Method for optimizing infrared spectrum detection parameters of ginseng leaves based on response surface method

Abstract

The invention discloses a method for optimizing infrared spectrum detection parameters of ginseng leaves based on a response surface method. A multi-level system with a signal-to-noise ratio, a peak height, a half-peak width and a peak area as evaluation indexes is constructed by taking scanning times, resolution and a ratio of a sample to KBr as variables. According to the method, a multi-index weight system is constructed through an analytic hierarchy process (AHP) for the first time, the weight values are as follows: the signal-to-noise ratio is 24.735%, the peak height is 14.022%, the half-peak width is 8.214% and the peak area is 53.029%, multi-factor collaborative optimization is converted into a mathematical problem based on quantitative scoring, and detection parameters are iteratively optimized by taking comprehensive scoring as an optimization target and combining a response surface method, so that the comprehensive performance of the detection parameters is improved. The optimal parameters of infrared spectrum detection of the ginseng leaves are determined as follows: the resolution is 4cm <-1 >, the scanning times are 32, and the mass ratio of the sample to KBr (mg) is 1: 100. According to the method, the problem of multi-factor collaborative optimization is solved by quantifying the index weight, the accuracy and sensitivity of spectrum detection are remarkably improved, and an efficient and accurate parameter optimization scheme is provided for infrared detection of traditional Chinese medicinal materials.

Description

Technical Field

[0001] The invention belongs to the field of traditional Chinese medicine detection, and particularly relates to a method for optimizing ginseng leaf infrared spectrum detection parameters based on response surface methodology. Background Art

[0002] As an important component of ginseng, ginseng leaves are rich in ginsenosides, polysaccharides and other active ingredients. Their quality inspection is of great significance to the comprehensive development and utilization of ginseng resources. Infrared spectroscopy technology is widely used in the field of Chinese medicinal material component analysis because of its advantages of rapidity, non-destructiveness and high sensitivity. By detecting ginseng leaves through infrared spectroscopy, its characteristic absorption peaks can be obtained, thereby analyzing the composition and content of the components and providing data support for quality evaluation. In addition, Han et al. (Han H, Zhang H, Wang Z, et al. Rapidly Identification of Ginseng from Different Origins Using Three-step Infrared Macro-Fingerprinting Technique[J]. Journal of Molecular Structure, 1310(2024): 138332.) explained in their infrared spectroscopy study of ginseng roots that the peak at 1630 cm -1 The amide I band nearby is a vibration absorption peak corresponding to C=O stretching vibration. 1630 cm -1 The peak shape and intensity differences around 1634cm can effectively reflect the differences in protein composition of ginseng from different origins. Song Wenjun (Song Wenjun. Research on infrared characteristic peaks of ginseng [J]. Journal of Tianjin Normal University, 2009, 29 (02): 63-65.) In the study of infrared characteristic peaks of ginseng, it is shown that 1634cm -1 The absorption peak at 1630cm may be the characteristic peak of saponin. The characteristic peaks of nitrogen-containing compounds such as proteins and amino acids in ginsenosides are one of the key indicators for infrared spectroscopy identification and component analysis. Its resolution and signal-to-noise ratio directly affect the accuracy of the detection results. However, in existing studies, the detection parameters optimization for such characteristic peaks of ginseng leaves (such as the number of scans, resolution at 1630cm -1 As a result, characteristic peaks may overlap, be distorted, or have weak signals due to improper parameter settings, making it difficult to accurately extract component information.

[0003] Currently, infrared spectroscopy for the analysis of traditional Chinese medicines (TCMs) relies heavily on empirical experience, lacking systematic optimization. Improper settings for key parameters such as scan count, resolution, and sample-to-KBr ratio can lead to low signal-to-noise ratios, peak distortion, and overlapping peaks, compromising the accuracy and sensitivity of component analysis.

[0004] In the study of using Fourier transform infrared spectroscopy to detect soil properties, Issam et al. (Issam B, Lotfi K, MS H, et al. Optimizing setup of scan number in FTIR spectroscopy using the moment distance index and PLS regression: application to soil spectroscopy [J]. Scientific Reports, 2021, 11 (1): 13358-13358.) found that the number of scans had a significant impact on spectral stability and prediction model accuracy: when the number of scans was less than 50, the standardized moment distance index fluctuated greatly and the spectral repeatability was poor, while after more than 50 times, the standardized moment distance index tended to be stable and the spectral similarity was significantly improved; the partial least squares regression model showed that with the increase in the number of scans, the correlation coefficient of soil property prediction generally increased, the cross-validation root mean square error decreased, and optimizing the number of scans could make up for the defect of small sample size. A high-precision model could still be constructed in 40 samples, confirming that optimizing the number of scans is a key step to improve the accuracy of FTIR spectral analysis.

[0005] In the study of Fourier transform infrared spectroscopy identification of high explosives, Krzysztof et al. (Banas K, Banas M A, Heussler PS, et al. Influence of spectral resolution, spectral range and signal-to-noise ratio of Fourier transform infra-red spectra on identification of high explosive substances [J]. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, 188 (2018): 106-112.) found that for high explosive samples, when the resolution was 4 cm -1 Although it can retain more spectral details, it will reduce the signal-to-noise ratio. -1 or 32cm -1 The spectral data at this time performs better in terms of cluster compactness and stability indicators. It can not only distinguish different explosives through characteristic peaks, but also improve identification reliability with a higher signal-to-noise ratio. This shows the necessity of selecting appropriate resolution parameters.

[0006] In addition, traditional methods do not fully consider the interactions between various parameters, making it difficult to achieve multi-parameter collaborative optimization, which limits the application effect of infrared spectroscopy technology in ginseng leaf detection.

[0007] Therefore, there is an urgent need for a scientific and effective method to systematically optimize the infrared spectroscopy detection parameters of ginseng leaves to improve detection accuracy and efficiency and provide a reliable technical means for ginseng leaf quality control.

[0008] Although existing related research has achieved certain results, it still has limitations. Qi Minghui et al. (Qi Minghui, Hu Xueting, Zhao Jing, et al. Optimization of near-infrared spectroscopy detection method for the content of Fritillaria cirrhosa in Fritillaria mixtures [J]. Journal of Tianjin University of Traditional Chinese Medicine, 41(06):767-773.) optimized the method of near-infrared spectroscopy detection of Fritillaria cirrhosa content in Fritillaria mixtures through orthogonal experimental design, and systematically studied the effects of four parameters, particle size, number of scans, resolution, and gain, on model performance. The results showed that the number of scans had the greatest impact on model accuracy. However, the traditional orthogonal experimental method has many limitations in the optimization of near-infrared spectroscopy detection parameters. Not only is the parameter combination consideration limited, but because the orthogonal experimental method assumes that the factors are independent of each other, it is impossible to effectively examine the comprehensive impact of the synergistic effect of multiple factors on the detection results.

[0009] In comparison, the present invention introduces the analytic hierarchy process (AHP) to show significant advantages. In terms of operational difficulty, by constructing processes such as judgment matrix and consistency test, complex multi-factor problems are converted into hierarchical and structured analysis, thereby lowering the professional threshold. In terms of cost-effectiveness, the weight of each indicator is clarified with the help of AHP, which reduces unnecessary experimental attempts and further reduces costs. In terms of parameter coverage, it can comprehensively consider indicators such as signal-to-noise ratio, peak height, half-peak width, peak area, and quantify the impact of each indicator on the results to avoid missing important information. For the interaction between factors, it can be integrated into a comprehensive score through weight distribution to truthfully reflect the synergistic effect of multiple factors. In the face of complex and changeable experimental environments, the method of the present invention performs parameter optimization based on a quantitative model, and the results have higher stability and accuracy, which can provide a reference for the optimization of infrared detection parameters of ginseng leaves and infrared detection of Chinese medicinal materials. Summary of the Invention

[0010] At present, there are significant deficiencies in parameter setting for infrared spectroscopic determination of Chinese medicinal materials: traditional methods rely on experience to set key parameters such as the number of scans, resolution, and the ratio of sample to KBr, and fail to construct a multi-index quantitative evaluation system. They ignore the differential importance of indicators such as signal-to-noise ratio, peak height, half-peak width, and peak area, as well as the interaction between parameters: the synergistic effect of the number of scans and resolution, and the antagonistic effect of resolution and sample ratio, which leads to single-factor optimization easily falling into local optimality, and parameter combinations often cause problems such as low signal-to-noise ratio, peak broadening, and overlapping characteristic peaks. It is difficult to accurately reflect the complex component information of Chinese medicinal materials such as ginseng leaves, limiting the detection accuracy and sensitivity. The purpose of the present invention is to disclose a method for optimizing infrared spectroscopic detection parameters of ginseng leaves based on the response surface method, aiming to construct a multi-index weight system through the hierarchical analysis method (AHP), to solve the core problem of parameter setting relying on experience and lack of quantitative basis for multi-factor collaborative optimization in traditional infrared spectroscopic detection, and to achieve systematic and precise optimization of detection parameters. The method comprises the following steps:

[0011] (1) Select dry, mold-free ginseng leaves, remove impurities and rotten parts, clean the surface stains with pure water, drain and place in a cool and ventilated place to dry naturally to constant weight (moisture content ≤ 8%). Use an ultra-fine grinder to grind the dried ginseng leaves, pass through a 150-mesh sieve (particle size ≤ 100 μm), and store the resulting powder in a sealed desiccator for future use to prevent moisture absorption.

[0012] (2) The number of scans (A), resolution (B), and the mass ratio of sample to KBr mixture (C) were determined as key variables, and the preliminary optimization range of each parameter was determined through single-factor experiments: Number of scans: Test the effect of 8 to 64 scans on the signal-to-noise ratio, and determine that the spectral signal-to-noise ratio is the highest when 32 scans are performed; Resolution: Test 2 to 16 cm -1 Separation of characteristic peaks within the range, 4cm -1 The peak shape is sharp and the baseline is stable; the ratio of sample to KBr: compared with the mixing ratio of 1:50 to 1:150 (mg:mg), the tablet transmittance is best at 1:100, the characteristic peak intensity is high and there is no baseline drift.

[0013] (3) The Box-Behnken response surface method was used to design a three-factor, four-level experiment, with the number of scans (A), resolution (B), and the mass ratio of sample to KBr mixture (C) as independent variables. At the same time, a multi-level evaluation index system including signal-to-noise ratio, peak height, half-peak width, and peak area was established, and the weight analysis of each evaluation index was performed using the analytic hierarchy process. The specific process was as follows: a judgment matrix was constructed, and after rigorous steps such as consistency testing, the weight values ​​of the four evaluation indicators of signal-to-noise ratio, peak height, half-peak width, and peak area were finally determined to be 24.735%, 14.022%, 8.214%, and 53.029%, respectively. The above weight values ​​were substituted into the evaluation index system, and the comprehensive score corresponding to each variable parameter combination was calculated. This was used as the response value to establish a comprehensive evaluation model of hierarchical analysis compound empowerment, providing a quantitative basis for subsequent parameter optimization.

[0014] (4) Based on the response surface experimental data, a quadratic multiple regression equation model was established: Y = 79.15 + 1.07*A-1.59*B+7.12*C+1.08*A*B-0.51*A*C-1.24*B*C-8.38*A 2 -4.10*B 2 -12.13*C 2 In the formula, Y is the comprehensive score, A is the number of scans, B is the resolution, and C is the ratio of sample to KBr. This equation was used to analyze and solve the model. Through mathematical optimization algorithms, the parameter combination that maximizes the comprehensive score Y was found. The optimal parameters for infrared detection of ginseng leaves were determined to be: 32 scans, a resolution of 4 cm -1 , the mixing ratio of sample and KBr is 1:100.

[0015] Beneficial progress

[0016] This study, for the first time, incorporates the Analytic Hierarchy Process (AHP) into the optimization of ginseng leaf infrared spectroscopy parameters. Through a structured weighting process involving the construction of a judgment matrix and consistency testing, the weights for each indicator were ultimately determined to be 24.735%, 14.022%, 8.214%, and 53.029%, respectively. This process transforms subjective experience into objective weighted data, avoiding the blind practice of setting parameters solely based on empirical experience in traditional methods. This allows the comprehensive evaluation score to truly reflect the detection performance under the synergistic effect of multiple indicators.

[0017] 2. By using the parameter combination optimized collaboratively by the analytic hierarchy process and the response surface methodology, the signal-to-noise ratio of the characteristic peak of the infrared spectrum of ginseng leaves was increased by 26%, and the half-peak width was reduced by 9.8% compared with the infrared spectrum under inefficient parameters, which intuitively demonstrated the effectiveness of the present invention in solving the problem of multi-factor collaborative optimization.

[0018] 3. By quantifying the weights of indicators, the challenge of multi-factor collaborative optimization was effectively addressed, fully accounting for the interactions between various parameters: The number of scans and resolution exhibited a weak synergistic effect. High resolution required sufficient scans to suppress noise and avoid signal distortion, and a simultaneous improvement in both slightly improved the overall score. Resolution and sample ratio exhibited an antagonistic effect. High resolution required a low sample ratio to ensure transparency and uniformity of the compressed tablets; otherwise, particle agglomeration could lead to enhanced light scattering and peak broadening. The interaction between the number of scans and sample ratio was weak, but a higher number of scans was required to accumulate signal at a low sample ratio. However, a high sample ratio might result in a limited improvement in the signal-to-noise ratio due to increased scans due to light scattering. The interactions between the various parameters were quantified into a comprehensive scoring model using the analytic hierarchy process. Combined with the quadratic multivariate regression equation constructed using the response surface methodology, the complex synergistic or antagonistic relationships between the parameters were converted into computable mathematical expressions. This prioritized the weights of core indicators such as peak area while taking into account secondary indicators such as signal-to-noise ratio and peak shape, achieving a dynamic balance between the number of scans, resolution, and sample ratio, significantly improving the overall performance of the detection parameters.

[0019] 4. Compared with the traditional method of setting parameters based on experience, this method is based on scientific experimental design and data analysis, and has stronger systematicity and repeatability: the parameter optimization range is determined by single-factor experiments, and then the response surface method is used for multi-factor collaborative optimization. The hierarchical analysis method is combined to construct a multi-index weight system, and indicators such as signal-to-noise ratio, peak height, half-peak width, and peak area are converted into quantitative comprehensive scores to avoid the limitations of a single indicator; the experimental process is strictly standardized, including sample pretreatment (cleaning, air drying, crushing and screening), data processing (baseline correction, smoothing) and indicator calculation method to reduce human errors; a quadratic multiple regression model (R) is established based on 17 sets of experimental data designed by Box-Behnken. 2 =0.9667), accurately predicting the comprehensive performance of parameter combinations, and three replicated validation experiments (CV = 2.69%) demonstrated stable and reliable results. This method breaks through the traditional "trial and error" model and establishes a scientific paradigm of "data-driven, model optimization, and standard replication," providing a replicable parameter optimization solution for infrared detection of ginseng leaves and other Chinese medicinal materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Response surface three-dimensional plot of the effects of scan number and resolution on comprehensive score

[0021] Figure 2 Response surface three-dimensional plot of the effects of scan number and sample-KBr mixing ratio on comprehensive score

[0022] Figure 3 Response surface three-dimensional plot of the effects of resolution and sample-KBr mixing ratio on comprehensive score

[0023] Figure 4Infrared spectra of the optimal parameters and the optimal parameters among the three single factors: single factor 1 is the number of scans; single factor 2 is the resolution; single factor 3 is the mass ratio of sample to KBr

[0024] Figure 5 Infrared spectra of ginseng leaves under optimal parameters and comparison file parameters DETAILED DESCRIPTION

[0025] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings and embodiments:

[0026] Example 1 Sample pretreatment

[0027] Select dry, mold-free ginseng leaves, remove impurities and rotten parts, rinse with pure water, drain, and air dry in a cool, well-ventilated area until constant weight is reached, with a moisture content of ≤8%. Grind the shade-dried ginseng leaves using an ultrafine grinder, pass through a 150-mesh sieve, and reduce the particle size to ≤100 μm. Store the resulting powder in a sealed desiccator until ready for use to prevent moisture absorption.

[0028] Example 2 Evaluation method

[0029] Characteristic peak selection: 1630~1660cm -1 The characteristic peak at is related to the vibration of the C=O double bond or amide I band in ginsenoside, the core active ingredient in ginseng leaves. It is one of the hallmark features of the infrared spectrum of ginseng leaves and can be used for qualitative analysis of ingredients and content evaluation. The characteristic peak here is selected to calculate half-peak width, peak height, peak area, signal-to-noise ratio and other information.

[0030] Calculation of half-peak width, peak height, and peak area: After converting transmittance and absorbance, performing baseline correction, and 13-point smoothing on the infrared spectrum using OMNIC 9 software, a one-dimensional infrared spectrum was obtained. -1 The characteristic peak at the position can be directly obtained, and the data information of half-peak width, peak height and peak area can be directly obtained.

[0031] Signal-to-noise ratio calculation: According to the formula SNR=h / σ noise Calculate the signal-to-noise ratio. Where SNR is the signal-to-noise ratio, h is the peak height of the characteristic peak, σ noise is the standard deviation of baseline noise. Select 1630~1660cm -1 The peak height of the characteristic peak is h, 3900~3600cm -1 There is no characteristic absorption peak in the range, which is selected as the baseline noise standard deviation data σ noise .

[0032] Example 3 Single factor experiment optimization

[0033] Infrared spectroscopy was performed using the number of scans, resolution, and the sample / KBr mass ratio as independent variables. Three replicates were performed under the same conditions, and the spectrum was averaged. The resulting infrared spectrum was plotted with wavelength as the abscissa and transmittance as the ordinate. Transmittance-absorbance conversion, baseline correction, and 13-point smoothing were performed using OMNIC 9 to generate a one-dimensional infrared spectrum.

[0034] After establishing a multi-level evaluation index system encompassing signal-to-noise ratio (SNR), peak height, half-width (FWHM), and peak area, the analytic hierarchy process (AHP) was first introduced to quantify the subjective weights of these multiple indicators. By constructing a pairwise comparison judgment matrix for the indicator layer, as shown in Table 1, and after a consistency test (CR = 0.078 < 0.1), the weights for each indicator were determined to be 24.735% for SNR, 14.022% for peak height, 8.214% for FWHM, and 53.029% for peak area. This prioritized peak area > SNR > peak height > FWHM, providing a scientific and quantitative basis for comprehensive scoring.

[0035] Table 1 Comparison and judgment matrix of indicator layers

[0036]

[0037] The weight values ​​determined by the hierarchical analysis method were substituted into the evaluation index system, and the comprehensive score of each parameter combination was calculated using the comprehensive score formula Y = (signal-to-noise ratio / maximum signal-to-noise ratio*0.25-half-peak width / maximum half-peak width*0.08+peak height / maximum peak height*0.14+peak area / maximum peak area*0.53)*100%. The multi-index optimization was converted into a single quantitative target, which significantly improved the efficiency and accuracy of parameter optimization.

[0038] The comprehensive scores of the single-factor experiment are shown in Tables 2 to 4. The optimal parameters screened out are 32 scans and a resolution of 4 cm. -1 The mass ratio of sample to KBr is 1:100.

[0039] Table 2 Single factor - comprehensive score of number of scans

[0040]

[0041] Table 3 Comprehensive score of single factor - resolution

[0042]

[0043] Table 4 Single factor - comprehensive score of signal-to-noise ratio

[0044]

[0045] Example 4 Response surface methodology optimization and verification

[0046] Based on the optimal parameters of the single-factor experiment, a three-factor three-level response surface experiment was established. The factor level table is shown in Table 5. Box-Behnken experimental design:

[0047] Table 5 Box-Behnken design factors and levels

[0048]

[0049] With the number of scans (A), resolution (B), and the mass mixing ratio of sample to KBr (C) as independent variables, a three-factor three-level experiment was designed, with a total of 17 groups of experiments, and the response value was the comprehensive score of Example 3.

[0050] The comprehensive scoring results of the design combinations and their corresponding experiments are shown in Table 6:

[0051] Table 6 Response surface design and experimental results

[0052]

[0053] Design Expert 12.0 software was used to process the data by binomial equation fitting and multiple linear regression. 2 ) and confidence level (P) were used as the model judgment criteria, and variance analysis was performed on the regression parameters of the response surface. The quadratic fitting equation was:

[0054] R 2 =79.15+1.07*A-1.59*B+7.12*C+1.08*A*B-0.51*A*C-1.24*B*C-8.38*A 2 -4.10*B 2 -12.13*C 2 .

[0055] The variance analysis is shown in Table 7. According to Table 7, the P value of the regression equation model is significant, indicating that the model has a good fit and the experimental method is feasible; the P value of the lack of fit term is not significant, indicating that the model error is small; the larger the F value of the factor and the smaller the P value, the greater the influence of the factor on the result. Therefore, the influence of each factor on the infrared spectrum is in the following order:

[0056] Table 7 Variance results of response surface experiment

[0057]

[0058]

[0059] Variance analysis showed that the F value of the sample to KBr mass ratio (C) was the largest, at 122.89, which was consistent with the conclusion in the hierarchical analysis method that "the peak area weight was the highest, at 53.029%", confirming the rationality of the weight distribution - the sample ratio directly affects the transmittance and peak area of ​​the tablet, thus playing a decisive role in the comprehensive score.

[0060] The optimal parameters obtained by response surface software calculation and optimization are: 32 scans, 4 cm resolution -1 The sample-to-KBr mixing ratio was 1:100. Three validation experiments yielded a signal-to-noise ratio of 98.196±0.370, a half-peak width of 78.11±0.483, a peak height of 25.09±0.417, and a peak area of ​​2867.41±0.531, all close to the theoretical values, confirming the accuracy and reliability of the experiment.

[0061] Comparison and analysis of infrared spectra after optimizing parameters in Example 5

[0062] The infrared spectra under three single-factor optimal parameters were selected as the control group: single factor 1 was 32 scans, and the rest of the data were initial values; single factor 2 was 4 cm resolution. -1 , the rest of the data are initial values; single factor 3 is the mass ratio of sample to KBr is 1:100, the rest of the data are initial values. The infrared spectrum under the optimal parameter combination is the experimental group, the experimental group parameters are 32 scan times, resolution 4cm -1 , the mass ratio of sample to KBr is 1:100.

[0063] based on Figure 4 , we can see that the control group with only optimized scanning times, single factor 1, has a high -1 In the characteristic peak region, the signal-to-noise ratio improved, but the peak shape was broad. Simply increasing the number of scans could increase signal acquisition, but the peak separation was poor due to insufficient resolution. The improper mass ratio of sample to KBr mixture reduced the transparency of the pressed tablet, limiting the intensity of the characteristic peak.

[0064] The control group, which optimized only the resolution, showed a sharp peak shape for single factor 2, but a low signal-to-noise ratio and a small peak area. High resolution improved peak separation, but insufficient scan times led to noise interference.

[0065] The control group, which optimized only the sample-to-KBr ratio, showed higher peak intensity for single factor 3, but a moderate signal-to-noise ratio and slightly broad peak shape. A 1:100 sample-to-KBr ratio optimized the transparency of the pressed tablet, but the scan number and resolution were not optimized in tandem, resulting in insufficient signal acquisition and an increased risk of peak overlap.

[0066] The characteristic peak of the experimental group with the optimal parameter combination was sharp, with a signal-to-noise ratio of up to 98.196±0.370, and the half-peak width was reduced to 78.11±0.483cm -1 The peak area reached 2867.41±0.531. Multi-parameter collaborative optimization eliminated the limitations of single-factor optimization. The matching of scan times and resolution improved signal quality and peak separation. The reasonable sample ratio ensured light transmittance and characteristic peak intensity.

[0067] Table 8 Quantitative comparison of key indicators

[0068]

[0069] Note: ** Indicates p<0.01 (very significant), *** Indicates p<0.001 (extremely significant), **** Indicates p<0.0001 (extremely high significance). The differences between the optimal parameter combination and each single factor group reached a significant level.

[0070] The interaction between the number of scans and the resolution, e.g. Figure 1 The three-dimensional response surface diagram shows that high resolution requires sufficient scanning times to avoid noise interference; the interaction between sample ratio and resolution, such as Figure 3 As shown, a low ratio of 1:100 significantly increases peak area at high resolution, confirming the rationality of the "peak area weight is the highest, at 53.029%." Traditional single-factor optimization focuses solely on a single metric, which can easily lead to the degradation of other metrics. For example, single factor 1 ignores resolution, and single factor 3 neglects scan count. This invention, however, integrates multiple metric weights through the Analytic Hierarchy Process (AHP) and combines it with the Response Surface Methodology (RSM) to achieve a dynamic balance between parameters, ultimately maximizing the overall score.

[0071] Example 6 Comparative Experiment

[0072] Scanning was performed according to the optimal parameters of Example 4: the number of scans was 32 times, and the resolution was 4 cm -1 The mass ratio of sample to KBr mixture is 1:100.

[0073] Document 1: Scanning times: 16 times, resolution: 4cm -1 The mass ratio of the sample to KBr mixture was 1:150. (Li Yun, Xu Furong, Zhang Jinyu, et al. Study on identification of origin of Panax notoginseng and prediction of saponin content by FTIR combined with chemometrics [J]. Spectroscopy and Spectral Analysis, 2017, 37(08): 2418-2423.); Reference 2: The number of scans was 16, and the resolution was 4 cm -1The mass ratio of the sample to KBr mixture is 3:200. (Fan Shuaishuai, Gao Han, Tian Wei. Rapid identification of three kinds of granules of ginseng, red ginseng and American ginseng by Fourier transform infrared spectroscopy [J]. Drug Evaluation Research, 2018, 41(12): 2242-2247.)

[0074] The signal-to-noise ratio, half-peak width, peak area, and peak height of the infrared spectra under the three parameter combinations were calculated, and it was found that the score for scanning according to the optimal parameters of Example 4 was the highest. Its advantage is that the appropriate number of scans can improve the signal-to-noise ratio, and the appropriate mass ratio of the sample to KBr makes the characteristic peaks of the spectrum more clearly resolved.

[0075] Table 9 Quantitative comparison of key indicators

[0076]

[0077] Note: **** The difference between the optimal parameter combination of the present invention and the reference group is statistically significant.

[0078] The resolution of the three sets of parameters is 4cm -1 However, when the ratio of the sample to KBr is 1:100, the light transmittance of the tablet is significantly improved. When the ratio is 1:150 in Document 1, the sample concentration is insufficient due to the low sample ratio, resulting in weak characteristic peak intensity and small peak area. When the ratio is 3:200 in Document 2, the sample ratio is too high, which easily causes particle agglomeration, resulting in enhanced light scattering and widening of the half-peak width. For example, the half-peak width in Document 2 is close to that of the present invention, but the peak area is smaller.

[0079] The present invention achieves a sharper peak shape through the collaborative optimization of "number of scans + sample ratio" and, at the same resolution, a half-peak width close to that of Reference 1, but an increase of 33.8% in peak area.

[0080] The comprehensive score of this invention is higher than that of Document 1 and Document 2. The core reasons are:

[0081] The analytic hierarchy process assigned the peak area the highest weight, which was 53.029%. The peak area of ​​the present invention was significantly higher than that of the control group, which significantly improved the comprehensive score. The multi-factor collaborative optimization avoided the limitations of single parameter adjustment.

[0082] In summary, the present invention discloses a method for optimizing the infrared spectrum detection parameters of ginseng leaves based on the response surface methodology. The core of the method is to introduce the analytic hierarchy process (AHP) for the first time to construct a multi-index weight system: by constructing a judgment matrix and performing a consistency test, the weights of the signal-to-noise ratio, peak height, half-peak width, and peak area are determined to be 24.735%, 14.022%, 8.214%, and 53.029%, respectively, and the subjective experience of core indicators such as peak area is converted into an objective quantitative basis. The comprehensive score calculated with this weight system is used as the optimization goal, and the optimal parameter combination is determined in combination with the response surface methodology: resolution 4cm -1 , 32 scans, and a sample-to-KBr ratio of 1:100. Experimental verification showed that the optimized characteristic peak signal-to-noise ratio reached 98.196±0.370, and the half-peak width was reduced by 9.8%, significantly improving detection accuracy and sensitivity. This method breaks through the limitations of traditional methods that rely on experience and ignore the synergistic effects of multiple factors, and provides a systematic "weighted quantification-model-driven" solution for infrared detection of traditional Chinese medicinal materials.

Claims

1. A method for optimizing infrared spectroscopy detection parameters of ginseng leaves based on response surface methodology, characterized in that: It includes the pre-processing steps of ginseng leaf samples, variable determination, indicator construction, weight calculation, score evaluation and parameter optimization steps; The pretreatment steps include washing, air drying, and sieving; the variable determination includes selecting the number of scans, resolution, and the ratio of sample to KBr as infrared spectrum detection variable parameters; the indicator construction includes establishing a multi-level evaluation indicator system including signal-to-noise ratio, peak height, half-peak width, and peak area; the weight calculation includes obtaining the combined weights of each evaluation indicator using the hierarchical analysis method; and the score evaluation includes substituting the combined weights into the evaluation indicator system to calculate the evaluation score of each variable parameter combination; The parameter optimization is based on response surface experiments, with the evaluation score as the optimization target.

2. The method according to claim 1, characterized in that Select ginseng leaves, remove rotten leaves, wash them, drain the water, dry them in the shade to constant weight, grind them in a grinder, and pass them through a 150-mesh sieve.

3. The method according to claim 1, characterized in that The single-factor experiment was used to optimize the infrared detection parameter variables of ginseng leaves, with the scanning times of 32 times and the resolution of 4 cm. -1 The mg mass ratio of sample to KBr is 1:

100.

4. The method according to claim 1, characterized in that The hierarchical analysis method was used to perform weight analysis on the four evaluation indicators of signal-to-noise ratio, peak height, half-peak width, and peak area. By constructing a judgment matrix and undergoing consistency tests and other steps, the weight values ​​of each indicator were finally determined to be 24.735%, 14.022%, 8.214%, and 53.029%, respectively; and the above weight values ​​were substituted into the evaluation index system to calculate the comprehensive score corresponding to each variable parameter combination.

5. The method according to claim 1, characterized in that In the response surface experimental design of ginseng leaf infrared detection parameters, a three-factor three-level experimental scheme is adopted; the detection conditions after single factor optimization in claim 3 are taken as the central value, and the comprehensive score calculated by determining the weight through the hierarchical analysis method in claim 4 is taken as the response value.

6. The method according to claim 1, characterized in that The quadratic multiple regression equation model established in the response surface is: R 2 =79.15+1.07*A-1.59*B+7.12*C+1.08*A*B-0.51*A*C-1.24*B*C-8.38*A 2 -4.10*B 2 -12.13*C 2 Where Y is the comprehensive score, A is the number of scans, B is the resolution, and C is the mass ratio of the sample to KBr.

7. The method according to claim 1, characterized in that The optimal parameters for infrared detection of ginseng leaves optimized in the response surface experiment were: 32 scans and a resolution of 4 cm. -1 The mixing mass ratio of sample to KBr was 1:100, and the results were consistent with those of single factor.

8. The method according to claim 1, characterized in that The analytic hierarchy process includes: constructing a pairwise comparison judgment matrix of the indicator layer (as shown in Table 1), determining the weight of each evaluation indicator through consistency test (CR < 0.1), wherein the peak area weight accounts for > 50%, providing a differentiated quantitative basis for parameter optimization.

Images