 11 categories, with an accuracy rate of 82.86%.

[0076] The model identification accuracy rates of Example 1 and Comparative Examples 1 to 3 are shown in Table 12. The results show that among the models established by the CARS, UVE, VIP, and SiPLS variable selection methods, SiPLS has the highest accuracy rate, greater than or equal to 90%, while the accuracy rates of the other variable selection methods are all below 90%.

[0077] Table 12 Model identification accuracy of Example 1 and Comparative Examples 1-3 Comparative Example 4 An application of a near-infrared spectroscopy-based method for identifying safflower quality includes the following steps: The SiPLS method in Example 1 was omitted, while other conditions remained the same as in Example 1, resulting in a full-spectrum PLS model. The full-spectrum PLS model has 12 latent variable factors. Thirty-five independent genuine and counterfeit safflower samples that were not involved in model training were used as an external validation set to identify 11 categories of genuine and counterfeit products, achieving an accuracy rate of 85.71%.

[0078] Application Example 1 An application of a near-infrared spectroscopy-based method for identifying safflower quality includes the following steps: 1. Preparation of safflower blind samples Fourteen completely independent safflower blind samples were randomly selected. All safflower blind samples were confirmed by morphological and microscopic identification according to the 2025 edition of the Chinese Pharmacopoeia. The composition of the safflower blind samples is as follows:

[0079] The study covered seven growth cycles from the early flowering stage to the late withering stage, with two samples per cycle, totaling 14 samples, all collected from the Shihezi planting base in Xinjiang.

[0080] 2. Identification of safflower blind samples The detection parameters and methods were exactly the same as in Example 1: a Bruker Tango-R near-infrared spectrometer was used, with the wavenumber range set to 12500 cm⁻¹. -1 ~4000cm -1 4cm resolution -1 The scan was performed 32 times, with the gold mirror blank as the background. Each blind sample was measured 3 times, and the average value was taken as the original spectral data. The spectral data was preprocessed by multivariate scattering correction. The preprocessed spectral data was then imported into the safflower periodic origin authenticity identification model constructed in Example 1, and the periodic label of the safflower blind sample was output.

[0081] 3. Identification results of safflower blind samples The blind sample prediction results for safflower cycle are shown in Table 13. The results show that the prediction results for 14 safflower samples from different cycles are completely consistent with the true values, with an accuracy of 100%.

[0082] Table 13 Results of blind sample prediction of safflower cycle Application Example 2 An application of a near-infrared spectroscopy-based method for identifying safflower quality includes the following steps: 1. Preparation of safflower blind samples Eight completely independent safflower blind samples were randomly selected. All safflower blind samples were confirmed by morphological and microscopic identification according to the 2025 edition of the Chinese Pharmacopoeia. The composition of the safflower blind samples is as follows:

[0083] Authentic safflower from different origins: 2 samples from each of the four main producing areas of Henan, Sichuan, Yunnan and Xinjiang, totaling 8 samples, all from different batches of commercially available compliant medicinal materials.

[0084] 2. Identification of safflower blind samples The detection parameters and methods were exactly the same as in Example 1: a Bruker Tango-R near-infrared spectrometer was used, with the wavenumber range set to 12500 cm⁻¹. -1 ~4000cm -1 4 cm resolution -1 The scan was performed 32 times, with the gold mirror blank as the background. Each blind sample was measured 3 times, and the average value was taken as the original spectral data. The spectral data was preprocessed by multivariate scattering correction. The preprocessed spectral data was then imported into the safflower cycle origin authenticity identification model constructed in Example 1, and the origin label of the sample to be tested was output simultaneously.

[0085] 3. Identification results of safflower blind samples The results of the blind sample identification of safflower from different origins are shown in Table 14. The results show that the identification results of the 8 safflower samples from different origins are completely consistent with the true values, with an accuracy rate of 100%.

[0086] Table 14 Results of blind sample identification of safflower origin Application Example 3 An application of a near-infrared spectroscopy-based method for identifying safflower quality includes the following steps: 1. Preparation of safflower blind samples A total of 35 completely independent safflower blind samples were randomly selected. All safflower blind samples were confirmed by morphological and microscopic identification according to the 2025 edition of the Chinese Pharmacopoeia. The composition of the safflower blind samples is as follows:

[0087] The study included 10 types of adulterants: talc powder, barite powder, alum, sugar water, bright red dye, lemon yellow dye, sunset yellow dye, acid orange II dye, carmine dye, and annatto orange dye. Two samples of each type were included, along with 15 samples of genuine safflower, for a total of 35 samples.

[0088] 2. Identification of safflower blind samples The detection parameters and methods were exactly the same as in Example 1: a Bruker Tango-R near-infrared spectrometer was used, with the wavenumber range set to 12500 cm⁻¹. -1 ~4000cm -1 4cm resolution-1 The scan was performed 32 times, with a blank gold mirror as the background. Each blind sample was measured three times, and the average value was taken as the raw spectral data. Multivariate scattering correction was used to preprocess the spectral data. The preprocessed spectral data was then imported into the safflower cycle origin authenticity identification model constructed in Example 1, and the authenticity label of the safflower blind sample was output.

[0089] 3. Identification results of safflower blind samples The identification results of Application Example 3 and Comparative Examples 1-4 are shown in Table 15. The accuracy rates of authenticity classification for Application Example 3 and Comparative Examples 1-4 are shown in Table 16. The results show that for 20 adulterated samples and 15 genuine safflower samples, the accuracy rate for the binary classification of authenticity is 100%, and the accuracy rate for the detailed classification of adulteration types is 91.43%.

[0090] Table 15 shows the authenticity classification results for Application Example 3 and Comparative Examples 1-4. Table 16. Accuracy of Authenticity Classification and Identification in Application Example 3 and Comparative Examples 1-4 in conclusion This invention collected near-infrared spectra of safflower samples from seven different growth stages, four different origins, and ten types of adulterated or stained samples. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used to construct a rapid identification model for safflower quality attributes. Results showed that after appropriate spectral preprocessing, PCA could effectively distinguish different categories of samples, especially clearly differentiating genuine from counterfeit samples. The PLS-DA model performed excellently across different growth stages, origins, and authenticity criteria, with validation set recognition rates mostly reaching 100%, demonstrating high model accuracy. The method established in this invention enables rapid and non-destructive identification of safflower growth stages, geographical origins, and authenticity attributes simultaneously, providing reliable technical support for the quality control of safflower medicinal materials and their products.

[0091] It should be noted that when numerical ranges are involved in this invention, it should be understood that both endpoints of each numerical range and any value between the two endpoints can be selected. Since the steps and methods used are the same as in the embodiments, preferred embodiments are described in this invention to avoid redundancy. Although preferred embodiments of this invention have been described, those skilled in the art, once they understand the inventive concept of this invention, can make other changes and modifications to these embodiments, and all such changes and modifications fall within the scope of this invention.

[0092] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. If such modifications and variations fall within the scope of equivalents of this invention, then this invention is also intended to include such modifications and variations.

Claims

1. A method for quality identification of safflower based on near-infrared spectroscopy, characterized in that, Includes the following steps: Collect genuine safflower samples from different growth cycles and origins, and prepare adulterated or dyed samples. Test samples of genuine and counterfeit safflower were placed at 12500cm. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range are used to obtain raw spectral data; Multivariate scattering correction is used to preprocess the original spectral data to correct the baseline shift and intensity changes caused by scattering in the spectral data, resulting in preprocessed spectral data; Using preprocessed spectral data as the dependent variable matrix and growth cycle, origin, and authenticity as the independent variable matrix, regression modeling was performed using partial least squares-discriminant analysis algorithm to obtain the safflower cycle origin authenticity identification model. Safflower samples were collected at 12500 cm. -1 ~4000cm -1 Near-infrared spectral data within the wavenumber range are used to import the near-infrared spectral data of the safflower sample to be tested into the safflower cycle origin authenticity identification model to determine the growth cycle, origin, and authenticity of the safflower sample.

2. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, It also includes methods for identifying adulteration or staining categories in counterfeit samples, comprising the following steps: The characteristic wavenumber variables were obtained by screening the preprocessed spectral data of counterfeit samples using the combined interval partial least squares method. Using the characteristic wavenumber variable as the dependent variable matrix and the categories of adulteration and staining as the independent variable matrix, a regression model was performed using the partial least squares-discriminant analysis algorithm to obtain a safflower adulteration and staining category identification model. The near-infrared spectral data of the counterfeit sample to be tested are imported into the safflower adulteration and staining category identification model to determine the category of adulteration or staining of the counterfeit sample to be tested.

3. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, The wavenumber in the characteristic wavenumber variable is 9933.814 cm. -1 ~7391.648cm -1 .

4. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, The different growth cycles are the early stage of initial flowering, the late stage of initial flowering, the early stage of full bloom, the middle stage of full bloom, the late stage of full bloom, the early stage of withering, and the late stage of withering.

5. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, The different production areas mentioned are Henan, Sichuan, Yunnan, and Xinjiang.

6. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, The term "admixture" refers to safflower that has been mixed with stone powder, sugar water, or alum water.

7. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 6, characterized in that, The stone powder is barite powder or talc powder.

8. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 1, characterized in that, The dyeing refers to safflower adulterated with synthetic dyes.

9. The method for identifying safflower quality based on near-infrared spectroscopy according to claim 8, characterized in that, The synthetic dye is any one of crimson, lemon yellow, sunset yellow, golden orange II, carmine, and annatto orange.

Images