A method for screening dried chilli peppers based on raw material indexes to predict the quality of chilli oil

By detecting key indicators of dried chili peppers and establishing a partial least squares regression model, the problem of the disconnect between dried chili pepper raw materials and chili oil quality prediction was solved, enabling accurate screening and quality prediction in the raw material stage, and reducing the trial-and-error costs and time costs for enterprises.

CN123065992APending Publication Date: 2026-11-10XIHUA UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611344666.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-01
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing technologies cannot accurately predict the quality of dried chili peppers after they are made into chili oil during the raw material procurement process, resulting in large fluctuations in product quality, high trial and error costs for enterprises, and strong subjectivity and poor reproducibility of sensory evaluation.

Method used

By testing key indicators of dried chili peppers, such as hexanol content, hexanol content, ethyl acetate content, moisture content, b* value, capsaicin content, and capsanthin content, a partial least squares regression model was established to predict the overall quality score of chili oil, and raw materials were selected according to the scores.

Benefits of technology

This approach enables the prediction of chili oil quality during the raw material testing stage, reducing trial-and-error costs and time costs. It also establishes objective and quantifiable evaluation standards, overcoming the subjectivity and poor reproducibility of sensory evaluation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123065992A_ABST
    Figure CN123065992A_ABST
Patent Text Reader

Abstract

The application discloses a kind of based on raw material index forecast chilli oil quality dry chilli screening method, belong to food analysis detection technical field.This method detects the n-hexanal, n-hexanol, ethyl acetate, moisture, b * Value, capsaicin and capsanthin content of 7 key indicators of dry chilli to be screened, into the partial least squares regression model constructed in advance, directly output the comprehensive quality prediction score of the chilli oil prepared by this dry chilli, according to the score from high to low order and select the front as oil preparation raw material.The model takes dry chilli key indicators as independent variables, and takes chilli oil comprehensive quality score as dependent variable;The latter is linearly weighted after being weighted by entropy weight method from 16 chilli oil core indicators significantly related to sensory score.This application quantitatively connects raw material evaluation and terminal product quality, without sample preparation and sensory evaluation, can judge raw material value in procurement link, overcome the subjective, single index and one-sided defects of sensory evaluation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of food analysis and testing technology, specifically relating to a method for screening dried chilies to predict the quality of chili oil based on raw material indicators. Background Technology

[0002] Chili peppers are a core raw material in the condiment processing industry, and dried chili peppers are the main form of chili pepper processing. Chili oil is made by pouring hot oil over dried chili powder; it has a bright red color and a rich aroma, and is used extensively in Sichuan cuisine and various other flavored foods. Dried chili peppers from different origins and varieties exhibit significant natural differences in appearance, nutritional quality, and volatile flavor compounds. These differences directly lead to batch-to-batch fluctuations in the color, aroma, and taste of the final chili oil product. Therefore, accurately assessing the quality of a batch of dried chili peppers after processing into chili oil during the raw material procurement stage is crucial for chili oil manufacturers to achieve stable product quality.

[0003] CN115777901A discloses a method for preparing chili oil, in which 30 people with experience in food sensory evaluation score the prepared chili oil in terms of color, aroma, and taste according to a segmented rule, and take the average score of each group as the product quality evaluation result. The disadvantages of this prior art are: its quality evaluation can only be carried out after the chili oil is actually prepared, which is a post-evaluation. Enterprises can only adopt a rough model of trial and error, resulting in high trial and error costs; moreover, the segmented scoring of color, aroma, and taste depends entirely on the subjective judgment of the evaluators, resulting in poor reproducibility and difficulty in horizontal comparison between different evaluation groups, making it impossible to use this to proactively rank candidate dried chili raw materials in the procurement stage.

[0004] In their paper, "Comprehensive Evaluation of the Quality and Processing Suitability Analysis of Dried Chili Peppers from Xinjiang" (Food Industry Technology, 2023, Vol. 44, No. 4, pp. 317-325), Nian Guofang et al. used 12 major dried chili pepper varieties cultivated in Xinjiang as materials. Through principal component analysis and cluster analysis, they conducted a comprehensive quality evaluation of dried chili pepper resources and determined that long chili peppers are suitable for making fermented chili sauce, chili threads, and chili oil; facing-heaven chili peppers are suitable for extracting capsaicin; and flat chili peppers are suitable for extracting capsaicin. The drawback of this approach is that its processing suitability conclusions remain at the qualitative inference level of category classification, without providing a quantitative model between dried chili pepper indicators and the quality score of a specific end product. Therefore, multiple candidate raw materials within the same category cannot be ranked, and the specific quality values ​​of the end product cannot be predicted.

[0005] In summary, existing technologies generally suffer from drawbacks such as a disconnect between raw material evaluation and end-product quality, a lack of forward-looking quantitative prediction capabilities, strong subjectivity in sensory evaluation, and one-sidedness in evaluation based on single physicochemical indicators. These shortcomings lead to large fluctuations in the quality of finished chili oil products, high trial-and-error costs for enterprises, and difficulty in effectively identifying the value of high-quality raw materials. Therefore, there is an urgent need for a method that can reveal the intrinsic quantitative correlation between the indicators of dried chili raw materials and the overall quality of chili oil, and based on this, accurately screen dried chili raw materials. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method for screening dried chilies based on raw material indicators to predict the quality of chili oil.

[0007] One of the objectives of this invention is to reveal the quantitative correlation between the intrinsic quality indicators of dried chili pepper raw materials and the overall quality of chili oil, and to establish an evaluation system that runs from the raw material end to the product end, so that the evaluation results of dried chili peppers can truly reflect their final quality performance after oil extraction, thereby providing chili oil processing enterprises with a raw material selection basis with practical application value.

[0008] The second objective of this invention is to provide a quantitative model that can directly output the comprehensive quality prediction score of chili oil during the raw material procurement process. This allows companies to know the quality performance of candidate raw materials after oil production without actual sample preparation and sensory evaluation, upgrading the extensive "try before use" model to a precise "evaluate before purchase" model, thereby reducing the trial-and-error costs, time costs, and resource consumption of companies.

[0009] The third objective of this invention is to establish an objective, quantifiable, and reproducible evaluation standard for dried chili raw materials by conducting multi-dimensional systematic testing and scientific dimensionality reduction of the appearance quality, nutritional quality, moisture content, and volatile flavor substances of dried chili peppers. This overcomes the shortcomings of subjective and poor reproducibility of artificial sensory evaluation and the one-sidedness of evaluation based on a single physicochemical indicator.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for screening dried chilies to predict chili oil quality based on raw material indicators includes: S1. Detect key indicators of the dried chili peppers to be screened. Key indicators include hexanal content, hexanol content, ethyl acetate content, moisture content, and b. * Value, capsaicin content and capsanthin content, of which b * The value represents the degree of yellow-blueness; S2. Substitute the key indicators detected in S1 into the pre-constructed comprehensive quality prediction model for chili oil, and output the comprehensive quality prediction score of the chili oil prepared from the dried chilies to be screened. The comprehensive quality prediction model for chili oil is a partial least squares regression model. Its independent variable is the key indicators of the dried chili sample, and its dependent variable is the comprehensive quality score of the chili oil corresponding to the dried chili sample. The comprehensive quality score is obtained by linearly weighting the core quality indicators of chili oil after weighting by the entropy weight method. S3. Sort the dried chilies to be screened from high to low according to the comprehensive quality prediction score, and select the dried chilies with the highest scores as the raw materials for making chili oil.

[0011] Furthermore, in S2, the comprehensive quality prediction model for chili oil is constructed according to the following steps: S21. Take n portions of dried chili peppers as modeling samples, n≥20, and test 21 physicochemical indicators and volatile flavor substances in each modeling sample. S22. The dimensionality of 21 physicochemical indicators and volatile flavor substances was reduced to obtain 15 quality indicators of dried chili peppers. S23. Each modeling sample was prepared into chili oil. The physicochemical indicators and volatile flavor substances of the chili oil were tested and sensory evaluation was conducted to determine 16 core quality indicators of chili oil that were significantly related to the sensory score. After weighting by entropy weight method, the comprehensive quality score of each chili oil was obtained by linear weighting. S24. Seven key indicators were selected from 15 dried chili quality indicators, namely ethyl acetate content, capsanthin content, and b... * The values ​​of water content, n-hexanol content, capsaicin content, and n-hexanaldehyde content were used as the independent variables. The comprehensive quality score obtained from S23 was used as the dependent variable. The optimal number of principal components was determined by leave-one-out cross-validation, and a partial least squares regression model was established.

[0012] Furthermore, in S22, dimensionality reduction includes two parts: dimensionality reduction of physicochemical indicators and dimensionality reduction of flavor substances. The dimensionality reduction of the physicochemical indicators is as follows: Pearson correlation analysis is performed on 21 physicochemical indicators. The strong correlation is determined by the absolute value of the correlation coefficient r ≥ 0.7 and p < 0.05. Only one representative indicator is retained in each group of strongly correlated indicators. Then, indicators with a coefficient of variation < 10% among n modeling samples are removed, resulting in 7 physicochemical indicators. Flavor compounds were reduced to the following: For key aroma compounds with a relative odor activity value >1, compounds with a detection rate <70% were first removed. Then, for samples where the compound was not detected, the missing value was supplemented with half of the minimum detection value of the compound in all samples. Pearson correlation analysis was then performed to remove compounds that were significantly correlated with other compounds and had low relative odor activity values, resulting in 8 key flavor compounds.

[0013] Furthermore, in S24, the method for selecting the 7 key indicators is as follows: principal component analysis is performed on 15 dried chili quality indicators, and principal components are extracted according to the principle of eigenvalue > 1; for the first principal component, indicators with an absolute loading value ≥ 0.79 are selected; for each principal component other than the first principal component, indicators with the largest absolute loading value and an absolute loading value ≥ 0.89 are selected; the above indicators are combined to obtain 7 key indicators.

[0014] Furthermore, in S23, the 16 core quality indicators for chili oil are: a * value, b * Value, capsaicin content, peroxide value, acid value, capsaicin content, 2-n-pentylfuran content, 2,3-dimethylpyrazine content, pinene content, 2,3-pentanedione content, 1-penten-3-one content, linalool content, 2,3-butanediol content, n-pentanol content, phenylacetaldehyde content, and (E)-2-heptenal content, of which a * The value represents the degree of redness and greenness.

[0015] Furthermore, in S23, the overall quality score is calculated according to the following steps: The first step is to divide the 16 core quality indicators into positive indicators, negative indicators, and moderate indicators. Among them, acid value, peroxide value, 1-penten-3-one, n-pentanol, and (E)-2-heptenal content are negative indicators, capsaicin, b* value, and phenylacetaldehyde content are moderate indicators, and the rest are positive indicators. The second step is to process the positive indicators by y ij =(x ij -x jmin ) / (x jmax -x jmin ), for negative indicators, by y ij =(x jmax -x ij ) / (x jmax -x jmin Standardize the range, and adjust the appropriate index according to y. ij =1 / (1+ABS(x ij - / 5)) standardize to obtain the standardized value y ij In the formula, i is the sample number, j is the index number, and x ij Let x be the measured value of the j-th indicator in the i-th chili oil sample. jmax With x jmin Let x be the maximum and minimum values ​​of the j-th indicator in all samples. kj The measured value of the j-th indicator of the k-th chili oil sample, which is among the top five in sensory evaluation; Third step, press Calculate the feature weight p of the i-th sample under the j-th indicator.ij And stipulate that when p ij =0 when p ij ×ln(p ij )=0; Press Calculate the information entropy e of the j-th indicator. j , where n is the number of samples; according to d j =1-e j Calculate the information utility value d j Press w j =d j / Σd j Calculate weight w j ; Step 4, press F i =Σ(w j ×y ij Linear weighting is used to obtain the overall quality score F of the i-th chili oil. i .

[0016] Furthermore, the partial least squares regression model has a principal component count of 1, and its regression equation is: Y = 0.4343 + 1.3398 × 10 -4 ×A-1.2367×10 -4 ×B+8.0504×10 -5 ×C-2.1023×D-1.8198×10 -3 ×E-0.0284×F+2.4485×10 -3 ×G; In the formula, Y is the predicted score for the overall quality of chili oil, dimensionless; A, B, and C are the relative contents of hexanal, hexanol, and ethyl acetate in dried chili peppers, respectively, all in μg / kg; D is the moisture content of dried chili peppers, expressed as a mass fraction and rounded to the decimal, dimensionless; E is the moisture content of dried chili peppers... * Values ​​are dimensionless; F represents the capsaicin content of dried chili peppers, in mg / g; G represents the capsanthin content of dried chili peppers, in mg / g.

[0017] Furthermore, in S23, the method for preparing chili oil is as follows: after removing the stem from the modeling sample, crush it into dried chili powder. Weigh the dried chili powder and rapeseed oil separately according to the mass ratio of dried chili powder to rapeseed oil of 24:100. Heat the rapeseed oil to 210°C and then turn off the heat. When the rapeseed oil cools to 180°C, immediately pour it into a container containing dried chili powder. Stir for 1 minute and then let it cool naturally. Fill the container and extract for 24 hours. Take the supernatant oil as chili oil.

[0018] The present invention has the following beneficial effects: 1. This invention establishes a quantitative prediction model that directly points to the comprehensive quality score of chili oil from the raw material indicators of dried chili peppers. This shifts the focus of raw material evaluation from the raw material itself to the end product, overcoming the shortcomings of the prior art where the evaluation object and the evaluated object are the same and the evaluation results stop at the raw material end. This invention enables forward prediction of the quality of oil made from unknown raw materials.

[0019] 2. The prediction of this invention can be completed in the raw material testing stage, without the need for actual preparation of chili oil or sensory evaluation. This overcomes the shortcomings of the existing technology, which requires sample preparation before evaluation and can only adopt a trial-before-use model. Enterprises can directly rank multiple candidate batches and select the best ones based on the prediction scores, which greatly reduces the cost of trial and error and time.

[0020] 3. The model input of this invention is the intrinsic quality index of the dried chili raw material itself rather than the production process parameters, and the output is a comprehensive quality score composed of three sensory attributes: appearance, aroma, taste and aftertaste, rather than a single qualified judgment. This overcomes the shortcomings of the existing technology, which cannot be used for raw material selection and has a single evaluation dimension.

[0021] 4. This invention uses Pearson correlation analysis, coefficient of variation elimination, and principal component analysis to progressively reduce dimensionality, simplifying 21 physicochemical indicators and 41 key aroma compounds into 7 key indicators. This results in fewer testing items, lower testing costs, and easier implementation in enterprise quality control processes. At the same time, the entropy weight method objectively assigns weights based on the dispersion of the data itself, overcoming the shortcomings of subjective human experience-based weighting and sensory scoring. The evaluation results are objective, quantifiable, and reproducible.

[0022] 5. This invention conducted a parallel comparison of models built using three index selection methods: Pearson correlation method, principal component analysis method, and partial least squares regression variable projection importance method. The comparison confirmed that the model built using the index set selected by principal component analysis has the best fitting and predictive ability, with a cumulative coefficient of determination R0. 2 The cumulative cross-validation determination coefficient Q reached 0.806. 2 It reached 0.731. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the construction of the comprehensive quality prediction model for chili oil in this invention. Figure 2 This is a flowchart of the dried chili pepper screening method of the present invention; Figure 3 This is a correlation analysis diagram showing the relationships between various physicochemical indicators of dried chili peppers according to the present invention; Figure 4 This is a correlation analysis diagram showing the relationships between key flavor compounds in dried chili peppers according to the present invention. Figure 5 This is a correlation analysis graph showing the relationship between the sensory score of chili oil aroma and volatile flavor compounds in this invention. Figure 6 This is a correlation analysis chart showing the relationship between various physicochemical indicators and sensory scores of the chili oil of this invention. Figure 7 This is a bar chart showing the variable projection importance values ​​of the dried chili pepper index in this invention. Detailed Implementation

[0024] The technical solution of the present invention will be further described clearly and completely below with reference to the accompanying drawings and embodiments. The described embodiments are only some embodiments of the present invention, and not all embodiments. Other embodiments obtained by those skilled in the art without creative effort are all within the protection scope of the present invention.

[0025] All dried chili peppers used in this invention must meet the relevant national standards and be manually selected before use to remove moldy, insect-damaged, and foreign matter. Unless otherwise specified, all reagents used are of analytical grade, and the water used is ultrapure water.

[0026] Example 1 like Figure 1 As shown in the figure, this embodiment provides the construction process of the comprehensive quality prediction model for chili oil, as follows: 1.1 Selection of Modeling Samples Twenty dried chili pepper samples were selected as modeling samples, covering mainstream cultivated varieties from major chili pepper producing areas in my country, including Guizhou, Yunnan, Sichuan, Henan, Xinjiang, and Hainan. These included varieties such as Erjingtiao, Mantianxing, Lantern Pepper, New Generation, Bullet Pepper, Small Millet Pepper, and Wrinkled Pepper, to ensure sufficient varietal diversity and regional representativeness. The numbers and names of the 20 modeling samples are shown in Table 1.

[0027] Table 1. 20 dried chili pepper samples used for modeling (S1-S20) and 5 validation samples (V1-V5) 1.2 Detection of initial quality indicators of dried chili peppers An initial quality index library was constructed for 20 modeling samples from four dimensions: appearance quality, nutritional quality, moisture content, and volatile flavor compounds. The library included 21 physicochemical indicators, namely: single fruit length, single fruit weight, pulp content, seed coat ratio, and L. * Value (representing brightness), a * Value (representing red-green hue), b * The following parameters are considered: color value (indicating yellow-blue tint), color value, carotenoid content, capsanthin content, browning degree, total phenolic content, vitamin C content, reducing sugar content, protein content, fat content, capsaicin content, dihydrocapsaicin content, Scoville index, spiciness, and moisture content. The detection methods for each indicator are as follows.

[0028] (1) Appearance quality indicators The length of a single chili pepper was measured using vernier calipers. For each sample, 20 uniformly shaped dried chili peppers were selected, and the length from the stem to the tip was measured. The average value was taken, and the unit was mm. The weight of a single chili pepper was measured using an electronic analytical balance. For each sample, 20 uniformly shaped dried chili peppers were selected, and the total weight was divided by the number of peppers to obtain the weight of a single chili pepper, and the unit was g. The method for determining the pulp content and seed-peel ratio was as follows: For each sample, 20 uniformly shaped dried chili peppers were selected, and the total weight was measured as m1. The peel and seeds were then manually separated. After removing the seeds, the total peel weight was measured as m2, and the total seed weight was m1 - m2. The pulp content was calculated as the percentage of peel weight to the total fruit weight, i.e., pulp content = m2 / m1 × 100%, and the unit was %; the seed-peel ratio was calculated as the ratio of seed weight to peel weight, i.e., seed-peel ratio = (m1 - m2) / m2, dimensionless. As can be seen from the above definition, there is a deterministic conversion relationship between pulp content and seed coat ratio: pulp content = 100 / (1 + seed coat ratio). The information of the two is highly overlapping, which is why the correlation coefficient between the two reaches -0.993 in Section 1.3 below.

[0029] L * value, a * Value and b * The determination method for the values ​​is as follows: A colorimeter is used. The colorimeter is turned on and preheated, and calibrated using the standard white and black plates provided with the instrument. The prepared chili powder sample is placed in a sample cup, and its L*, a*, and b* values ​​are measured in the dark. Each sample is measured at least three times. * The value represents brightness, a * The value represents the red-green tint, b * The values ​​represent the yellow-blue tint, and all three are dimensionless.

[0030] Method for determining color value: Accurately weigh 1 g of chili powder sample, record its mass as m, and place it in a 100 mL volumetric flask. Add acetone solvent to about 2 cm below the mark and extract in a water bath at 25℃ in the dark for 1 h. After making up to volume with acetone solvent, let it stand in the dark for another 1 h to allow the solution to separate into layers.

[0031] Transfer 0.4 mL of the supernatant to a 10 mL volumetric flask and dilute to the mark with acetone. Measure the absorbance at 460 nm (absorbance controlled within the range of 0.300–0.700). Calculate the color value using the following formula: In the formula: E represents the valence of paprika tint; A represents the absorbance of the sample solution at 460 nm; K represents the dilution factor of the supernatant; m represents the sample mass (g); and 100 represents the conversion factor.

[0032] Determination of browning degree: Accurately weigh 1.000 g of dried chili powder, add 10 mL of distilled water, and let stand for 2 h. Take the supernatant and dilute it 10 times by volume. Then take 5 mL of the diluted solution, add an equal volume of 95% ethanol solution, centrifuge at 7800 r / min for 10 min, and take the supernatant to measure its absorbance at a wavelength of 420 nm. The absorbance value is used to characterize the browning degree of the sample. Each sample is repeated 3 times, and the average value is taken.

[0033] (2) Nutritional quality indicators The capsaicin and dihydrocapsaicin contents were determined according to the high-performance liquid chromatography method in GB / T 21266—2007 "Determination of Capsaicinoids and Expression of Spiciness in Chili Peppers and Chili Products", and the results were expressed in mg / g. The Scoville index and spiciness were calculated using the same standard: the total capsaicin content was the sum of the capsaicin and dihydrocapsaicin contents; Scoville index = total capsaicin content (mg / kg) × 16.1; spiciness = Scoville index / 100. Both are dimensionless values.

[0034] The capsanthin content was determined as follows: First, accurately weigh a certain amount of capsanthin standard, prepare a series of standard solutions of varying concentrations, and measure the absorbance of each standard solution at a wavelength of 460 nm. Plot a standard curve with capsanthin concentration on the x-axis and absorbance on the y-axis. For the capsanthin content determination, weigh 0.50 g of pulverized chili powder into a 100 mL beaker, add 15 mL of a methanol-tetrahydrofuran mixed solution (1:1, v / v), and extract using ultrasound in a 40℃ water bath for 30 min. After filtration, return the filter paper and residue to the beaker, add another 15 mL of methanol-tetrahydrofuran mixed solution, and extract using ultrasound for 10 min. Repeat this step twice. Combine the extracts and bring the volume to 50 mL. Take 500 μL of the extract, dilute it with the same mixed solvent, and bring the volume to 25 mL. Measure the absorbance at a wavelength of 460 nm, and calculate the capsanthin content in the sample based on the standard curve. Each sample was measured in triplicate.

[0035] Determination of carotenoid content: Accurately weigh 0.1000g of pulverized dried chili pepper sample, add 10mL of acetone solution, and let stand for 16h under light-protected conditions to allow the pigments to fully leach out. Dilute the resulting supernatant 10 times by volume and measure the absorbance at wavelengths of 663nm, 646nm, and 470nm, respectively. The carotenoid content is calculated using the following formula: ρ 叶绿素a (mg / L) = 12.21A 663nm -2.81A 646nm ρ 叶绿素b (mg / L) = 20.13A 646nm -5.03A663nm ρ 类胡萝卜素 (mg / L)=(1000A 470nm -3.27ρ 叶绿素a -104ρ 叶绿素b ) / 229 Carotenoid content (mg / g) = ρ 类胡萝卜素 ×V×N / (m×100) In the formula, A663nm, A646nm, and A470nm are the absorbances of the supernatant at wavelengths of 663nm, 646nm, and 470nm, respectively, and ρ is the absorbance of the supernatant at wavelengths of 663nm, 646nm, and 470nm. 叶绿素a ρ 叶绿素b ρ 类胡萝卜素 The values ​​represent the mass concentrations of chlorophyll a, chlorophyll b, and carotenoids, respectively, in mg / L; V is the total volume of the extract in mL; m is the mass of the dried chili pepper sample in g; and N is the dilution factor.

[0036] The total phenol content was determined using the Folin-Ciocalteu method: 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0 mL of 100 μg / mL gallic acid standard solution were respectively pipetted into graduated test tubes. 0.5 mL of 50% Folin-Ciocalteu reagent was added and mixed well. Distilled water was added to a final volume of 8.5 mL, and the mixture was allowed to stand for 10 min. Then, 1.5 mL of 20% sodium carbonate solution was added and mixed well. After standing at room temperature for 60 min, the absorbance was measured at 760 nm. A standard curve was plotted: y = 0.9855x + 0.0553 (R²). 2 =0.9998); Accurately weigh 1.00 g of pulverized dried chili powder, add 20 mL of 80% methanol solution, ultrasonically extract for 30 min, then let stand for 30 min in the dark. Collect the supernatant and store at 4℃. Complete the determination within 24 h; accurately pipette 1 mL of the supernatant, and perform the remaining operations as with the standard curve, repeating three times and taking the average. The total phenol content is calculated according to the following formula: Total phenol content (mg / g) = Z × V × N / (Vs × M × 1000) In the formula, Z is the mass of gallic acid in μg, V is the total volume of the extract in mL, Vs is the volume of the extract taken during the determination in mL, M is the sample weight in g, and N is the dilution factor.

[0037] Determination of reducing sugar content: Take 0, 0.2, 0.4, 0.6, 0.8, 1.0, and 1.2 mL of 1 mg / mL glucose standard solution into graduated test tubes, respectively. Add distilled water to a final volume of 2 mL, add 1.5 mL of DNS reagent, and heat in a boiling water bath for 5 min (seal the tube during heating). Cool and dilute to a final volume of 10 mL. Measure the absorbance at 540 nm. Plot a standard curve; the regression equation is: y = 2.6560x − 0.1490 (R² + π / 2)².2 =0.9956); Weigh 1.000 g of dried chili powder, add a small amount of distilled water, grind evenly, transfer to a graduated test tube, then add distilled water to the 25 mL mark and extract in an 80℃ water bath for 30 min; after cooling, filter and wash the residue with distilled water (washing with 20 mL of distilled water), collect the filtrate in a 50 mL volumetric flask and dilute to volume, mix thoroughly and set aside; accurately pipette 1.0 mL of the above extract, add 1.5 mL of DNS reagent and mix well, the remaining operations are the same as the standard curve determination. Each sample is measured three times and the average value is taken; reducing sugar content mg / g = C×V×N / Vs×W, where: C is glucose content (mg); V is the total volume of dried chili extract (mL); Vs is the volume of extract taken during the determination (mL); W is the sample weight (g); N is the dilution factor.

[0038] Determination of Vitamin C content: Take 1 mg / mL L-ascorbic acid standard solution, and pipette 0.0, 0.1, 0.2, 0.4, 0.6, 0.8, and 1.0 mL respectively. Add oxalic acid-EDTA solution to make up to 5 mL. Then, add 500 μL of 3% metaphosphoric acid-acetic acid solution, 1 mL of 5% sulfuric acid solution, and 5% ammonium molybdate solution sequentially, mix well, and place in a 30℃ water bath for 15 min. After color development, dilute to 25 mL and measure the absorbance at 760 nm. Plot a standard curve y = 0.0052x - 0.0144 (R²). 2 =0.9970); Accurately weigh 1.5000g of pulverized dried chili powder into a mortar, add 5mL of oxalic acid-EDTA solution, grind evenly, transfer to a 25mL volumetric flask, and rinse the mortar several times with oxalic acid-EDTA solution to ensure complete sample transfer. After adjusting the volume, transfer to a centrifuge tube and centrifuge at 5000r / min for 10min. Collect the supernatant for later use; take 1mL of the supernatant and add oxalic acid-EDTA solution to make up to 5mL. The remaining operations are the same as the standard curve. Each sample is measured three times and the average value is taken. The vitamin C content is calculated according to the following formula: Vitamin C content (mg / 100g) = S×V×N / (Vs×M×100) In the formula, S is the mass of L-ascorbic acid in mg, V is the total volume of the extract in mL, Vs is the volume taken during the determination in mL, M is the sample weight in g, and N is the dilution factor.

[0039] The protein content was determined according to the automatic Kjeldahl nitrogen analyzer method in GB 5009.5—2025 "National Food Safety Standard - Determination of Protein in Food".

[0040] The fat content was determined according to the Soxhlet extraction method in GB 5009.6—2025, "National Food Safety Standard - Determination of Fat in Food".

[0041] (3) Moisture content Moisture content was determined according to the direct drying method in GB 5009.3—2016 "National Food Safety Standard - Determination of Moisture in Food", and expressed as a mass fraction. When substituting into the regression equation, the moisture content was taken as a decimal, for example, a mass fraction of 11.8% was represented as 0.118.

[0042] (4) Volatile flavor compounds The headspace solid-phase microextraction-gas chromatography-mass spectrometry method was used for determination. Pretreatment conditions: Accurately weigh 2.0 g of sample into a 15 mL headspace vial, add 7 mL of saturated sodium chloride solution and 2 μL of 1.013 μg / μL 2,4,5-trimethylthiazole internal standard, tighten the cap, equilibrate in a 60 °C water bath for 20 min, extract with a 50 / 30 μm divinylbenzene / carbon molecular sieve / polydimethylsiloxane extraction head for 30 min, and then desorb at the injection port for 5 min.

[0043] Gas chromatography conditions: The column was a DB-5MS capillary column (30m × 320μm × 0.25μm); the carrier gas was helium with a purity greater than 99.999%; the carrier gas flow rate was 1 mL / min; the injection port temperature was 250℃, and splitless injection was used. The temperature program was as follows: initial temperature 40℃, held for 5 min; then increased to 150℃ at 4℃ / min, held for 3 min; then increased to 180℃ at 3℃ / min, held for 2 min; finally, increased to 250℃ at 8℃ / min, held for 3 min.

[0044] Mass spectrometry conditions: Signal acquisition was performed in full scan mode, with a scan range of 50 m / z to 550 m / z; ionization was performed by electron impact ionization with an electron impact energy of 70 eV; the ion source temperature was 230 °C and the quadrupole temperature was 150 °C.

[0045] The results were qualitatively determined by searching the NIST20 spectral library. The relative content of each volatile compound was calculated using the internal standard method, with units of μg / kg. The relative odor activity values ​​of each compound were further calculated using the formula shown in equation (1). B=C / T (1) In the formula: B is the relative odor activity value, dimensionless; C is the content of the compound relative to the internal standard, in μg / kg; T is the aroma threshold of the compound, in μg / kg, and the value is derived from publicly available aroma threshold manuals and relevant literature. Compounds with a relative odor activity value >1 are identified as key aroma compounds.

[0046] A total of 137 volatile flavor compounds were detected in 20 modeling samples, including 41 key aroma compounds with a relative odor activity value >1.

[0047] 1.3 Dimensional Reduction of Quality Indicators for Dried Chili Peppers (1) Dimensionality reduction of physicochemical indicators Pearson correlation analysis was performed on 21 physicochemical indicators, with a correlation coefficient r ≥ 0.7 and p < 0.05 used as the criterion for strong correlation. The analysis results are as follows: Figure 3 As shown: Single fruit weight was significantly negatively correlated with capsaicin, dihydrocapsaicin, Scoville index, and spiciness, with correlation coefficients of -0.804, -0.813, -0.777, and -0.814, respectively; capsaicin was significantly positively correlated with dihydrocapsaicin, Scoville index, and spiciness, with correlation coefficients of 0.962, 0.893, and 0.996, respectively; single fruit length was significantly positively correlated with pulp content, with a correlation coefficient of 0.933; seed coat ratio was significantly negatively correlated with pulp content and vitamin C, with correlation coefficients of -0.993 and -0.817, respectively, and significantly positively correlated with fat, with a correlation coefficient of 0.851; pulp content was significantly negatively correlated with fat, with a correlation coefficient of -0.846; carotenoids were significantly negatively correlated with capsaicin, color value, and a... * The values ​​showed a significant positive correlation, with correlation coefficients of 0.824, 0.801, and 0.755, respectively.

[0048] Based on this, the strongly correlated indicators among the 21 physicochemical indicators can be grouped into three groups: the first group includes single fruit weight, capsaicin, dihydrocapsaicin, Scoville index, and spiciness; the second group includes single fruit length, pulp content, seed coat ratio, vitamin C, and fat; and the third group includes carotenoids, capsanthin, color value, and a. * Value. Only one representative indicator was retained in each group: capsaicin was retained in the first group, pulp content in the second group, and capsanthin in the third group. Indicators not included in the above three groups were moisture content, browning degree, and b. * Value, reducing sugar, L * The study included seven items: total phenols, total protein, and total phenols.

[0049] For the seven indicators that did not enter the strongly correlated group, their coefficients of variation among the 20 modeling samples were further examined. Indicators with coefficients of variation <10%, i.e., insufficient discrimination between samples, were removed. It was calculated that total phenols, protein, and L... * The coefficients of variation for the values ​​were 8.2%, 7.4%, and 6.1%, respectively, all <10%, and were therefore discarded; moisture content, browning degree, and b * The coefficients of variation for the values ​​and reducing sugars were 12.6%, 18.3%, 14.7% and 21.5% respectively, all ≥10%, and were retained.

[0050] In summary, after dimensionality reduction, the physicochemical indicators retained capsaicin content, pulp content, moisture content, capsanthin content, browning degree, and b... * There are 7 items in total, including the value and reducing sugar content.

[0051] (2) Dimensional reduction of volatile flavor compounds Since key flavor compounds are needed for subsequent modeling, missing values ​​in the flavor data must first be processed. Of the 41 key aroma compounds, 26 were removed due to a detection rate <70% (i.e., not detected or below the limit of quantitation in more than 30% of samples), leaving 15. For the remaining compounds, missing values ​​were supplemented in the undetected samples with half the minimum detection value of that compound in all samples to reduce the impact of missing values ​​on subsequent statistical analysis results.

[0052] The 15 key aroma compounds retained are: isovaleraldehyde, n-hexanol, (E)-2-heptenal, benzaldehyde, phenylacetaldehyde, nonanal, β-cyclocitral, n-pentanol, n-hexanol, isovalerol, 2-ethylhexanol, benzyl alcohol, linalool, acetic acid, and ethyl acetate.

[0053] Pearson correlation analysis was performed on 15 key aroma compounds, and the results are as follows: Figure 4 As shown: Benzaldehyde was significantly positively correlated with n-pentanol, with a correlation coefficient of 0.622; β-cyclocitral was significantly positively correlated with n-hexanol and 2-ethylhexanol, with correlation coefficients of 0.651 and 0.619, respectively; n-hexanol was significantly positively correlated with nonanol (p<0.05); n-hexanol was significantly positively correlated with (E)-2-heptenal, with a correlation coefficient of 0.712; phenylacetaldehyde was significantly positively correlated with linalool, with a correlation coefficient of 0.937; acetic acid was significantly positively correlated with linalool, with a correlation coefficient of 0.635; n-hexanol was significantly positively correlated with 2-ethylhexanol, with a correlation coefficient of 0.774; benzyl alcohol was significantly positively correlated with isoamyl alcohol and acetic acid, with correlation coefficients of 0.754 and 0.671, respectively; isoamyl alcohol was significantly positively correlated with acetic acid, with a correlation coefficient of 0.663.

[0054] To improve the stability of the subsequent partial least squares regression model and avoid multicollinearity among variables, compounds with strong representativeness and high relative odor activity values ​​were retained for those with significant correlations (p<0.05). Accordingly, benzaldehyde was eliminated due to its significant correlation with n-pentanol; β-cyclocitral, nonanol, and (E)-2-heptenal were eliminated due to their significant correlations with n-hexanol or 2-ethylhexanol; linalool was eliminated due to its significant correlation with phenylacetaldehyde; and isopentanol and acetic acid were eliminated due to their significant correlations with benzyl alcohol, for a total of 7 eliminated compounds. Finally, 8 key flavor compounds—isopentanol, n-hexanol, phenylacetaldehyde, n-pentanol, n-hexanol, benzyl alcohol, ethyl acetate, and 2-ethylhexanol—were selected for subsequent modeling analysis.

[0055] It should be noted that the correlation coefficient between n-hexanol and 2-ethylhexanol is 0.774, which is significant. However, they belong to two different categories of flavor compounds, namely straight-chain primary alcohols and branched-chain alcohols, and they fall under different principal components in the principal component analysis in section 1.8: n-hexanol has a loading of 0.900 on the second principal component, while the absolute values ​​of the loadings of 2-ethylhexanol on all six principal components are all <0.5 (the maximum being 0.476). The information carried by the two has limited overlap, so both are retained.

[0056] After the two-part dimensionality reduction, a total of 15 quality indicators for dried chili peppers were obtained, which are denoted as follows: X1 isovaleraldehyde content, X2 n-hexanol content, X3 phenylacetaldehyde content, X4 n-pentanol content, X5 n-hexanol content, X6 benzyl alcohol content, X7 ethyl acetate content, X8 2-ethylhexanol content, X9 pulp content, X... 10 Moisture content, X 11 Browning degree, X 12 b * Value, X 13 Reducing sugar content, X 14 Capsaicin content, X 15 Capsaicin content.

[0057] 1.4 Preparation of chili oil Twenty modeling samples were prepared into chili oil. The preparation method was as follows: dried chilies were stemmed and then pulverized into uniform dried chili powder using a grinder. 24g of dried chili powder was weighed and placed in a clean stainless steel bowl. Separately, 100g of rapeseed oil was weighed, heated to 210℃, and then the heat was turned off. When it cooled to 180℃, it was immediately added to the stainless steel bowl containing the dried chili powder and stirred rapidly for 1 minute, followed by natural cooling. After the oil was cooked, it was poured into 400mL glass bottles and extracted for 24 hours. The supernatant oil was then collected to obtain the chili oil samples, totaling 20 samples, which corresponded one-to-one with the 20 modeling samples.

[0058] 1.5 Detection and Sensory Evaluation of Chili Oil Quality Indicators (1) Physicochemical indicators Eight physicochemical indicators were measured in 20 chili oil samples, in the following order: L * value, a * value, b * The study included measuring the following parameters: color value, capsaicin content, moisture content, acid value, and peroxide value. The color was determined by spreading 10g of chili oil evenly in a 5mL opaque glass and measuring the L value of the chili oil using a colorimeter. * value, a * Value and b *Capsaicin content was determined by high performance liquid chromatography (HPLC) according to GB / T 21266—2007, and the result is expressed in mg / kg. Capsanthin content was determined by the same method as the determination of capsanthin in dried chili peppers in section 1.2 above, and the result is expressed in g / kg. Moisture content was determined by direct drying method according to GB 5009.3—2016. Acid value was determined according to GB 5009.229—2025 "National Food Safety Standard - Determination of Acid Value in Food". Peroxide value was determined according to GB 5009.227—2023 "National Food Safety Standard - Determination of Peroxide Value in Food - Direct Titration Method". Each sample was tested in triplicate.

[0059] (2) Volatile flavor compounds The headspace volatile components of chili oil samples were analyzed using headspace solid-phase microextraction-gas chromatography-mass spectrometry (HS-SPME-GC-MS). Pretreatment conditions: 2.0 g of chili oil sample was accurately weighed into a 15 mL extraction flask, 5 μL of 1.013 μg / μL 2,4,5-trimethylthiazole internal standard was added, the flask was tightened, and the sample was heated in a 60 °C water bath for 20 min, followed by extraction for 30 min, and desorption at the injection port for 5 min. Gas chromatography conditions: A DB-5MS capillary column (30 m × 320 μm × 0.25 μm) was used, helium (purity > 99.999%) was used as the carrier gas, the flow rate was 1.2 mL / min, the injection port temperature was 250 °C, and splitless injection was used; the temperature program was an initial temperature of 40 °C held for 2 min, then increased to 240 °C at a rate of 6 °C / min. Mass spectrometry conditions: Full scan mode was used, with a scan range of 41 m / z to 330 m / z. Ionization was performed by electron impact ionization at an energy of 70 eV. The ion source temperature was 200 °C, and the quadrupole temperature was 150 °C. Qualitative analysis was performed using the NIST20 library. Quantitative analysis was performed using the internal standard method, with 2,4,5-trimethylthiazole as the internal standard. The relative content of each compound was calculated based on the ratio of its peak area to the internal standard peak area. Relative odor activity analysis identified 42 key aroma compounds with a relative odor activity value >1 in the chili oil.

[0060] (3) Sensory evaluation A professional sensory evaluation team of 16 evaluators was formed. The evaluators consisted of 16 graduate students from the School of Food Science and Technology (8 men and 8 women), aged 20-30, all in good health and without adverse reactions to chili oil products. All evaluators had received professional training in chili oil sensory evaluation before the evaluation and possessed normal olfactory and gustatory sensitivity. Tobacco, spices, and other irritating substances were prohibited within 2 hours prior to the evaluation. The evaluation was conducted in a temperature- and humidity-controlled, odor-free sensory laboratory. Samples were submitted with random three-digit codes. During the evaluation, evaluators were provided with purified water and cucumber slices to clean their mouths, reducing flavor residue interference between different samples. The evaluation dimensions included appearance, aroma, taste, and aftertaste, each scored on a 10-point scale according to the scoring criteria in Table 2. Higher scores indicated better quality in that dimension. Each sample was evaluated three times, and the average score from all 16 evaluators was taken as the sensory score for that dimension.

[0061] Table 2 Sensory Evaluation Scoring Criteria for Chili Oil Appearance (look) Aroma (smell) Taste Aftertaste Score The chili oil is dark in color, has poor transparency, contains a lot of sediment, has a thick texture, fine foam, and large suspended particles. The chili oil has a strong burnt and pungent flavor. The chili oil is not spicy enough, lacks the flavor of cooked chili peppers, and has a burnt and bitter taste. The aroma dissipates quickly in the mouth, or the aftertaste is bitter, dry, or has a monotonous burning sensation. 0~3 points The chili oil is pale yellow in color, has good transparency, a small amount of sediment, a thick texture, and a few fine bubbles. The chili oil is clear and transparent with small suspended particles. The rapeseed oil in the chili oil has a rich and strong aroma, which overpowers the aroma of the chili peppers. Chili oil can be quite spicy or mildly spicy, while cooked chili peppers have a weaker flavor and a slightly burnt or bitter taste. The aroma lingers in the mouth for a reasonable amount of time, and the mouth feels clean after the spiciness subsides, with no obvious bitterness or greasy residue. 4~6 points The chili oil is bright red in color, has good transparency, no sediment, and a thick consistency. The surface is clean and free of bubbles, and the chili oil is clear and transparent with no suspended matter. Chili oil has a smoky, roasted, nutty, and rich aroma of rapeseed oil. The chili oil has a moderate spiciness, a rich flavor from cooked chilies, and no burnt or bitter taste. It has a pleasant aftertaste, lingers in the mouth for a long time, and has no unpleasant aftertaste. 7~10 points 1.6 Determination of Core Quality Indicators for Chili Oil Using sensory scores of appearance, aroma, taste, and aftertaste as dependent variables, and the contents of 8 physicochemical indicators and 42 key aroma compounds as independent variables, Pearson correlation analysis was performed on each of them.

[0062] Of the eight physicochemical indicators, the L* value and moisture content showed no significant correlation with the sensory scores for appearance, aroma, taste, and aftertaste (p>0.05), and were therefore excluded. Thus, only a was retained among the physicochemical indicators. * value, b * The six items are: acid value, capsanthin content, capsanthin content, acid value, and peroxide value.

[0063] like Figure 5 As shown, 12 volatile compounds were significantly correlated with the aroma sensory score of chili oil (p<0.05), namely 2-n-pentylfuran, 2,3-dimethylpyrazine, D-limonene, pinene, 3-hydroxy-2-butanone, 2,3-pentanedione, 2,3-butanediol, linalool, phenylacetaldehyde, (E)-2-heptenal, 1-penten-3-one, and n-pentanol. Among them, 2,3-pentanedione and 2,3-dimethylpyrazine contribute roasted and charred aromas to chili oil; 3-hydroxy-2-butanone and phenylacetaldehyde contribute buttery and sweet aromas; 2-n-pentylfuran and linalool contribute fruity and woody aromas; 1-penten-3-one contributes a spicy and pungent aroma to chili oil and is significantly negatively correlated with the aroma score; the remaining compounds are all significantly positively correlated with the aroma score.

[0064] Further elimination of collinear variables was performed on the above 12 compounds: D-limonene and pinene belong to the terpenoid class and have a correlation coefficient of 0.862; 3-hydroxy-2-butanone and 2,3-pentanedione belong to the short-chain carbonyl class and have a correlation coefficient of 0.831. One of them with a stronger correlation to the aroma score was retained, namely pinene and 2,3-pentanedione, while D-limonene and 3-hydroxy-2-butanone were removed, resulting in 10 aroma compounds remaining.

[0065] like Figure 6 As shown, the indicators that were significantly correlated with the sensory evaluation of chili oil were capsanthin content and a * Value and b * The correlation coefficients were 0.735, 0.860, and -0.895, respectively (p<0.05). The indicators that were significantly correlated with the sensory scores of chili oil taste and aftertaste were capsaicin content, acid value, and peroxide value, with correlation coefficients of 0.697, -0.838, and -0.662, respectively (p<0.05).

[0066] In summary, 6 out of 8 physicochemical indicators and 10 out of 42 key aroma compounds were retained, together forming 16 core quality indicators for chili oil used in comprehensive quality evaluation, in the following order: a * value, b * Value, capsanthin content, peroxide value, acid value, capsanthin content, 2-n-pentylfuran content, 2,3-dimethylpyrazine content, pinene content, 2,3-pentanedione content, 1-penten-3-one content, linalool content, 2,3-butanediol content, n-pentanol content, phenylacetaldehyde content, and (E)-2-heptenal content.

[0067] 1.7 Calculation of the overall quality score of chili oil The entropy weighting method was used to objectively assign weights to 16 core quality indicators, and the comprehensive quality score was calculated based on the linear weighting method. Information entropy reflects the dispersion of quality indicator data, and information entropy is inversely proportional to the weights; that is, the higher the information entropy, the smaller the weight coefficient. The specific steps are as follows: The first step is to classify the indicators. Based on the correlation between the indicators and sensory scores in Section 1.6 above, the acid value, peroxide value, 1-penten-3-one, n-pentanol, and (E)-2-heptenal content are classified as negative indicators, meaning that the higher the value, the worse the quality of the chili oil; capsaicin, b* value, and phenylacetaldehyde content are classified as moderate indicators; the rest are classified as positive indicators, meaning that the higher the value, the better the quality of the chili oil.

[0068] The second step is range standardization. Let x... ij Let x be the measured value of the j-th indicator in the i-th chili oil sample, i=1,2,…,n, where n is the sample size and j=1,2,…,16; jmax With x jminThese are the maximum and minimum values ​​of the j-th indicator in all samples, respectively. ij Let x be the value of the j-th indicator in the i-th chili oil sample after range standardization, dimensionless, and ranging from 0 to 1. kj The measured value of the j-th indicator of the k-th chili oil in the top five sensory evaluations. Positive indicators are standardized according to formula (2), negative indicators according to formula (3), and moderate indicators according to formula (4): y ij =(x ij -x jmin ) / (x jmax -x jmin (2) y ij =(x jmax -x ij ) / (x jmax -x jmin (3) y ij =1 / (1+ABS(x ij - / 5)) (4) The third step is to calculate the information entropy and entropy weight. Calculate the feature weight of the i-th sample under the j-th indicator using equation (5), calculate the information entropy of the j-th indicator using equation (6), calculate the information utility value using equation (7), and calculate the weight using equation (8): (5) (6) d j =1-e j (7) w j =d j / Σ j=1 16 d j (8) In the formula, p ij e represents the feature weight of the i-th sample under the j-th indicator, which is dimensionless; j Let d be the information entropy of the j-th index, which is dimensionless; j w is the information utility value of the j-th indicator, which is dimensionless; j Let be the weight of the j-th indicator, dimensionless; n is the sample size, n=20 in this embodiment; j is the index of the core quality indicator, j=1,2,…,16; ln represents the natural logarithm to the base e; when p ij When =0, p is defined ij ×ln(p ij )=0.

[0069] Step 4: Linear weighting. Calculate the overall quality score F of the i-th chili oil sample according to equation (9). i : F i =Σ j=1 16 (w j ×y ij (9) In the formula, F i The overall quality score of the i-th chili oil sample is dimensionless; from equations (2) and (3), we can know that y ij The value of the j-th indicator in the i-th chili oil sample is standardized by range, dimensionless, and ranges from 0 to 1; w j Let be the weight of the j-th indicator, which is dimensionless; from equation (8), we know that the sum of the 16 weights is 1, therefore the overall quality score F i The value range is also 0~1, and the higher the score, the better the overall quality of the chili oil.

[0070] The information entropy, information utility value, and weight of the 16 core quality indicators calculated according to the above steps are shown in Table 3.

[0071] Table 3. Information entropy and entropy weight of each core quality indicator of chili oil Sensory attributes index <![CDATA[Information entropy e j > <![CDATA[information utility value d j > <![CDATA[weight w j > Appearance <![CDATA[a * Value 0.9076 0.0924 0.0654 Appearance <![CDATA[b * Value 0.9359 0.0641 0.0454 Appearance Capsaicin content 0.9276 0.0724 0.0512 Taste / Aftertaste Peroxide value 0.9429 0.0571 0.0404 Taste / Aftertaste Acid value 0.9435 0.0565 0.0400 Taste / Aftertaste Capsaicin content 0.8837 0.1163 0.0832 aroma 2-n-pentylfuran content 0.8732 0.1268 0.0897 aroma 2,3-Dimethylpyrazine content 0.9007 0.0993 0.0703 aroma longleaf pinene content 0.8839 0.1161 0.0821 aroma 2,3-Pentanedione content 0.8900 0.1100 0.0778 aroma 1-Penten-3-one content 0.9496 0.0504 0.0357 aroma Linalool content 0.8900 0.1100 0.0779 aroma 2,3-Butanediol content 0.9287 0.0713 0.0504 aroma n-Pentanol content 0.9797 0.0203 0.0144 aroma Benzaldehyde content 0.8407 0.1593 0.1127 aroma (E)-2-heptenal content 0.9094 0.0906 0.0641 As shown in Table 3, among the appearance attributes, chromaticity a * The weight coefficient of the value is the largest. Among the taste and aftertaste attributes, the weight coefficient of capsaicin content is the largest. Among the aroma attributes, the weight coefficients of phenylacetaldehyde, 2-n-pentylfuran, pinene, and linalool are relatively large. The comprehensive quality scores of 20 chili oil samples were calculated and ranked according to formula (9). The results showed that the comprehensive quality scores of chili oils made from Hainan Erjingtiao, Guizhou Erjingtiao, and Yunnan Erjingtiao ranked in the top 3, with better comprehensive quality; while the comprehensive quality scores of chili oils made from Yunnan Mantianxing, Xinjiang Mantianxing, Henan Denglongjiao, and Guizhou Denglongjiao were relatively poor.

[0072] 1.8 Screening of Key Indicators for Dried Chili Peppers To control the number of independent variables and avoid overfitting, key indicators need to be further screened from the 15 dried chili quality indicators obtained in Section 1.3 above. This embodiment uses Pearson correlation analysis, principal component analysis, and partial least squares regression variable projection importance method for screening, and compares the performance of the models built using these three methods in parallel. It should be noted that the three methods screen independently, and the number of candidate indicators obtained is not the same as the specific items. The final indicators are determined by the fitting and predictive ability of the model built in Section 1.9.

[0073] During the screening phase, a dependent variable matrix was constructed based on six core indicators of chili oil. The selection rule for these six indicators was as follows: according to the weights in Table 3, the indicator with the highest weight among the appearance attributes, i.e., 'a', was chosen. * The values ​​were determined by selecting the capsaicin content, which has the highest weight among the taste and aftertaste attributes, and the four most weighted indicators among the aroma attributes: phenylacetaldehyde content (0.1127), 2-n-pentylfuran content (0.0897), pinene content (0.0821), and linalool content (0.0779). These six dependent variables were denoted as Y1 (capsaicin content in chili oil), Y2 (a...), and Y3 (a...). * Value, phenylacetaldehyde content of Y3 chili oil, 2-n-pentylfuran content of Y4 chili oil, pinene content of Y5 chili oil, and linalool content of Y6 chili oil.

[0074] (1) Pearson correlation analysis Pearson correlation analysis was performed on 15 quality indicators of dried chili peppers and 6 core indicators of chili oil. The results are shown in Table 4.

[0075] Table 4. Correlation coefficients between dried chili quality indicators and core indicators of chili oil Dried chili index <![CDATA[Y1]]> <![CDATA[Y2]]> <![CDATA[Y3]]> <![CDATA[Y4]]> <![CDATA[Y5]]> <![CDATA[Y6]]> <![CDATA[X1 Isovaleraldehyde]]> -0.128 0.058 -0.036 0.245 -0.466* -0.104 <![CDATA[X2 n-hexanal]]> 0.048 0.020 0.171 -0.168 0.042 -0.015 <![CDATA[X3 phenylacetaldehyde]]> 0.000 0.446* 0.366 0.280 0.768** 0.595** <![CDATA[X4 n-pentanol]]> 0.525* -0.281 -0.054 -0.284 -0.313 -0.269 <![CDATA[X5 n-hexanol]]> 0.214 -0.032 0.114 0.017 -0.020 -0.041 <![CDATA[X6 Benzyl alcohol]]> 0.332 -0.032 -0.089 0.033 0.315 -0.003 <![CDATA[X7 ethyl acetate]]> -0.050 0.844** 0.507* 0.703** 0.604** 0.695** <![CDATA[X82-Ethylhexanol]]> 0.028 -0.457* -0.347 -0.435 -0.315 -0.490* <![CDATA[X9 pulp content]]> -0.434 0.693** 0.567** 0.627** 0.353 0.668** <![CDATA[X 10 Moisture content 0.275 -0.721** -0.422 -0.675** -0.695** -0.716** <![CDATA[X 11 Browning degree -0.009 -0.525* -0.138 -0.454* -0.460* -0.376 <![CDATA[X 12 b * Value 0.089 -0.793** -0.630** -0.800** -0.460* -0.723** <![CDATA[X 13 reducing sugars -0.043 0.465* 0.109 0.592** -0.059 0.035 <![CDATA[X 14 Capsaicin 0.953** -0.077 -0.173 -0.182 -0.105 -0.395 <![CDATA[X 15 Capsaicin -0.072 0.866** 0.476* 0.773** 0.484* 0.579** Note: In the table, * indicates p < 0.05, and ** indicates p < 0.01.

[0076] Table 4 shows that phenylacetaldehyde (X3), ethyl acetate (X7), pulp content (X9), and moisture content (X) are all related to the concentrations of phenylacetaldehyde (X3), ethyl acetate (X7), pulp content (X9), and moisture content (X9). 10 b * Value (X) 12 ), capsaicin (X) 14 ) and capsanthin (X) 15 The seven items are strongly correlated with the core indicators of chili oil and can be used as candidate variables for model construction.

[0077] (2) Principal component analysis Principal component analysis was performed on 15 quality indicators of dried chili peppers. Six principal components were extracted according to the principle that the eigenvalue is greater than 1. The loadings of each indicator on the six principal components are shown in Table 5. The eigenvalues, contribution rates and cumulative contribution rates of each principal component are shown in Table 6.

[0078] Table 5 Principal component loadings of dried chili pepper quality indicators index 1st principal component 2nd principal component 3rd principal component 4th principal component 5th principal component 6th principal component Isovaleraldehyde -0.062 -0.214 0.011 0.180 0.874 0.053 hexanol 0.016 0.208 -0.122 -0.129 0.080 0.911 phenylacetaldehyde 0.755 -0.023 0.028 -0.404 -0.275 0.262 n-Pentanol -0.161 -0.243 0.831 -0.347 0.095 -0.168 n-Hexyl alcohol -0.014 0.900 0.027 0.152 -0.026 0.037 benzyl alcohol 0.090 0.836 0.072 -0.127 -0.244 0.178 Ethyl acetate 0.808 0.181 -0.148 0.384 -0.106 -0.049 2-Ethylhexanol -0.476 -0.403 0.285 0.284 -0.294 0.448 Pulp content 0.713 -0.287 -0.351 -0.154 0.356 0.074 Moisture content -0.821 0.307 0.226 -0.140 0.279 0.120 Browning degree -0.677 -0.104 0.019 -0.064 0.262 -0.203 <![CDATA[b * Value -0.904 -0.091 -0.023 -0.062 -0.225 0.223 reducing sugars 0.212 0.005 -0.038 0.866 0.168 -0.092 Capsaicin -0.066 0.274 0.902 0.158 -0.066 0.022 Capsaicin 0.799 0.211 -0.036 0.285 0.329 -0.092 Table 6. Eigenvalues ​​and variance contribution rates of each principal component project 1st principal component 2nd principal component 3rd principal component 4th principal component 5th principal component 6th principal component Eigenvalues 4.984 2.384 1.755 1.572 1.141 1.054 Contribution rate / % 33.225 15.890 11.697 10.478 7.610 7.028 Cumulative contribution rate / % 33.225 49.115 60.812 71.290 78.900 85.928 As shown in Tables 5 and 6, the six principal components can reflect 85.928% of the information from the original 15 indicators and can be used to characterize the quality characteristics of dried chili peppers. The first principal component has the highest contribution rate at 33.225%. Ethyl acetate, capsanthin, phenylacetaldehyde, and pulp content have high positive loadings on this principal component. Moisture content and b... * The first principal component has a high loading rate, indicating that it integrates information on the color characteristics and some flavor compounds of dried chili peppers. The second principal component has a contribution rate of 15.890%, with high loading rates for n-hexanol and benzyl alcohol, reflecting the changing characteristics of volatile alcohol flavor compounds. The third principal component has a contribution rate of 11.697%, with high loading rates for capsaicin and n-pentanol, mainly related to spiciness compounds and some alcohol flavor compounds. The fourth principal component has a contribution rate of 10.478%, with a high loading rate for reducing sugars, reflecting the changes in sugar compounds. The fifth and sixth principal components have contribution rates of 7.610% and 7.028%, respectively, with isovaleraldehyde and n-hexanol as the main characteristics, revealing the changing patterns of volatile aldehydes.

[0079] Based on this, representative indicators are extracted according to the following rules: For the first principal component, indicators with an absolute loading value ≥ 0.79 are selected; for the second to sixth principal components, indicators with the largest absolute loading value and an absolute loading value ≥ 0.89 are selected respectively. Accordingly, b is extracted from the first principal component. * Four items were selected: pH (-0.904), moisture content (-0.821), ethyl acetate (0.808), and capsanthin (0.799). Phenylacetaldehyde (0.755), pulp content (0.713), and browning degree (-0.677) were not extracted because they did not meet the threshold. Hexanol (0.900) was extracted as the second principal component; capsanthin (0.902) as the third principal component; and hexanaldehyde (0.911) as the sixth principal component. The reducing sugar (0.866) and isovaleraldehyde (0.874), which had the largest absolute loading values ​​on the fourth and fifth principal components, respectively, did not meet the threshold of 0.89 and were therefore not extracted. These two thresholds can be adjusted appropriately according to the number of target variables; the lower the threshold, the more indicators are selected.

[0080] Based on this, principal component analysis was used to screen for ethyl acetate, capsanthin, and beta-carotene. * The seven candidate indicators are: value, moisture content, n-hexanol, capsaicin, and n-hexanal.

[0081] (3) Partial Least Squares Regression Variable Projection Importance Method Partial least squares regression analysis was conducted using 15 dried chili quality indicators as independent variables and 6 core chili oil indicators as dependent variables to obtain the projected importance values ​​of each variable. A projected importance value > 1 was used as the criterion for determining important variables. Figure 7As shown, the indicators with a variable projection importance value > 1 are ethyl acetate, capsanthin, capsaicin, moisture content, pulp content, and phenylacetaldehyde, totaling 6 items, with variable projection importance values ​​of 1.370, 1.340, 1.335, 1.324, 1.263, and 1.170, respectively. The variable projection importance values ​​of the remaining 9 indicators are all < 1. Therefore, the above 6 candidate indicators were obtained by screening using the variable projection importance method.

[0082] 1.9 Establishment of Partial Least Squares Regression Model Using the key indicators of dried chili peppers obtained through the three methods described above as independent variables and the comprehensive quality score of chili oil obtained in Section 1.7 as the dependent variable, a partial least squares regression model was established, and the cumulative coefficient of determination R0 was used as the dependent variable. 2 With cumulative cross-validation determination coefficient Q 2 The model's fit and predictive ability were evaluated, and the results are shown in Table 7.

[0083] Table 7. Performance Comparison of Three Index Screening Methods and Their Adopted Models Filtering methods The indicators obtained through screening <![CDATA[Cumulative R 2 > <![CDATA[Cumulative Q 2 > Pearson correlation method <![CDATA[Phenylacetaldehyde, capsanthin, ethyl acetate, pulp content, moisture content, capsaicin, b * value]]> 0.715 0.652 Importance of Variable Projection Method Capsaicin, ethyl acetate, moisture content, phenylacetaldehyde, capsanthin, pulp content 0.732 0.661 Principal component analysis <![CDATA[Ethyl acetate, capsanthin, b * value, moisture content, n-hexanol, capsaicin, n-hexanal]]> 0.806 0.731 Table 7 shows that the cumulative R-squared value of the partial least squares regression model built based on the indicators selected by principal component analysis is... 2 0.806, cumulative Q 2 The value was 0.731, which was significantly better than the models built by the other two screening methods. Therefore, the seven indicators selected by principal component analysis were used to build the final model.

[0084] Based on ethyl acetate content, capsanthin content, and b * Seven indicators—value, moisture content, n-hexanol content, capsaicin content, and n-hexanal content—were used as independent variables, with the overall quality score of chili oil as the dependent variable. Leave-one-out cross-validation was employed to determine the optimal number of principal components. When one principal component was extracted, the cumulative Q... 2 It reaches its maximum value of 0.731, at which point the accumulated R... 2 The regression coefficient was 0.806, the root mean square error of the training set was 0.028, and the mean absolute error was 0.021, indicating that the model has good fitting effect and predictive ability. The regression coefficients of each variable are shown in Table 8.

[0085] Table 8 Regression coefficients of the partial least squares regression model Independent variable symbol unit Regression coefficient constant term — — 0.4343 hexanal content A μg / kg <![CDATA[1.3398×10 -4 ]]> n-Hexol content B μg / kg <![CDATA[-1.2367×10 -4 <!-- 15 -->]]> Ethyl acetate content C μg / kg <![CDATA[8.0504×10 -5 ]]> Moisture content D Dimensionless (in decimal form) -2.1023 <![CDATA[b * Value E Dimensionless <![CDATA[-1.8198×10 -3 ]]> Capsaicin content F mg / g -0.0284 Capsaicin content G mg / g <![CDATA[2.4485×10 -3 ]]> Based on this, the regression equation for the comprehensive quality prediction model of chili oil is obtained, as shown in equation (10): Y = 0.4343 + 1.3398 × 10 -4 ×A-1.2367×10 -4 ×B+8.0504×10 -5 ×C-2.1023×D-1.8198×10-3 ×E-0.0284×F+2.4485×10 -3 ×G (10) In the formula, Y is the predicted score for the overall quality of chili oil, dimensionless; A, B, and C are the relative contents of hexanal, hexanol, and ethyl acetate in dried chili peppers, respectively, all in μg / kg; D is the moisture content of dried chili peppers, expressed as a mass fraction and rounded to the decimal, dimensionless; E is the moisture content of dried chili peppers... * Values ​​are dimensionless; F represents the capsaicin content of dried chili peppers, in mg / g; G represents the capsanthin content of dried chili peppers, in mg / g.

[0086] The signs of the regression coefficients in equation (10) are consistent with the correlation analysis results in Table 4: moisture content and b * The values ​​showed a significant negative correlation with the core indicators of chili oil, with a negative regression coefficient. Ethyl acetate and capsanthin showed a significant positive correlation with the core indicators of chili oil, with positive regression coefficients. Although capsaicin showed a highly significant positive correlation with the capsaicin content of chili oil (r=0.953), the total weight of aroma indicators in the overall quality score of chili oil reached 0.5924, significantly higher than the total weight of taste and aftertaste indicators (0.1636). However, capsaicin from dried chili peppers showed a negative correlation with aroma compounds such as phenylacetaldehyde, 2-n-pentylfuran, pinene, and linalool in chili oil. Therefore, its net effect on the overall quality score was negative, and the regression coefficient was negative, which is not contradictory.

[0087] To verify the practical application effect of the established PLSR prediction model, data were collected from dried chili pepper samples V1-V5 in Table 1 and their corresponding chili oil samples according to the indicators and data processing methods used in the modeling process. The key quality indicator X of the five dried chili pepper samples was substituted into the regression equation Y= The predicted values ​​for the five chili oils were calculated as follows: 0.4343 + 1.3398 × 10⁻⁴ × hexanal - 1.2367 × 10⁻⁴ × hexanol + 8.0504 × 10⁻⁵ × ethyl acetate - 2.1023 × moisture content - 1.8198 × 10⁻³ × b* value - 0.0284 × capsaicin + 2.4485 × 10⁻³ × capsanthin. The results are shown in Table 9. These results indicate a clear correlation between the predicted values ​​and the overall quality scores of the five chili oils. Kendall's coefficient of harmony was used to test the consistency between the predicted and measured values. The results showed that the Kendall correlation coefficient between the predicted and measured overall quality scores was 0.800 (p = 0.05), reaching a critical significance level, indicating that the model can accurately distinguish the overall quality of different samples.

[0088] Table 9. Predicted and Actual Values ​​of 5 Types of Verified Chili Oil variety Overall quality forecast Overall quality actual score V1 0.3294 0.4831 V2 0.3340 0.3936 V3 0.3351 0.4876 V4 0.3106 0.2749 V5 0.3172 0.3828 Comparative Example 1 In addition to replacing the independent variables in the modeling with those obtained by Pearson correlation analysis, such as phenylacetaldehyde, capsaicin, ethyl acetate, pulp content, moisture content, and capsaicin and beta-carotene, the modeling independent variables were also replaced. * Apart from the seven indicators, the remaining steps are the same as in Example 1. The cumulative R-squared value of the constructed partial least squares regression model... 2 0.715, cumulative Q 2 The value was 0.652, which was lower than that in Example 1. The reason is that the Pearson correlation method only examines the univariate correlation between the independent and dependent variables. Among the seven selected indicators, six (b* value, moisture content, ethyl acetate, capsanthin, phenylacetaldehyde, and pulp content) all had absolute loading values ​​≥0.7 on the first principal component, and their information highly overlapped. All seven indicators only covered the first and third principal components and did not touch the alcohol, sugar, and aldehyde information carried by the second, fourth, fifth, and sixth principal components.

[0089] Comparative Example 2 Except for replacing the independent variables in the modeling with the six indicators obtained by the variable projection importance method—capsaicin, ethyl acetate, moisture content, phenylacetaldehyde, capsanthin, and pulp content—the remaining steps were the same as in Example 1. The cumulative R-squared value of the constructed partial least squares regression model was... 2 0.732, cumulative Q 2 The value is 0.661. Although the variable projection importance method only includes 6 indicators, the number of items with an absolute value ≥0.7 on the first principal component (5 items) is less than that of the Pearson correlation method (6 items), and the redundancy is slightly lower. Therefore, its R² and Q² are slightly higher than those of Comparative Example 1. However, its selected indicators still only cover the first and third principal components and do not touch the alcohol and aldehyde flavor information carried by the second and sixth principal components. Therefore, it is still significantly lower than that of Example 1.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for screening dried chilies to predict chili oil quality based on raw material indicators, characterized in that, Includes the following steps: S1. Detect key indicators of the dried chili peppers to be screened. Key indicators include hexanal content, hexanol content, ethyl acetate content, moisture content, and b. * Value, capsaicin content and capsanthin content, of which b * The value represents the degree of yellow-blueness; S2. Substitute the key indicators detected in S1 into the pre-constructed comprehensive quality prediction model for chili oil, and output the comprehensive quality prediction score of the chili oil prepared from the dried chilies to be screened. The comprehensive quality prediction model for chili oil is a partial least squares regression model. Its independent variable is the key indicators of the dried chili sample, and its dependent variable is the comprehensive quality score of the chili oil corresponding to the dried chili sample. The comprehensive quality score is obtained by linearly weighting the core quality indicators of chili oil after weighting by the entropy weight method. S3. Sort the dried chilies to be screened from high to low according to the comprehensive quality prediction score, and select the dried chilies with the highest scores as the raw materials for making chili oil.

2. The method for screening dried chili peppers according to claim 1, characterized in that, In S2, the comprehensive quality prediction model for chili oil is constructed according to the following steps: S21. Take n portions of dried chili peppers as modeling samples, n≥20, and test 21 physicochemical indicators and volatile flavor substances in each modeling sample. S22. The dimensionality of 21 physicochemical indicators and volatile flavor substances was reduced to obtain 15 quality indicators of dried chili peppers. S23. Each modeling sample was prepared into chili oil. The physicochemical indicators and volatile flavor substances of the chili oil were tested and sensory evaluation was conducted to determine 16 core quality indicators of chili oil that were significantly related to the sensory score. After weighting by entropy weight method, the comprehensive quality score of each chili oil was obtained by linear weighting. S24. Seven key indicators were selected from 15 dried chili quality indicators, namely ethyl acetate content, capsanthin content, and b... * The values ​​of water content, n-hexanol content, capsaicin content, and n-hexanaldehyde content were used as the independent variables. The comprehensive quality score obtained from S23 was used as the dependent variable. The leave-one-out cross-validation method was used to determine the optimal number of principal components and a partial least squares regression model was established.

3. The method for screening dried chili peppers according to claim 2, characterized in that, In S22, dimensionality reduction includes two parts: dimensionality reduction of physicochemical indicators and dimensionality reduction of flavor substances. The dimensionality reduction of the physicochemical indicators is as follows: Pearson correlation analysis is performed on 21 physicochemical indicators. The strong correlation is determined by the absolute value of the correlation coefficient r ≥ 0.7 and p < 0.

05. Only one representative indicator is retained in each group of strongly correlated indicators. Then, indicators with a coefficient of variation < 10% among n modeling samples are removed, resulting in 7 physicochemical indicators. Flavor compounds were reduced to the following: For key aroma compounds with a relative odor activity value >1, compounds with a detection rate <70% were first removed. Then, for samples where the compound was not detected, the missing value was supplemented with half of the minimum detection value of the compound in all samples. Pearson correlation analysis was then performed to remove compounds that were significantly correlated with other compounds and had low relative odor activity values, resulting in 8 key flavor compounds.

4. The method for screening dried chili peppers according to claim 2, characterized in that, In S24, the method for screening the 7 key indicators is as follows: principal component analysis is performed on 15 dried chili quality indicators, and principal components are extracted according to the principle that the eigenvalue is >1. For the first principal component, the index with an absolute loading value ≥ 0.79 is selected; for each principal component other than the first principal component, the index with the largest absolute loading value and an absolute loading value ≥ 0.89 is selected; the above indices are combined to obtain 7 key indices.

5. The method for screening dried chili peppers according to claim 2, characterized in that, In S23, the 16 core quality indicators for chili oil are: a * value, b * Value, capsaicin content, peroxide value, acid value, capsaicin content, 2-n-pentylfuran content, 2,3-dimethylpyrazine content, pinene content, 2,3-pentanedione content, 1-penten-3-one content, linalool content, 2,3-butanediol content, n-pentanol content, phenylacetaldehyde content, and (E)-2-heptenal content, of which a * The value represents the degree of redness and greenness.

6. The method for screening dried chili peppers according to claim 5, characterized in that, In S23, the overall quality score is calculated using the following steps: The first step is to divide the 16 core quality indicators into positive indicators, negative indicators, and moderate indicators. Among them, acid value, peroxide value, 1-penten-3-one, n-pentanol, and (E)-2-heptenal content are negative indicators, capsaicin, b* value, and phenylacetaldehyde content are moderate indicators, and the rest are positive indicators. The second step is to process the positive indicators by y ij =(x ij -x jmin ) / (x jmax -x jmin ), for negative indicators by y ij =(x jmax -x ij ) / (x jmax -x jmin Standardize the range, and adjust the appropriate index according to y. ij =1 / (1+ABS(x ij - / 5)) standardize to obtain the standardized value y ij In the formula, i is the sample number, j is the index number, and x ij Let x be the measured value of the j-th indicator in the i-th chili oil sample. jmax With x jmin Let x be the maximum and minimum values ​​of the j-th indicator in all samples. kj The measured value of the j-th indicator of the k-th chili oil sample, which is among the top five in sensory evaluation; Third step, press Calculate the feature weight p of the i-th sample under the j-th indicator. ij And stipulate that when p ij =0 when p ij ×ln(p ij )=0; Press Calculate the information entropy e of the j-th indicator. j , where n is the number of samples; according to d j =1-e j Calculate the information utility value d j Press w j =d j / Σd j Calculate weight w j ; Step 4, press F i =Σ(w j ×y ij Linear weighting is used to obtain the overall quality score F of the i-th chili oil. i .

7. The method for screening dried chili peppers according to claim 2, characterized in that, The partial least squares regression model has a principal component count of 1, and its regression equation is: Y=0.4343+1.3398×10 -4 ×A-1.2367×10 -4 ×B+8.0504×10 -5 ×C-2.1023×D-1.8198×10 -3 ×E-0.0284×F+2.4485×10 -3 ×G; In the formula, Y is the predicted score for the overall quality of chili oil, dimensionless; A, B, and C are the relative contents of hexanal, hexanol, and ethyl acetate in dried chili peppers, respectively, all in μg / kg; D is the moisture content of dried chili peppers, expressed as a mass fraction and rounded to the decimal, dimensionless; E is the moisture content of dried chili peppers... * Values ​​are dimensionless; F represents the capsaicin content of dried chili peppers, in mg / g; G represents the capsanthin content of dried chili peppers, in mg / g.