Method for predicting relative correction factor of monomer raw material in atomized product and quantitative detection method

Abstract

This invention relates to the field of aerosol analysis technology, and in particular to a method for predicting and quantitatively detecting the relative correction factor of monomeric raw materials in aerosols. The prediction method includes: obtaining the molecular structure of the target monomeric raw material; calculating multiple predefined molecular structure descriptors; and substituting the molecular structure descriptors into a prediction model established through multivariate quadratic regression fitting to calculate the predicted relative correction factor of the target monomer. This invention also provides a quantitative detection method based on the prediction model. This method requires only molecular structure information to accurately predict the relative correction factor, completely eliminating dependence on expensive and scarce chemical standards, significantly reducing detection costs, simplifying the process, and improving the efficiency of quantitative analysis of aerosol components and product development.

Description

Technical Field

[0001] This invention relates to the field of atomized material analysis technology, and in particular to a method for predicting and quantitatively detecting the relative correction factor of monomeric raw materials in atomized materials. Background Technology

[0002] Atomized products are mainly composed of propylene glycol (PG), glycerol (VG), nicotine, and various flavorings and fragrances. The precision and stability of the atomized product formulation directly affect the product's sensory quality, safety, and compliance. Therefore, accurate quantitative analysis of key components in atomized products, especially the diverse range of monomeric raw materials, has become a core aspect of quality control and safety supervision in the industry.

[0003] Currently, the industry commonly employs gas chromatography-flame ionization detector (GC-FID) combined with the internal standard method for quantitative analysis of atomized components. This method requires obtaining the relative correction factor (RCF) of the target component relative to the internal standard, which is a key parameter for quantitative calculation.

[0004] In the existing technology, the methods for obtaining relative correction factors are mainly divided into two categories: (1) Experimental determination method: This is the mainstream and necessary method in the current industry. For example, the detection method disclosed in patent CN201911293221A still requires the preparation of standard samples first, and the correction factor is calculated by measuring the peak area of ​​the instrument. This method is essentially dependent on the physical acquisition and detection of high-purity chemical standards. (2) Model prediction method: In the field of analytical chemistry, in order to reduce the dependence on standards, researchers have tried to establish quantitative structure-activity relationship models between the molecular structure of compounds and the analytical response value. For example, Du Xihua et al. ("Using Topological Indices") m The paper "T Predicts the Relative Molecular Weight Correction Factor of Compounds in Thermal Conductivity" successfully predicted the correction factors of various organic compounds on thermal conductivity detectors based on molecular topological indices. This demonstrated a stable correlation between molecular structure descriptors and correction factors, providing an important theoretical reference for obtaining correction factors through calculation rather than experimentation. Furthermore, in other fields (such as CN120992538A and CN120890922A), near-infrared spectroscopy, hyperspectral imaging, and other techniques combined with machine learning algorithms have been used to establish component content prediction models, enabling rapid and non-destructive detection of certain specific systems.

[0005] However, applying the aforementioned existing technologies to the quantitative analysis of monomeric raw materials in atomized products still faces significant challenges and limitations: experimental determination methods reveal fundamental flaws in atomized product analysis; the types of flavor monomers used in atomized products are numerous and constantly updated, and many monomer standards (especially specialty, rare, or newly synthesized flavors) are scarce, expensive, or even unavailable on the market; this results in extremely high costs and cumbersome processes for the one-to-one testing model, severely restricting the rapid detection of multi-component, large-volume samples and the efficiency of product development. Existing model prediction methods cannot directly solve the problems in this field: the research of Du Xihua et al. targets thermal conductivity detectors (TCD) and specific topological indices, while the mainstream detector for atomized products is the flame ionization detector (FID), and the response principles of the two are different, so the models cannot be directly applied or transferred. Spectroscopic prediction models in other fields (such as the detection of tea polyphenols and glycyrrhizin) are completely different in technical approach from the atomized product component analysis based on chromatographic separation principles, and cannot be used to predict the key parameter in chromatographic analysis—the relative correction factor.

[0006] Therefore, the atomized products industry urgently needs a universal method that can break free from dependence on physical standards and quickly and accurately predict the relative correction factor of FID based solely on molecular structure information, in order to overcome the bottleneck of quantitative analysis, reduce costs, and improve efficiency. Summary of the Invention

[0007] Given the problems of existing quantitative analysis of atomized components, such as heavy reliance on standards, low efficiency, high cost, and difficulty in adapting to complex and variable fragrance systems, this invention systematically constructs a dedicated prediction model based on molecular structure descriptors. This completely changes the traditional method of relying on physical standards for quantification, and provides a brand-new technical path for rapid, accurate, and low-cost analysis and product development of atomized components.

[0008] The first aspect of this invention provides a method for predicting the relative correction factor of monomer raw materials in atomized products, comprising the following steps:

[0009] Obtain molecular structure information of the target monomer raw material;

[0010] Based on the molecular structure information, the values ​​of multiple predefined molecular structure descriptors are determined;

[0011] The numerical value of the molecular structure descriptor is substituted into a pre-established relative correction factor prediction model to obtain the predicted relative correction factor of the monomer raw material relative to the internal standard; wherein, the relative correction factor prediction model represents the functional relationship between the relative correction factor of the monomer raw material and the numerical value of its molecular structure descriptor.

[0012] A second aspect of the present invention provides a method for quantitative detection of monomeric raw materials in atomized products, comprising the following steps:

[0013] S1. Add a known concentration of internal standard solution to the nebulized sample to be tested, and use gas chromatography to detect it, and obtain the chromatographic peak areas of the target monomer raw material and the internal standard.

[0014] S2. Using the above-mentioned relative correction factor prediction method, predict the relative correction factor f of the target monomer raw material. i ;

[0015] S3. Calculate the content Ci of the monomer raw material in the sample to be tested according to the following formula:

[0016]

[0017] Among them, S i Let C1 be the peak area of ​​component i, C1 be the concentration of the internal standard, S1 be the peak area of ​​the internal standard, and f be the peak area of ​​component i. i , where is the predicted relative correction factor obtained by the relative correction factor prediction model for component i, V is the sample volume, and m is the sample mass.

[0018] Compared with the prior art, the present invention has the following advantages:

[0019] 1. The relative factor prediction method for monomer raw materials in atomized products provided by this invention only requires obtaining the molecular structure information of the monomer raw materials. By calculating the value of its structure descriptor and substituting it into the prediction model, its relative correction factor in GC-FID detection can be obtained, which solves the quantitative analysis bottleneck caused by the scarcity, high cost or difficulty in obtaining fragrance monomer standards in the atomized product industry.

[0020] 2. Traditional internal standard methods require measured correction factors for each monomer, which is cumbersome and costly. This invention, through a single modeling step, can cover hundreds of monomers. In actual testing, only one GC-FID analysis of the sample is needed to quickly complete the simultaneous quantification of multiple components based on the model prediction results, improving analytical efficiency by tens of times and significantly reducing detection costs.

[0021] 3. This invention constructs a dedicated multiple quadratic regression prediction model based on a large-scale training set containing 312 monomers. The coefficient of determination (Rsq) of the prediction model is greater than 0.91, and the proportion of predicted values ​​with relative errors within ±10% to measured values ​​exceeds 88%. Validated on an independent validation set, the model is stable and reliable, fully meeting the accuracy and reproducibility requirements of industrial quantitative analysis.

[0022] 4. The prediction method and quantitative detection method established in this invention provide a new, efficient, accurate, and low-cost solution for the product development, formula analysis, production quality control, and market compliance testing of atomized substances, and have good application prospects and promotion value. Attached Figure Description

[0023] Figure 1 This is a scatter plot comparing the predicted and measured values ​​of the monomer raw material relative correction factor model in an embodiment of the present invention.

[0024] Figure 2 This is a bar chart showing the distribution of the ratio of the model-predicted relative correction factor to the measured value in an embodiment of the present invention. Detailed Implementation

[0025] The following specific embodiments further illustrate the prediction method and quantitative detection method of the relative correction factor of monomer raw materials in the atomized material of the present invention.

[0026] It should be noted that, unless there is a conflict, the various features in the embodiments of this invention can be combined with each other, and all are within the protection scope of this invention. Furthermore, the terms "first" and "second" used in this invention do not limit the data or execution order, but only distinguish between identical or similar items with essentially the same function and effect.

[0027] This invention provides a method for predicting the relative correction factor of monomer feedstock in atomized products. By establishing a dedicated prediction model through regression analysis, a model can be established that correlates the molecular structural characteristics of monomer feedstock with its response behavior on gas chromatography-flame ionization detector (GC-FID). This allows for high-precision prediction of the relative correction factor of monomer feedstock relative to an internal standard (such as n-heptadecane) without the need for a pure standard of the monomer.

[0028] Based on the above prediction method, this invention further provides a quantitative detection method for monomeric raw materials in atomized products. The quantitative detection method combines the prediction model with conventional GC-FID analysis and the internal standard method: first, the chromatographic peak areas of the target monomer and the internal standard are obtained through GC-FID detection; then, the relative correction factor of the monomer is obtained using the prediction model; finally, the accurate content is calculated by substituting it into the internal standard method formula. The quantitative detection method provided by this invention completely eliminates the dependence on numerous standards. Furthermore, the prediction and quantitative methods described in this invention can also be further applied to fields such as rapid fragrance imitation in atomized products, formula analysis, and production quality control.

[0029] The first aspect of this invention provides a method for predicting the relative correction factor of monomer raw materials in atomized products, comprising the following steps:

[0030] (1) Obtain the molecular structure information of the target monomer raw material.

[0031] Molecular structure information refers to data that uniquely identifies the chemical structure of a target monomer raw material. It typically includes the molecular formula, structural formula, molecular weight, functional group information, and stereochemical information. For known compounds, the molecular formula, structural formula, and molecular weight can be obtained by consulting standard spectra or literature using their chemical name or CAS number. For example, the molecular structure information of leaf ester isobutyrate (CAS: 41519-23-7) can be obtained from the *Handbook of Fragrance Chemistry* or the PubChem database. Alternatively, cheminformatics software (such as ChemDraw, OpenBabel, and RDKit) can be used to directly read or calculate the molecular structure information.

[0032] (2) Based on the molecular structure information, determine the values ​​of multiple predefined molecular structure descriptors.

[0033] Studies have shown that combustion enthalpy can be calculated by summing the contributions of each atom in the molecule. Furthermore, specific molecular structures (such as unsaturated bonds, heteroatoms, and cyclic structures) can alter the combustion path and ionization efficiency of a compound. Therefore, it is necessary to select structural features that comprehensively reflect the types and numbers of atoms, as well as the interatomic connections, as independent variables.

[0034] Based on the above principles, this invention screens out molecular structure descriptors that can effectively characterize the molecular structure of monomer raw materials and significantly affect their response behavior on GC-FID from multiple dimensions such as carbon skeleton characteristics, functional group characteristics, unsaturation characteristics, heteroatom characteristics, branching degree characteristics, cyclic structure characteristics, and molecular weight characteristics.

[0035] In some embodiments of the present invention, the plurality of predefined molecular structure descriptors include at least one of the following types of descriptors:

[0036] A carbon atom descriptor includes at least one of a C atom, a CH group, a CH2 group, and a CH3 group; the number of C atoms refers to a carbon atom that is tetravalent, fully bonded, and not connected to a hydrogen atom.

[0037] Functional group descriptors include at least one of the following: >C=O group, >C=C< group, N=C< group, -OH or -SH group, >NH group, -NH2 group, >COC< or >CSC< group.

[0038] The heteroatom descriptor includes the N primitive number; the N atom refers to a nitrogen atom that is fully trivalent and not connected to a hydrogen atom.

[0039] Branch structure descriptor, including at least one of carbon branches and the total number of atoms in the branches excluding hydrogen.

[0040] Ring structure descriptors include at least one of the following: non-aromatic ring, aromatic ring, π-π conjugated bond, and total number of ring atoms.

[0041] Substituent position descriptor, including the ratio of the largest substituent position to the total number of atoms excluding H.

[0042] Electronic structure parameter descriptor, including the O value of C-1 / 2; the O value of C-1 / 2 refers to: the total number of carbon atoms (C) in the molecule − 0.5 × the total number of oxygen atoms (O) − the total number of nitrogen atoms (N).

[0043] A molecular weight ratio descriptor, which includes the ratio of the molecular weight of the monomer raw material to the molecular weight of the internal standard.

[0044] In some specific embodiments of the present invention, to further improve the prediction accuracy of the model, the molecular structure descriptors include the following 21 types: X1: C atom (a tetravalent carbon atom that is fully bonded and not connected to hydrogen atoms); X2: CH group; X3: CH2 group; X4: CH3 group; X5: >C=O group; X6: >C=C< group; X7: N=C< group; X8: -OH or -SH group; X9: N atom (a trivalent nitrogen atom that is fully bonded and not connected to hydrogen atoms); X10: >NH group; X11: -NH2 group; X12: > COC < or > CSC < (ether bond or thioether bond); X13: carbon branch; X14: total number of atoms in the branch excluding hydrogen; X15: non-aromatic ring; X16: aromatic ring; X17: π-π conjugated bond; X18: total number of cyclic atoms; X19: maximum position of substituent / total number of atoms excluding H (ratio of maximum position of substituent to total number of atoms excluding hydrogen); X20: O value of C-1 / 2 (total number of carbon atoms in the molecule − 0.5 × total number of oxygen atoms); X21: ratio of the molecular weight of the monomer raw material to the molecular weight of the internal standard (e.g., Mi / M(C17)).

[0045] The numerical values ​​of the above molecular structure descriptors can be obtained manually or directly using molecular structure analysis software.

[0046] (3) Construction of the relative correction factor prediction model.

[0047] The construction of the relative correction factor prediction model includes the following steps:

[0048] S1. Select a variety of individual raw materials to form a training dataset.

[0049] A training dataset is composed of various monomeric raw materials, which should cover common aroma compound categories in atomized products to enhance the generalization ability of the model. At least 200 monomeric raw materials are selected. In one specific embodiment of the invention, 312 common atomized product monomeric raw materials are collected, covering major aroma compound categories such as esters, alcohols, aldehydes, ketones, acids, lactones, alkenes, and heterocyclic compounds.

[0050] S2. Obtain the molecular structure formula of each monomer raw material in the training dataset, and determine the values ​​of multiple predefined molecular structure descriptors.

[0051] S3. Obtain the detection value of the relative correction factor of each monomer raw material in the training dataset;

[0052] Specifically, the determination was performed using gas chromatography-flame ionization detector (GC-FID) combined with the internal standard method.

[0053] The internal standard is selected from substances with good thermal stability, no reaction with sample components, and good separation from the chromatographic peaks of the target component, such as any one of n-heptadecane, its homologues, or heterocyclic compounds. Accurately weigh each monomer raw material standard, dissolve and dilute to volume with an internal standard solution of known concentration to prepare a series of standard solutions of known concentration.

[0054] The chromatographic conditions include: a polar column; an injection port temperature of 230-250°C; a detector temperature of 250-280°C, and a programmed temperature ramp. Specifically, the programmed temperature ramp is as follows: an initial column temperature of 40-60°C, held for 1-3 minutes, then increased at 8-12°C / min to 170-190°C, and then further increased at 8-12°C / min to 240-260°C, held for 3-7 minutes. It should be noted that the above chromatographic conditions are merely examples. Those skilled in the art can appropriately adjust the column type, temperature ramp, and other parameters according to actual analytical needs, and all such adjusted technical solutions fall within the scope of protection of this invention.

[0055] Under the above conditions, inject and analyze the samples, and record the peak areas of each monomeric raw material standard and internal standard. Calculate the relative correction factor of each monomeric raw material relative to the internal standard using the internal standard method formula. Specifically, the relative correction factor Fi is calculated using the following formula:

[0056] Fi=

[0057] Where Ai is the peak area of ​​standard i, As is the peak area of ​​internal standard, Cs is the concentration of internal standard in standard solution, and Ci is the concentration of standard i in standard solution.

[0058] Using the above method, the relative correction factor detection values ​​of each monomer raw material in the training dataset are obtained.

[0059] S4. Using the numerical value of the molecular structure descriptor as the independent variable and the relative correction factor detection value of the monomer raw material as the dependent variable, perform regression analysis to fit and obtain the relative correction factor prediction model.

[0060] In some embodiments of the present invention, the regression analysis is a multiple quadratic regression analysis, and the prediction model is a multiple quadratic regression model. The multiple quadratic regression model can simultaneously consider the linear and quadratic nonlinear effects of each descriptor, more accurately reflecting the complex relationship between molecular structure and FID response. The least squares method is used to perform regression fitting on the training set data to obtain the linear and quadratic coefficients of each descriptor in the model.

[0061] After the model is fitted, the goodness of fit needs to be evaluated to determine the model's ability to interpret the training data and its fitting effect. This invention uses the following metrics to comprehensively evaluate the goodness of fit:

[0062] 1) Coefficient of determination (R²) 2 R0 is used to measure how well a model explains the variation in the dependent variable. 2 The closer the value is to 1, the better the model fits the data.

[0063] 2) Adjusted coefficient of determination (R²) 2 Considering the influence of the number of independent variables, for R 2 Corrections should be made to avoid artificially inflating the goodness of fit by adding independent variables.

[0064] 3) Standard error (S): Used to measure the average deviation between the predicted value and the measured value. The smaller the S, the higher the prediction accuracy of the model.

[0065] 4) Distribution of the ratio of predicted values ​​to measured values: The percentage of samples in the training set whose ratio of predicted values ​​to measured values ​​is in the range of 90% to 110% directly reflects the prediction accuracy of the model.

[0066] In some embodiments of the present invention, the prediction model is validated using a validation dataset containing multiple (e.g., at least 50) monomer raw materials. Specifically, the numerical values ​​of the molecular structure descriptors of each monomer raw material in the validation dataset are substituted into the prediction model to calculate its prediction relative correction factor; the prediction relative correction factor is compared with the detected value to evaluate the prediction accuracy of the prediction model.

[0067] In a specific embodiment of the present invention, 200 monomeric raw materials were used as a training set for multiple quadratic regression analysis to fit and obtain the prediction model. The coefficient of determination R of the model was calculated. 2 It is 91.35%, and the adjusted R 2 The accuracy rate was 89.26%, the standard error S was 0.0474, and the samples with a predicted value to measured value ratio in the range of 90% to 110% accounted for 88.5% of the total training set. These indicators show that the multivariate quadratic regression model has an excellent fitting effect on the training data, and the selected 21 molecular structure descriptors can effectively explain the variation of the relative correction factor.

[0068] (4) Substitute the numerical value of the molecular structure descriptor into the pre-established relative correction factor prediction model to obtain the predicted relative correction factor of the monomer raw material relative to the internal standard; wherein, the relative correction factor prediction model represents the functional relationship between the relative correction factor of the monomer raw material and the numerical value of its molecular structure descriptor.

[0069] By substituting the molecular structure descriptor value of the target monomer raw material into the pre-established relative correction factor prediction model, the predicted relative correction factor of the monomer raw material relative to the internal standard can be obtained.

[0070] A second aspect of the present invention provides a method for quantitative detection of monomeric raw materials in atomized products, comprising the following steps:

[0071] S1. Add a known concentration of internal standard solution to the nebulized sample to be tested, and use gas chromatography to detect it, and obtain the chromatographic peak areas of the target monomer raw material and the internal standard.

[0072] In some embodiments of the present invention, the internal standard is other thermally stable substances not listed in the monomer raw material catalog, preferably any one of n-heptadecane, homologues of n-heptadecane, and heterocyclic compounds, more preferably n-heptadecane; the concentration of the internal standard in the present invention is 400~600 mg / L.

[0073] In some embodiments of the present invention, the detector for the gas chromatograph is a flame ionization detector.

[0074] The chromatographic conditions were the same as those used during the model construction described above. Specifically: a polar column; injection port temperature of 230–250°C; detector temperature of 250–280°C; and a programmed temperature ramp. The programmed temperature ramp was as follows: initial column temperature 40–60°C, held for 1–3 min, then increased at 8–12°C / min to 170–190°C, and then increased at 8–12°C / min to 240–260°C, held for 3–7 min.

[0075] S2. Using the above-mentioned relative correction factor prediction method, predict the relative correction factor f of the target monomer raw material. i .

[0076] S3. Calculate the content Ci of the monomer raw material in the sample to be tested according to the following formula:

[0077]

[0078] Among them, S i Let C1 be the peak area of ​​component i, C1 be the concentration of the internal standard, S1 be the peak area of ​​the internal standard, and f be the peak area of ​​component i. i, where is the predicted relative correction factor obtained by the relative correction factor prediction model for component i, V is the sample volume, and m is the sample mass.

[0079] This invention directly obtains the predicted relative correction factors for each component through a relative correction factor prediction model, eliminating the need to prepare separate standards and measure correction factors for each analyte. Compared to traditional methods, this invention simplifies quantitative analysis from a one-sample-one-test approach to requiring only one internal standard for simultaneous quantification of multiple components. This effectively overcomes the bottleneck in quantitative analysis caused by the wide variety of aromatic raw materials and the difficulty or high cost of obtaining standards, significantly improving analytical efficiency and reducing detection costs.

[0080] Before further describing specific embodiments of the present invention, it should be understood that the scope of protection of the present invention is not limited to the specific embodiments described below; it should also be understood that the terminology used in the embodiments of the present invention is for describing specific embodiments and not for limiting the scope of protection of the present invention.

[0081] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. In addition to the specific methods, apparatus, and materials used in the embodiments, this invention can be implemented using any prior art methods, apparatus, and materials similar to or equivalent to those described in the embodiments of this invention, based on the knowledge of the prior art possessed by one of ordinary skill in the art and the description of this invention.

[0082] Unless otherwise stated, the experimental methods, detection methods, and preparation methods disclosed in this invention all employ conventional techniques in analytical chemistry and related fields. Unless otherwise stated, all materials and equipment used in this invention are commercially available.

[0083] Example 1: Construction of a Relative Correction Factor Prediction Model

[0084] 1. Main instruments and reagents

[0085] Instrument: Agilent 8890 gas chromatograph equipped with a flame ionization detector (FID).

[0086] Column: DB-FATWAXUI (30m×0.25mm×0.25μm).

[0087] Reagents: Isopropanol (chromatographic grade); n-Heptadecane (chromatographic grade, used as internal standard); various nebulizer monomer raw material standards involved in the examples (purity >95%).

[0088] Preparation of internal standard solution: Accurately weigh 0.5000 g of n-heptadecane into a 1 L volumetric flask, and dilute to volume with isopropanol to prepare a 500 mg / L internal standard solution.

[0089] 2. Obtain the measured value of the relative correction factor.

[0090] Accurately weigh approximately 0.0500 g of monomer raw material standard into a 10 mL volumetric flask, dilute to the mark with the above C17 internal standard solution, and vortex to mix. Transfer 2 mL to a sample vial.

[0091] Sample injection and analysis were performed under the following GC-FID conditions: Injector temperature 250℃; split ratio 20:1; injection volume 0.1μL; carrier gas N2, flow rate 1.0mL / min; temperature program: 50℃ held for 2min, increased to 180℃ at 10℃ / min and held for 5min, then increased to 250℃ at 10℃ / min and held for 5min; FID temperature 275℃, hydrogen 40mL / min, air 400mL / min, and make-up nitrogen 30mL / min.

[0092] The relative correction factor measured value (F_measured) is calculated by the following formula: F_measured = (Ai / Ci) / (A_internal_standard / C_internal_standard), where Ai and A_internal_standard are the peak areas of the monomer raw material and the internal standard, respectively, and Ci and C_internal_standard are the concentrations of the monomer raw material and the internal standard, respectively.

[0093] 3. Establishment of the relative correction factor prediction model

[0094] 3.1 Data Preparation and Molecular Structure Descriptor Calculation

[0095] Measured F values ​​were collected for 312 common atomizing monomer raw materials (covering esters, alcohols, aldehydes, ketones, acids, lactones, alkenes, heterocyclic compounds, etc.). For each monomer raw material, 21 molecular structure descriptors (X1 to X21) were calculated based on its molecular structure formula. The values ​​of the 21 molecular structure descriptors are described as follows: X1: Number of C atoms, i.e., the number of tetravalent carbon atoms that are fully bonded and not connected to hydrogen atoms; X2: Number of CH groups; X3: Number of CH2 groups; X4: Number of CH3 groups; X5: Number of >C=O groups; X6: Number of >C=C< groups; X7: Number of N=C< groups; X8: Number of -OH or -SH groups; X9: Number of N atoms, i.e., the number of trivalent nitrogen atoms that are fully bonded and not connected to hydrogen atoms; X10: Number of >NH groups; X1 1: Number of -NH2 groups; X12: Number of groups >COC < or >CSC <; X13: Number of carbon branches; X14: Total number of atoms in the branches excluding hydrogen; X15: Number of non-aromatic rings; X16: Number of aromatic rings; X17: Number of π-π conjugated bonds; X18: Total number of cyclic atoms; X19: Maximum position of substituent / Total number of atoms excluding H; X20: O value of C-1 / 2; X21: Ratio of the molecular weight of the fragrance monomer to the molecular weight of n-heptadecane, Mi / M (C17).

[0096] The following is a list of the calculation process for molecular structure descriptors of some monomer raw materials:

[0097] (1) Ethyl 2-methylbutyrate, with the following structural formula: The molecular formula is C7H 14 O2, molecular weight = 130, number of C atoms = 1, number of CH atoms = 1, number of CH2 atoms = 2, number of CH3 atoms = 3, number of >C=O atoms = 1, number of >C=C< atoms = 0, number of -N=C< atoms = 0, number of -O(S)H atoms = 0, number of N atoms = 0, number of >NH atoms = 0, number of -NH2 atoms = 0, number of >CO(S)-C< atoms = 1, number of carbon branches = 1, total number of branched atoms (excluding H) = 1, number of non-aromatic rings = 0, number of aromatic rings = 0, number of π-π conjugated bonds = 0, number of cyclic atoms = 0, maximum position of substituent / total number of atoms excluding H = 4 / 9 = 0.44, O in C-1 / 2 = 6, Mi / M(C17) = 130 / 240 = 0.54;

[0098] (2) 2-Acetylfuran, structural formula is The molecular formula is C6H6O2, the molecular weight is 110, the number of C atoms is 2, the number of CH atoms is 3, the number of CH2 atoms is 0, the number of CH3 atoms is 1, the number of >C=O atoms is 1, the number of >C=C< atoms is 2, the number of -N=C< atoms is 0, the number of -O(S)H atoms is 0, the number of N atoms is 0, the number of >NH atoms is 0, the number of -NH2 atoms is 0, the number of >CO(S)-C< atoms is 1, the number of carbon branches is 1, the total number of branches (excluding H) is 3, the number of non-aromatic rings is 0, the number of aromatic rings is 1, the number of π-π conjugated bonds is 3, the number of cyclic atoms is 5, the maximum position of substituents / the total number of atoms excluding H is 1 / 8 = 0.13, the number of O atoms in C-1 / 2 is 5, and the Mi / M(C17) = 110 / 240 = 0.46.

[0099] (3) Ethyl acetoacetate, structural formula is Its molecular formula is C6H 10 O3, molecular weight = 130, number of C atoms = 2, number of CH atoms = 0, number of CH2 atoms = 2, number of CH3 atoms = 2, number of >C=O atoms = 2, number of >C=C< atoms = 0, number of -N=C< atoms = 0, number of -O(S)H atoms = 0, number of N atoms = 0, number of >NH atoms = 0, number of -NH2 atoms = 0, number of >CO(S)-C< atoms = 1, number of carbon branches = 0, total number of branches (excluding H) = 0, number of non-aromatic rings = 0, number of aromatic rings = 0, number of π-π conjugated bonds = 0, number of cyclic atoms = 0, maximum position of substituent / total number of atoms excluding H = 4 / 9 = 0.44, O in C-1 / 2 = 4.5, Mi / M(C17) = 130 / 240 = 0.54.

[0100] The calculation results of molecular structure descriptors for some monomer raw materials are shown in Table 1 below.

[0101] Table 1. Calculation results of molecular structure descriptors for some monomer raw materials.

[0102]

[0103] 3.2 Model Establishment

[0104] The 312 samples were randomly divided into a training set (200 samples) and a validation set (112 samples). Using 21 molecular structure descriptors from the training set as independent variables and the corresponding measured F-values ​​as the dependent variable, a multiple quadratic regression analysis was performed to fit the predicted model equation. The model fit index was the coefficient of determination R0. 2 =91.35%, adjusted R 2 = 89.26%, standard error S=0.0474.

[0105] F 预测 = 0.5577-0.0336X1-0.0511X2-0.0145X3+0.0116X4-0.0342X5+0.0305X6+0.0150X7+0.0389X8+0.0174X9-0.0056X10-0.0016X11+ 0.0055X12+0.0319X13-0.0061X14-0.1695X15-0.1064X16+0.0132X17+0.02323X18-0.114X19+0.1546X20-1.288X21-0.00417X1 2 +0.00215X2 2 +0.000181X3 2 -0.00209X4 2 +0.0153X5 2 -0.00099X6 2 -0.0279X7 2 -0.01193X8 2 -0.00806X12 2 -0.00538X13 2 +0.000457X14 2 +0.0683X15 2 +0.0130X16 2 +0.00037X17 2 -0.001087X18 2 +0.040X19 2 -0.003668X20 2 +0.386X21 2

[0106] The model fit indices are: coefficient of determination Rsq = 91.35%, adjusted Rsq = 89.26%, and standard error S = 0.0474. 88.5% of the samples in the model set had a predicted-to-measured value ratio within the range of 90% to 110%, indicating a good model fit.

[0107] 3.3 Model Validation

[0108] The model was independently validated using 112 samples from the validation set. The molecular structure descriptors of each sample were substituted into the prediction model to obtain the F-prediction, which was then compared with the experimental F-prediction.

[0109] Table 2 below shows the calculation results of molecular structure descriptors for some monomer raw materials in the validation set.

[0110] Table 2. Calculation results of molecular structure descriptors for some monomer raw materials in the validation set.

[0111]

[0112] Table 3 below shows the ratios of the predicted to the measured relative correction factors for some monomer raw materials in the validation set.

[0113] Table 3. Ratios of predicted to measured relative correction factors for monomer raw materials in the validation set.

[0114]

[0115] Table 4 below shows the statistical results of the ratio intervals between the predicted and measured relative correction factors of some monomer raw materials in the validation set.

[0116] Table 4. Range of ratios between predicted and measured relative correction factors for monomer raw materials in the validation set.

[0117]

[0118] Validation results show that 69.64% of the samples had a F-ratio between predicted and observed values ​​that was in the range of 90%–110%, and a cumulative total of 92.85% of the samples had a ratio that was in the range of 80%–120%, indicating that the model has reliable predictive ability. A scatter plot comparing predicted and observed values ​​is shown below. Figure 1 The ratio distribution is shown in Figure 2 .

[0119] Example 2: Relative correction factor prediction for isobutyrate leaf ester

[0120] (1) Obtain the molecular structural formula of isobutyrate leaf ester (CAS: 41519-23-7):

[0121] .

[0122] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=1, X2=3, X3=3, X4=3, X5=1, X6=1, X7=0, X8=0, X9=0, X10=0, X11=0, X12=1, X13=1, X14=1, X15=0, X16=0, X17=0, X18=0, X19=0.25, X20=9.0, X21=0.71.

[0123] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.7357.

[0124] (4) The F of isobutyric acid leaf ester was obtained through experimental measurement. 实测 =0.7219.

[0125] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =101.9%, indicating good prediction accuracy.

[0126] Example 3: Relative correction factor prediction for 2-isopropyl-4-methylthiazole

[0127] (1) Obtain the molecular structural formula of 2-isopropyl-4-methylthiazole (CAS: 15679-13-7):

[0128] .

[0129] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=2, X2=2, X3=0, X4=3, X5=0, X6=1, X7=1, X8=0, X9=0, X10=0, X11=0, X12=1, X13=2, X14=4, X15=0, X16=1, X17=2, X18=5, X19=0.33, X20=6.50, X21=0.59.

[0130] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.6520.

[0131] (4) Through experimental measurement, the F of 2-isopropyl-4-methylthiazole was obtained. 实测 =0.6610.

[0132] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =98.6%, indicating good prediction accuracy.

[0133] Example 4: Prediction of the relative correction factor for β-pinene

[0134] (1) Obtain the molecular structural formula of β-pinene (CAS: 18172-67-3):

[0135] .

[0136] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=2, X2=2, X3=4, X4=2, X5=0, X6=1, X7=0, X8=0, X9=0, X10=0, X11=0, X12=0, X13=5, X14=4, X15=2, X16=0, X17=0, X18=7, X19=0.40, X20=10.0, X21=0.57.

[0137] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.9493.

[0138] (4) Through experimental measurement, the F of β-pinene was obtained. 实测 =0.9024.

[0139] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =105.2%, indicating good prediction accuracy.

[0140] Example 5: Prediction of the relative correction factor for valerate

[0141] (1) Obtain the molecular structural formula of valeric acid (CAS: 109-52-4):

[0142] .

[0143] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=1, X2=0, X3=3, X4=1, X5=1, X6=0, X7=0, X8=1, X9=0, X10=0, X11=0, X12=0, X13=0, X14=0, X15=0, X16=0, X17=0, X18=0, X19=0.14, X20=4.00, X21=0.42.

[0144] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.5592.

[0145] (4) Through experimental measurement, the F of valeric acid was obtained. 实测 =0.5678.

[0146] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =98.5%, indicating good prediction accuracy.

[0147] Example 6: Relative Correction Factor Prediction for Damascene

[0148] (1) Obtain the molecular structural formula of daumatenone (CAS: 23696-85-7):

[0149] .

[0150] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=4, X2=4, X3=1, X4=4, X5=1, X6=3, X7=0, X8=0, X9=0, X10=0, X11=0, X12=0, X13=4, X14=8, X15=1, X16=0, X17=4, X18=6, X19=0.21, X20=12.50, X21=0.79.

[0151] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.8876.

[0152] (4) Through experimental measurement, the F of damascenone was obtained. 实测 =0.8753.

[0153] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =101.4%, indicating good prediction accuracy.

[0154] Example 7: Relative Correction Factor Prediction for D-Limonene

[0155] (1) Obtain the molecular structural formula of D-limonene (CAS: 5989-27-5):

[0156] .

[0157] (2) Calculate the values ​​of its 21 molecular structure descriptors, which are: X1=2, X2=2, X3=4, X4=2, X5=0, X6=2, X7=0, X8=0, X9=0, X10=0, X11=0, X12=0, X13=2, X14=4, X15=1, X16=0, X17=0, X18=6, X19=0.40, X20=10.00, X21=0.57.

[0158] (3) Substitute the above molecular structure descriptor values ​​into the prediction model equation to calculate F. 预测 =0.9496.

[0159] (4) Through experimental measurement, the F of D-limonene was obtained. 实测 =0.9370.

[0160] (5) Evaluation of prediction accuracy, F 预测 / F 实测 =101.4%, indicating good prediction accuracy.

[0161] Example 8: Quantitative Detection of Monomer Raw Materials in Atomized Samples

[0162] Isovaleraldehyde was detected in a certain nebulized product sample. The quantitative detection steps for isovaleraldehyde are as follows:

[0163] (1) Sample pretreatment: Accurately weigh 0.0503 g (accurate to 0.0001 g) of the nebulized sample to be tested and place it in a 10 mL volumetric flask. Dilute to the mark with a known concentration of n-heptadecane (C17) internal standard solution (concentration of 500 mg / L, solvent isopropanol), and vortex for 2 min to mix thoroughly. Take an appropriate amount of the mixture into a gas chromatograph vial for testing.

[0164] (2) Obtaining chromatographic data: The processed samples were analyzed using gas chromatography-flame ionization detector (GC-FID). Chromatographic conditions: injection port temperature 250℃; split ratio 20:1; injection volume 0.1μL; carrier gas N2, flow rate 1.0mL / min; temperature program: 50℃ for 2min, increased to 180℃ at 10℃ / min and held for 5min, then increased to 250℃ at 10℃ / min and held for 5min; FID temperature 275℃, hydrogen 40mL / min, air 400mL / min, and nitrogen purge 30mL / min. Analysis yielded chromatograms, showing good separation between isovaleraldehyde and internal standard C17. The peak area of ​​isovaleraldehyde was Si=1928; the peak area of ​​internal standard C17 was S1=59542.

[0165] (3) Predicting the relative correction factor: After determining that the target compound is isovaleraldehyde (CAS: 590-86-3), based on its molecular structural formula (CH3)2CHCH2CHO, the values ​​of its 21 molecular structure descriptors were calculated as follows: X1=0, X2=2, X3=1, X4=2, X5=1, X6=0, X7=0, X8=0, X9=0, X10=0, X11=0, X12=0, X13=1, X14=1, X15=0, X16=0, X17=0, X18=0, X19=0.50, X20=4.50, X21=0.36 (molecular weight 86.13 / 238.41). The above molecular structure descriptors were substituted into the expression of the multiple quadratic regression model for calculation, and f was obtained. i =0.6288.

[0166] (4) Calculate the isovaleraldehyde content: Substitute the experimental data and predicted values ​​into the internal standard method calculation formula:

[0167]

[0168] Given C1 = 500 mg / L, sample mass m = 0.0503 g, and final volume V = 0.01 L.

[0169] Isovaleraldehyde content C i The calculation is as follows:

[0170]

[0171] (5) Result verification and accuracy assessment

[0172] To evaluate the accuracy of this quantitative method, a standard solution of known concentration was prepared using isovaleraldehyde standard (purity ≥98%), and the concentration was determined under the same chromatographic conditions. The measured content of isovaleraldehyde in the sample was 5.03 mg / g using the external standard method. The accuracy of the prediction method was calculated as: (5.12 / 5.03) × 100% = 102.0%.

[0173] The results show that the quantitative results obtained based on the prediction model of this invention are in high agreement with the measured results of the standard, with a relative error within ±5%, indicating good accuracy.

[0174] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for predicting the relative correction factor of monomer raw materials in atomized products, characterized in that, Includes the following steps: Obtain molecular structure information of the target monomer raw material; Based on the molecular structure information, the values ​​of multiple predefined molecular structure descriptors are determined; The numerical value of the molecular structure descriptor is substituted into a pre-established relative correction factor prediction model to obtain the predicted relative correction factor of the monomer raw material relative to the internal standard; wherein, the relative correction factor prediction model represents the functional relationship between the relative correction factor of the monomer raw material and the numerical value of its molecular structure descriptor.

2. The relative correction factor prediction method as described in claim 1, characterized in that, The construction of the relative correction factor prediction model includes the following steps: A training dataset was composed of multiple monomeric raw materials. Obtain the molecular structure formulas of each monomer raw material in the training dataset, and determine the values ​​of multiple predefined molecular structure descriptors therein; Obtain the detection values ​​of the relative correction factors for each monomer raw material in the training dataset; Using the numerical value of the molecular structure descriptor as the independent variable and the relative correction factor detection value of the monomer raw material as the dependent variable, regression analysis is performed to fit the relative correction factor prediction model.

3. The relative correction factor prediction method as described in claim 2, characterized in that, Includes one or more of the following characteristics: (a) The training dataset contains at least 200 monomeric raw materials; (b) The plurality of predefined molecular structure descriptors include at least one of the following types of descriptors: Carbon atom descriptors include at least one of C atom, CH group, CH2 group, and CH3 group; Functional group descriptors include at least one of the following: >C=O group, >C=C< group, N=C< group, number of -OH or -SH groups, >NH group, -NH2 group, >COC< or >CSC< group; Heteroatom descriptors, including N atoms; Branch structure descriptor, including at least one of carbon branches and the total number of atoms in the branches excluding hydrogen; Ring structure descriptors include at least one of the following: non-aromatic ring, aromatic ring, π-π conjugated bond, and total number of ring-forming atoms; Substituent position descriptor, including the ratio of the largest substituent position to the total number of atoms excluding H; Electronic structure parameter descriptor, including the O value of C-1 / 2; A molecular weight ratio descriptor, which includes the ratio of the molecular weight of the monomer raw material to the molecular weight of the internal standard.

4. The relative correction factor prediction method as described in claim 3, characterized in that, The predefined molecular structure descriptors specifically include the following 21 types: C atom, CH group, CH2 group, CH3 group, >C=O group, >C=C< group, N=C< group, -OH or -SH group, N atom, >NH group, -NH2 group, >COC< or >CSC< group, carbon branch, total number of atoms in the branch excluding hydrogen, non-aromatic ring, aromatic ring, π-π conjugated bond, total number of cyclic atoms, maximum position of substituent / total number of atoms excluding H, O value of C-1 / 2, and the ratio of the molecular weight of the monomer raw material to the molecular weight of the internal standard.

5. The relative correction factor prediction method as described in claim 4, characterized in that, The regression analysis is a multiple quadratic regression analysis, and the prediction model is a multiple quadratic regression model.

6. The relative correction factor prediction method according to any one of claims 2 to 5, characterized in that, The prediction model is validated using a validation dataset containing multiple monomer raw materials. The prediction relative correction factor is compared with the detected value to evaluate the prediction accuracy of the prediction model.

7. A method for quantitative detection of monomer raw materials in atomized products, characterized in that, Includes the following steps: S1. Add a known concentration of internal standard solution to the sample to be tested, and use gas chromatography to detect it, and obtain the chromatographic peak areas of the target monomer raw material and the internal standard. S2. Using the method described in any one of claims 1 to 6, predict the relative correction factor f of the target monomer raw material. i ; S3. Calculate the content C of the monomer raw material in the sample to be tested according to the following formula. i : Among them, S i Let C1 be the peak area of ​​component i, C1 be the concentration of the internal standard, S1 be the peak area of ​​the internal standard, and f be the peak area of ​​component i. i , where is the predicted relative correction factor obtained by the relative correction factor prediction model for component i, V is the sample volume, and m is the sample mass.

8. The quantitative detection method as described in claim 7, characterized in that, Includes one or more of the following characteristics: (a) The internal standard is any one of n-heptadecane, its homologues, or heterocyclic compounds; (b) The concentration of the internal standard is 400~600 mg / L; (c) The detector used in the gas chromatography is a flame ionization detector.

9. The quantitative detection method as described in claim 7, characterized in that, In step S1, the detection conditions include: the chromatographic column is a polar chromatographic column, the injection port temperature is 230~250℃, the detector temperature is 250~280℃, and a programmed temperature rise is used.

10. The quantitative detection method as described in claim 9, characterized in that, Includes one or more of the following characteristics: (a) The polar chromatographic column is a polyethylene glycol stationary phase column; (b) The temperature program is as follows: the initial column temperature is 40~60℃, held for 1~3 min, then increased to 170~190℃ at 8~12℃ / min, and then increased to 240~260℃ at 8~12℃ / min, held for 3~7 min.

Images