Rod chart assisted optimization of dispersible liquid-liquid microextraction back-extraction deodorization method for xiangdan injection

Abstract

The application provides a stick chart assisted optimization method for dispersing liquid-liquid microextraction-back extraction deodorization of Xiangdan injection, and relates to the field of traditional Chinese medicine preparation quality control. The method comprises the following steps: sample solution pretreatment, liquid phase analysis, DAD information extraction and stick chart generation, and stick chart information matching. Through dispersing liquid-liquid microextraction-back extraction technology, deodorization substances are enriched, and corresponding high performance liquid chromatography-diode array detector (HPLC-DAD) analysis data are extracted to form a stick chart, so as to realize multi-scale characterization of deodorization components in Xiangdan injection. The application first proposes a stick chart assisted dispersing liquid-liquid microextraction-back extraction method for detecting deodorization components in Xiangdan injection, and provides a new idea for traditional Chinese medicine quality control and evaluation.

Description

Technical Field

[0001] This invention belongs to the field of quality control of traditional Chinese medicine preparations, and specifically relates to a method for quality control of the aromatic component in Xiangdan injection. Specifically, it involves a method for optimizing the microextraction-back-extraction of aromatic components in the dispersion of Xiangdan injection using bar graphs. Background Technology

[0002] Xiangdan injection is a combination of saturated aromatic water of Dalbergia odorifera and water-extracted alcohol-precipitated salvia miltiorrhiza in a certain ratio of raw materials. Salvia miltiorrhiza is the principal drug of this preparation, and its phenolic acid components are the main active ingredients of Xiangdan injection. Mei Chao [Mei Chao. Study on the effect of Dalbergia odorifera on the absorption kinetics and mechanism of Salvia miltiorrhiza [D]. Changchun University of Traditional Chinese Medicine, 2011] proposed that the presence of Dalbergia odorifera can promote the distribution of the effective components of Salvia miltiorrhiza into the body, reduce the clearance rate, and play a synergistic role; Gao Shan et al. [Gao Shan et al. Study on the quality of Dalbergia odorifera in Xiangdan injection [J]. Chinese Journal of Drug Evaluation, 2017, 34(2):104-106.] proposed that there are many manufacturers of Xiangdan injection nowadays, and the Dalbergia odorifera components contained in the injection have great differences. The current standard for Xiangdan injection WS3-B-3289-98 only limits the effective components of Salvia miltiorrhiza and ignores the evaluation of Dalbergia odorifera components, which makes the quality control of Dalbergia odorifera in Xiangdan injection more serious. To further ensure the stability of the therapeutic effect of Xiangdan injection, it is necessary to establish a universal method for evaluating the quality of Dalbergia odorifera in formulations.

[0003] Lu Qian et al. [Lu Qian et al. Simultaneous determination of water-soluble components and nerolidol in Xiangdan injection by column-switching chromatography [J]. Chinese Journal of Traditional Chinese Medicine, 2008, 33(23):2776-2780] proposed that the content of Dalbergia odorifera components and Salvia miltiorrhiza components in Xiangdan injection differs significantly, and conventional analytical methods are difficult to detect trace amounts of Dalbergia odorifera components. Therefore, certain enrichment methods must be used to detect Dalbergia odorifera components in the preparation. However, the column-switching chromatography method used in this literature to enrich Dalbergia odorifera components has high organic solvent consumption, long extraction time, and low extraction efficiency and enrichment rate.

[0004] Dispersion-liquid microextraction (DLME) is a sample pretreatment technique developed based on traditional liquid-liquid extraction (LLME). Compared to traditional LLME, by adding dispersants or using physical methods such as ultrasound, vortexing, or air-assisted extraction, the organic extractant rapidly forms tiny droplets in the aqueous phase, thereby increasing dispersibility and the contact area between the aqueous phase and the extractant. This allows the target compound to be rapidly transferred between the sample solution and the extractant. This method uses low amounts of organic solvent, has short extraction times, and high extraction efficiency and enrichment rates, meeting the needs of modern green chemistry. Currently, there are no reports on using dispersion-liquid microextraction-back-extraction as a pretreatment method for Xiangdan injection to enrich its components. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a bar graph-assisted method for optimizing the dispersion liquid-liquid microextraction-back-extraction of Dalbergia odorifera in Xiangdan injection. By enriching the Dalbergia odorifera compound group through dispersion liquid-liquid microextraction-back-extraction technology, the corresponding high performance liquid chromatography-diode array detector (HPLC-DAD) analysis data is extracted to form a bar graph, thereby achieving multi-scale characterization of Dalbergia odorifera components in Xiangdan injection.

[0006] This invention provides a method for optimizing the dispersion liquid-liquid microextraction-back-extraction of Xiangdan injection with bar graph-assisted optimization, comprising: sample pretreatment, liquid phase analysis, DAD information extraction and bar graph generation, and bar graph information matching, wherein the sample pretreatment includes dispersion liquid-liquid microextraction-back-extraction technology.

[0007] Preferably, the dispersion liquid-liquid microextraction includes vortex-assisted dispersion liquid-liquid microextraction.

[0008] More preferably, the rotation speed of the vortex-assisted dispersion liquid-liquid microextraction is 3000-4000 rpm.

[0009] Preferably, the volume of the extractant in the dispersive liquid-liquid microextraction is 100-400 μL.

[0010] Preferably, the extractant for the dispersive liquid-liquid microextraction is a n-alkane.

[0011] More preferably, the extractant is n-tetradecane.

[0012] Preferably, the extraction time for dispersive liquid-liquid microextraction is 30-120 s.

[0013] Preferably, the volume of the extractant used for back-extraction is 50-200 μL.

[0014] Preferably, the extraction time for back-extraction is 30-120 seconds.

[0015] Preferably, the liquid phase analysis is performed using a high-performance liquid chromatography-diode array detector.

[0016] Preferably, the chromatographic conditions for liquid phase analysis are as follows: Agilent ZorBax SB C18 column (4.6 mm × 250 mm, 5.0 μm); column temperature 30 °C; flow rate 1.0 mL·min⁻¹; detection wavelength of DAD detector 200-400 nm; injection volume 10 μL; and dual mobile phase gradient elution.

[0017] More preferably, the two mobile phases are acetonitrile and pure water.

[0018] More preferably, the gradient elution program is as follows: 0-8 min, acetonitrile 34%, pure water 66%; 8-22 min, acetonitrile linearly increases from 34% to 48%, pure water linearly decreases from 66% to 52%; 22-30 min, acetonitrile linearly increases from 48% to 53%, pure water linearly decreases from 52% to 47%; 30-42 min, acetonitrile linearly increases from 53% to 65%, pure water linearly decreases from 47% to 35%; 42-50 min, acetonitrile linearly increases from 65% to 90%, pure water linearly decreases from 35% to 10%; 50-55 min, acetonitrile linearly increases from 90% to 95%, pure water linearly decreases from 10% to 5%; 55-66 min, acetonitrile 95%, pure water 5%.

[0019] Preferably, the DAD information extraction and bar chart generation includes the following steps:

[0020] Step (1) Export DAD data from the high performance liquid chromatography workstation and use the maximum value filter to quickly locate the retention information of the substance peaks;

[0021] Step (2) uses the airPLS method for band-by-band baseline correction, and uses a relaxed transformation method to find the peak start and end points based on the baseline and slope information.

[0022] Step (3) merge the same material peaks by whether the starting point and the ending point are subsets of each other, and retain the band information with the largest peak response to form a bar chart.

[0023] Preferably, the bar graph information matching is performed by setting the bar graph of Xiangdan injection as a benchmark, and then matching each peak of Xiangdan injection with the bar graphs of Jiangxiang test sample and Danshen test sample.

[0024] Preferably, the traversal matching process includes:

[0025] S1) Determine whether there is a peak in the sample to be matched that has a relative retention time of no more than ±2% with the reference peak. If so, decompose the peak interval data of the reference peak and the peak to be matched by SVD, and reconstruct the data after retaining 95% of the largest principal component. Otherwise, it is considered that there is no corresponding matching peak for the reference peak, and the matching of the next reference peak begins.

[0026] S2) The spectra of the peak elution ranges of the reference peak and the peak to be matched are superimposed and normalized to form an average spectrum;

[0027] S3) If the cosine similarity between the two peaks is greater than 0.97, the peaks are saved in the successfully matched sequence. If the cosine similarity is greater than or equal to 0.97 and greater than or equal to 0.85, the spectra of each retained time point in the peak interval of the peak to be matched are extracted and compared with the average spectrum of the reference peak. If there are peak intervals in the peak interval of the peak to be matched with a similarity greater than 0.97, the peaks are considered to be successfully matched; otherwise, the peaks are considered to be unmatched. If the cosine similarity is less than 0.85, the peaks are considered to be unmatched. Finally, if there are two or more peaks in the sample to be matched that are successfully matched with the reference peak, the peak with the largest cosine similarity is taken as the matching result.

[0028] Compared with existing technologies, the bar graph-assisted optimization method for the micro-extraction-back-extraction of Xiangdan injection dispersion has the following advantages:

[0029] (1) The bar graph-assisted optimization method for the dispersion liquid-liquid microextraction-back-extraction of the fragrant powder described in this invention has strong selective enrichment and can basically eliminate the interference of the Danshen component and enrich the fragrant powder component.

[0030] (2) The bar graph-assisted optimization method for the dispersion liquid-liquid microextraction-back-extraction of Xiangdan injection described in this invention does not require special instruments or equipment, and the amount of reagents used is small. Compared with the traditional liquid-liquid microextraction commonly used as a pretreatment method for Xiangdan injection, the pretreatment time is greatly shortened.

[0031] (3) The bar graph-assisted optimization method for the dispersion liquid-liquid microextraction-back-extraction of the fragrant powder injection described in this invention first discovered a large number of fragrant powder component peaks in HPLC analysis. Previous liquid phase analysis could only point out individual fragrant powder peaks or enrich only a single indicator substance.

[0032] (4) The bar graph-assisted optimization method for the dispersion liquid microextraction-back extraction of styrax in Xiangdan injection described in this invention can quickly grasp the properties of the analytical solution through full-band information extraction. Compared with traditional fingerprint spectrum, it can more comprehensively and thoroughly reflect the information of styrax components in Xiangdan injection. Attached Figure Description

[0033] Figure 1 Chromatograms of the distillate of Dalbergia odorifera at different wavelengths;

[0034] Figure 2 This is a rod-shaped diagram of the sandalwood distillate constructed according to the present invention;

[0035] Figure 3 Results of solvent selection for the dispersible liquid-liquid microextraction-back-extraction process;

[0036] Figure 4 Results of solvent screening for back-extraction in the dispersible liquid-liquid microextraction-back-extraction procedure;

[0037] Figure 5Results of extractant volume optimization in the dispersible liquid-liquid microextraction-back-extraction process;

[0038] Figure 6 Optimization results of extraction time in the dispersive liquid-liquid microextraction-back-extraction process;

[0039] Figure 7 Results of back-extraction agent volume optimization in dispersion liquid-liquid microextraction-back-extraction process;

[0040] Figure 8 Optimization results of back-extraction time in the dispersion liquid-liquid microextraction-back-extraction procedure;

[0041] Figure 9 Comparison of chromatograms of Dalbergia odorifera, Salvia miltiorrhiza, and Xiangdan injection at 210 nm wavelength;

[0042] Figure 10 The bar graph comparison results of the Dalbergia odorifera test sample, Salvia miltiorrhiza test sample, and Xiangdan injection constructed for this invention;

[0043] Figure 11 The bar graph of Xiangdan injection was transformed into a classic chromatogram with optimal absorption of each peak across the entire wavelength range;

[0044] Figure 12 A schematic diagram of common peaks calibrated for fingerprinting methodology;

[0045] Figure 13 Comparison of chromatograms of Xiangdan injection at 210 nm wavelength after dispersive liquid-liquid microextraction and conventional liquid-liquid extraction;

[0046] Figure 14 The results of bar graph comparison of Xiangdan injection after dispersion liquid-liquid microextraction and traditional liquid-liquid extraction. Detailed Implementation

[0047] It is worth noting that the raw materials used in this invention are all commercially available products, and their sources are not specifically limited.

[0048] Example 1

[0049] (1) Analysis of sample source and preparation

[0050] The test sample of Danshen was prepared by decocting Danshen with water three times. The first decoction was prepared with 8 times the amount of water, and the second and third decoctions were prepared with 6 times the amount of water. Each decoction was prepared for 1.5 hours. The filtrate was concentrated and subjected to three alcohol precipitation processes to obtain the final product.

[0051] Dalbergia odorifera sample: Dalbergia odorifera (wind-separated to remove impurities) is soaked in water, steam distilled, and about 500 mL of distillate is collected. The distillate is refrigerated for more than 24 hours, the oil layer is separated, and the aqueous solution is collected in a separate container.

[0052] Xiangdan Injection: The actual product of Zhengda Qingchunbao Pharmaceutical Co., Ltd. (batch number: 2303022, specification: 10mL / vial).

[0053] (2) Sample pretreatment method

[0054] The samples of Dalbergia odorifera and Salvia miltiorrhiza, along with the Xiangdan injection solution, were prepared according to the following steps: 200 μL of tetradecane was added to a 5 mL centrifuge tube, followed by 3 mL of the sample solution. The tube was vortexed at 3000 rpm for 60 s, and then centrifuged at 4000 rpm for 3 min to obtain the extract. The upper organic liquid was aspirated using a microsyringe and transferred to a 1 mL centrifuge tube. 50 μL of acetonitrile was added to the 1 mL centrifuge tube, and the tube was vortexed at 3000 rpm for 90 s. The tube was then centrifuged again at 4000 rpm for 3 min to obtain the back-extract. The lower back-extract (acetonitrile solution) was directly aspirated using a disposable syringe with a needle and injected into a liquid chromatography vial with an inner tube. The sample preparation was then complete.

[0055] (3) Liquid phase analysis process

[0056] The chromatographic column was an Agilent ZorBax SB C18 (4.6 mm × 250 mm, 5.0 μm); the column temperature was 30 ℃; the flow rate was 1.0 mL·min⁻¹; the detection wavelength of the DAD detector was 200-400 nm; the injection volume was 10 μL; acetonitrile was used as mobile phase A, and pure water was used as mobile phase B; the gradient elution program is shown in Table 1.

[0057] Table 1 Gradient elution conditions

[0058]

[0059] (4) DAD information extraction and bar chart generation

[0060] DAD data of the corresponding samples were exported from the chromatography workstation. A maximum value filter was applied to the exported data using a programming language to quickly locate the retention times of each peak. Baseline correction was performed band by band using the airPLS method. Finally, based on baseline and slope information, a relaxed transformation method was used to find the peak start and end points. Peaks of the same substance were merged based on whether their start and end points were subsets of each other, retaining the band information with the largest peak response to form a sequence such as... Figure 2 The bar chart shown.

[0061] (5) Bar chart information matching

[0062] By setting a bar chart of a specific sample as a baseline, each peak of that peak is matched against the bar charts of the remaining samples. Specifically, the matching process includes: determining if there is a peak in the sample to be matched whose relative retention time to the baseline peak is no more than ±2%. If so, the peak interval data of both the baseline peak and the peak to be matched are decomposed using SVD, and the data is reconstructed after retaining 95% of the largest principal components. The spectra of the peak intervals of the baseline peak and the peak to be matched are superimposed and normalized to form an average spectrum. If the cosine similarity between the two peaks is greater than 0.97, it is saved in the successfully matched sequence. If the cosine similarity is greater than 0.85 and less than 0.97, the spectra of each retention time point in the peak interval of the peak to be matched are extracted and compared with the average spectrum of the baseline peak. If there is a peak interval point with a similarity greater than 0.97, it is considered a successful match; otherwise, it is considered a failed match. If the cosine similarity is less than 0.85, it is considered a no-match. Finally, if there are two or more peaks in the sample to be matched that successfully match the baseline peak, the one with the largest cosine similarity is taken as the matching result.

[0063] Example 2: Solvent Screening for Dispersion Liquid-Liquid Microextraction-Back-Extraction Process

[0064] The types of extractants used in dispersion-liquid microextraction and back-extraction were investigated using a bar graph program, with *Dalbergia odorifera* as the target analyte. Chromatograms at different wavelengths are shown below. Figure 1 The results of the bar chart formation are shown below. Figure 2 Multiple batches of experiments confirmed that the stable occurrence of the aromatic compounds' baseline peaks is marked with an asterisk (*). Under constant conditions, the extraction effects of different types of n-alkanes (heptane, nonane, dodecane, and tetradecane) were screened, and the bar graph results are shown below. Figure 3 As shown in the figure. The results indicate that tetradecane has a better enrichment effect than other alkanes. Compared with alkanes with lower boiling points such as heptane, it is less volatile and better meets the requirements of dispersion liquid-liquid microextraction procedures. Different types of back-extraction agents (methanol, acetonitrile) were screened, and the bar chart results are shown in the figure. Figure 4 As shown. Acetonitrile is less volatile than methanol and has a more ideal extraction effect, therefore acetonitrile was chosen as the back-extraction solvent.

[0065] Example 3: Optimization of Dispersion Liquid-Liquid Microextraction-Back-Extraction Process

[0066] The volume of the extractant, extraction time, and the volume and time of the back-extraction solution in the liquid-liquid microextraction were optimized using bar graphs and peak matching procedures. Using *Dalbergia odorifera* as the target analyte, the effects of extractant volume, extraction time, back-extraction agent volume, and back-extraction time on the overall extraction efficiency of the stably appearing *Dalbergia odorifera* peak group were investigated, with the *Dalbergia odorifera* sample solution volume, centrifugation speed, and centrifugation time fixed. The weights of each indicator in the comprehensive evaluation were adjusted using the CRITIC method. Finally, the scores of each single-factor optimization scheme were comprehensively evaluated using TOPSIS, and the highest score was taken as the optimization result. The comprehensive score results are shown in Table 2. The final optimized microextraction conditions were 200 μL extractant, 60 s extraction time, 50 μL back-extraction solution, and 90 s back-extraction time. The optimization results are shown in Table 2. Figure 5-8 .

[0067] Table 2 Optimization Score Table for Dispersion Liquid-Liquid Microextraction

[0068]

[0069]

[0070] Example 4 Characterization of the Dalbergia odorifera component in Xiangdan injection

[0071] The optimized dispersion-liquid microextraction-back-extraction procedures of Examples 2 and 3 were applied to the *Salvia miltiorrhiza* test sample, the *Dalbergia odorifera* test sample, and *Xiangdan* injection. The chromatograms of the three at 210 nm are shown below. Figure 9 The results of the bar chart formation for the three elements are shown below. Figure 10 Taking the bar graph of Xiangdan injection as an example, the bar graph results were converted into a classic chromatogram with optimal absorption across the entire wavelength range, as shown in the following figure. Figure 11 As shown.

[0072] Using the bar graph of Xiangdan injection as a benchmark, peak matching was performed on the other two samples, and the assignment results of the corresponding components are shown in Table 3.

[0073] Table 3. Classification of corresponding components in Salvia miltiorrhiza and Dalbergia odorifera.

[0074]

[0075] Example 5: Methodological Investigation of Fingerprint Spectroscopy for Dalbergia odorifera Components in Xiangdan Injection

[0076] (1) Precision test

[0077] The Xiangdan injection sample was prepared using the pretreatment method of Example 1. 10 μL was injected each time, for a total of 6 injections. The relative retention time and relative peak area of ​​each peak were determined according to the chromatographic conditions. The results are as follows: Figure 12As shown, the relative retention times and areas of the remaining 10 common peaks were calculated using peak ④ as the reference peak. Tables 4 and 5 provide the corresponding results. The RSDs of the relative retention times and relative peak areas of each chromatographic peak are 0.06%-0.12% and 0.13%-0.71%, respectively, indicating good instrument precision.

[0078] Table 4 Precision Test - Relative Retention Time (n=6)

[0079]

[0080] Table 5 Precision Test - Relative Peak Area (n=6)

[0081]

[0082]

[0083] (2) Stability test

[0084] Samples of Xiangdan injection were prepared using the pretreatment method of Example 1. 10 μL of each sample was injected at 0, 2, 4, 6, 8, 10, 12, and 24 hours. Chromatograms were recorded, and the relative retention time and relative peak area of ​​each peak were calculated. The results are shown in Tables 6 and 7. The RSDs were 0.06%–0.21% and 0.28%–2.20%, respectively, indicating that the test solution had good stability within 24 hours.

[0085] Table 6 Stability Test - Relative Retention Time

[0086]

[0087] Table 7 Stability Test - Relative Peak Area

[0088]

[0089]

[0090] (3) Repeatability test

[0091] Six samples of the same batch of Xiangdan injection solution were prepared using the pretreatment method in Example 1. 10 μL of each sample was injected into the chromatograph, the chromatograms were recorded, and the relative retention time and relative peak area of ​​each peak were calculated. The results are shown in Tables 8 and 9. The RSDs were 0.04%-0.15% and 0.49%-2.47%, respectively, indicating that the method has good repeatability.

[0092] Table 8 Repeatability Tests - Relative Retention Times (n=6)

[0093]

[0094] Table 9 Repeatability Tests - Relative Peak Areas (n=6)

[0095]

[0096] Comparative experiment

[0097] Sample solution (XD-M) was obtained by pretreatment of Xiangdan injection solution according to the dispersion-liquid-liquid microextraction-back-extraction steps in Example 1; 10 mL of Xiangdan injection solution from the same batch was extracted three times with n-hexane, 8 mL each time, and the extracts were combined. The extracts were evaporated to dryness and reconstituted with anhydrous methanol to 1 mL to obtain sample solution (XD-U). The classical chromatogram at 210 nm obtained under the chromatographic conditions is shown below. Figure 13 As shown, the bar chart data is as follows: Figure 14 As shown in the figure. The results indicate that the pretreatment method for dispersive liquid-liquid microextraction (DLME) yields significantly better peak enrichment, more peaks, less solvent consumption, and shorter operation time compared to traditional DLME.

[0098] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, and is not intended to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention do not depart from the essence and scope of the technical solution of the present invention.

Claims

1. A method for optimizing the microextraction-back-extraction of a fragrant powder dispersion using bar graph-assisted microextraction, characterized in that, The method includes: sample pretreatment, liquid phase analysis, DAD information extraction and bar chart generation, and bar chart information matching; The sample pretreatment includes dispersion liquid-liquid microextraction-back-extraction technology; The extractant for the dispersion liquid-liquid microextraction is n-alkane; the volume of the extractant for the dispersion liquid-liquid microextraction is 200 μL; and the extraction time for the dispersion liquid-liquid microextraction is 60 s. The volume of the extractant used in the back-extraction was 50 μL, and the extraction time was 90 s. The chromatographic conditions for the liquid phase analysis were as follows: an Agilent ZorBax SB C18 column with dimensions of 4.6 mm × 250 mm and a diameter of 5.0 μm; a column temperature of 30 °C; and a flow rate of 1.0 mL / min. -1 The detection wavelength of the DAD detector is 200-400 nm; the injection volume is 10 μL, and a dual-mobile-phase gradient elution is used. The dual mobile phases are acetonitrile and pure water, and the gradient elution program is as follows: 0-8 min, acetonitrile 34%, pure water 66%; 8-22 min, acetonitrile linearly increases from 34% to 48%, pure water linearly decreases from 66% to 52%; 22-30 min, acetonitrile linearly increases from 48% to 53%, pure water linearly decreases from 52% to 47%; 30-42 min, acetonitrile linearly increases from 53% to 65%, pure water linearly decreases from 47% to 35%; 42-50 min, acetonitrile linearly increases from 65% to 90%, pure water linearly decreases from 35% to 10%; 50-55 min, acetonitrile linearly increases from 90% to 95%, pure water linearly decreases from 10% to 5%; 55-66 min, acetonitrile 95%, pure water 5%.

2. The method according to claim 1, characterized in that, The dispersion liquid-liquid microextraction includes vortex-assisted dispersion liquid-liquid microextraction.

3. The method according to claim 2, characterized in that, The rotation speed of the vortex-assisted liquid-liquid microextraction is 3000-4000 rpm.

4. The method according to claim 1, characterized in that, The DAD information extraction and bar chart generation includes the following steps: Step (1) Export DAD data from the high performance liquid chromatography workstation and use the maximum value filter to quickly locate the retention information of the substance peaks; Step (2) uses the airPLS method for band-by-band baseline correction, and uses a relaxed transformation method to find the peak start and end points based on the baseline and slope information; Step (3) merge the same material peaks by whether the starting point and the ending point are subsets of each other, and retain the band information with the largest peak response to form a bar chart.

5. The method according to claim 1, characterized in that, The bar graph information matching is based on the bar graph of Xiangdan injection, and each peak is matched against the bar graphs of Jiangxiang test sample and Danshen test sample. The traversal matching process includes: S1) Determine whether there is a peak in the sample to be matched that has a relative retention time of no more than ±2% with the reference peak. If so, decompose the peak interval data of the reference peak and the peak to be matched by SVD, and reconstruct the data after retaining 95% of the largest principal component. If not, it is considered that there is no corresponding matching peak for the reference peak, and the matching of the next reference peak begins. S2) The spectra of the peak elution ranges of the reference peak and the peak to be matched are superimposed and normalized to form an average spectrum; S3) If the cosine similarity between the two peaks is greater than 0.97, the peaks are saved in the successfully matched sequence. If the cosine similarity is greater than or equal to 0.97 and greater than or equal to 0.85, the spectra of each retained time point in the peak interval of the peak to be matched are extracted and compared with the average spectrum of the reference peak. If there are peak intervals in the peak interval of the peak to be matched with a similarity greater than 0.97, the peaks are considered to be successfully matched; otherwise, the peaks are considered to be unmatched. If the cosine similarity is less than 0.85, the peaks are considered to be unmatched. Finally, if there are two or more peaks in the sample to be matched that are successfully matched with the reference peak, the peak with the largest cosine similarity is taken as the matching result.

Images