Method, device, equipment and medium for predicting smoke release amount of cigarette leaf group formula

Abstract

The invention provides a method, a device, equipment and a medium for predicting the smoke release amount of a cigarette leaf group formula. The method comprises the following steps: processing original tobacco leaves to obtain a tobacco leaf sample; acquiring spectral data of the tobacco leaf sample, and acquiring chemical components of a first preset type by adopting a prediction model of a tobacco near-infrared big data platform; sorting the spectral data of the tobacco leaf samples by adopting a K-S algorithm, and selecting a second preset number of tobacco leaf samples as a to-be-analyzed data set; establishing a tobacco leaf raw material conventional smoke index release amount prediction model based on the chemical components on the basis of the first preset type of chemical components and the to-be-analyzed data set; according to the technical scheme provided by the invention, the problem of predicting the smoke release amount of the tobacco leaf group formula after various tobacco leaves are randomly combined in a digital formula can be solved.

Description

Technical Field

[0001] This document relates to the technical field of smoke emission prediction, and in particular to a method, device, equipment and system for predicting the smoke emission of a tobacco leaf group formula. Background Art

[0002] The goals of cigarette product design are mainly sensory quality and conventional index release, including smoke tar, nicotine and CO. Tobacco raw materials are an important factor affecting the release of conventional smoke indicators of cigarette products. Rapid and accurate prediction of the release of conventional smoke indicators of tobacco raw materials is of great significance for the realization of digital design of cigarette products. At present, the current national standard detection methods for smoke tar, nicotine and CO are complex, have a long detection cycle, produce waste liquid, have a relatively high analysis cost, and the preparation of tobacco raw material sample cigarettes is complex.

[0003] Near-infrared spectroscopy analysis technology has been widely used in tobacco leaf research due to its many advantages, including chemical composition, origin and aroma characteristics. Existing research includes: establishing a mathematical model of tobacco leaf tar production and spectrum by directly scanning tobacco leaves; predicting the nicotine content in the total particulate matter of smoke, but the error is large (average relative deviation 8.89%); combining conventional test data and near-infrared spectral data to quickly determine the nicotine content in cigarette smoke; building a robust PLS smoke release prediction model to improve the adaptability to spectral changes; establishing a prediction model for nicotine, tar and moisture by scanning Cambridge filters; establishing a nicotine release correction model by scanning Cambridge filters with near-infrared after smoking with a smoking machine; and establishing a tar and nicotine correction model for the extract using near-infrared transmission spectroscopy.

[0004] The above research on predicting conventional smoke emission by near infrared analysis technology has the following shortcomings: it only predicts a single indicator or two indicators, the research is not systematic and comprehensive, and the practical application is limited; the use of a smoking machine to capture smoke does not fully utilize the advantages of near infrared rapid detection; and the problem of predicting smoke from leaf group formulas after random combination of various tobacco leaves in digital formulas is not solved. Therefore, this application provides a method for predicting the smoke emission of tobacco leaf group formulas to solve the above technical problems. Summary of the invention

[0005] The embodiments of the present invention provide a method, device, equipment and medium for predicting the smoke release of a cigarette leaf group formula, aiming to solve the above-mentioned problem.

[0006] The present invention provides a method for predicting the smoke release of a tobacco leaf group formula, comprising:

[0007] S1. obtaining original tobacco leaves, and processing the original tobacco leaves to obtain tobacco leaf samples;

[0008] S2. Using a spectrum collector to obtain spectral data of the tobacco leaf sample, and using a prediction model of a tobacco near-infrared big data platform to obtain a first preset type of chemical component;

[0009] S3, after sorting the spectral data of the tobacco leaf samples by using the KS algorithm, selecting a second preset number of tobacco leaf samples as the data set to be analyzed;

[0010] S4, establishing a prediction model for the release of conventional smoke indicators of tobacco leaf raw materials based on chemical components based on the first preset type of chemical components and the data set to be analyzed, wherein the conventional smoke indicators include: tar, nicotine and CO;

[0011] S5, establishing a smoke emission prediction model for a tobacco leaf group formula according to the conventional smoke emission prediction model for tobacco leaf raw materials based on chemical composition;

[0012] S6. Predicting the conventional smoke index release of the tobacco leaf group formula according to the tobacco leaf group formula smoke release prediction model.

[0013] The present invention provides a device for predicting the smoke release amount of a tobacco leaf group formula, comprising:

[0014] A sample acquisition module is used to acquire original tobacco leaves and process the original tobacco leaves to acquire tobacco leaf samples;

[0015] A chemical component acquisition module, which uses a spectrum collector to acquire spectral data of the tobacco leaf sample, and uses a prediction model of a tobacco near-infrared big data platform to acquire a first preset type of chemical component;

[0016] The module for the data set to be analyzed uses a KS algorithm to sort the spectral data of the tobacco leaf samples, and then selects a second preset number of tobacco leaf samples as the data set to be analyzed;

[0017] A tobacco leaf raw material release model module is used to establish a tobacco leaf raw material conventional smoke index release prediction model based on chemical components of the first preset type and the data set to be analyzed, wherein the conventional smoke indexes include: tar, nicotine and CO;

[0018] A leaf group formula release model module is used to establish a tobacco leaf group formula smoke release prediction model based on the conventional smoke index release prediction model of tobacco leaf raw materials based on chemical composition;

[0019] The model prediction module predicts the conventional smoke index release of the cigarette leaf group formula according to the cigarette leaf group formula smoke release prediction model.

[0020] An embodiment of the present invention provides an electronic device, including:

[0021] processor; and,

[0022] A memory arranged to store computer executable instructions, which when executed cause the processor to perform the steps of the method for predicting the smoke delivery of a cigarette leaf group blend as described above.

[0023] An embodiment of the present invention provides a storage medium for storing computer executable instructions, which, when executed, implement the steps of the method for predicting the smoke release of a cigarette leaf group blend as described above.

[0024] By adopting the embodiment of the present invention, firstly, the advantage of near-infrared rapid detection is utilized to directly scan the tobacco end sample, and the sample pre-treatment is simple; then, representative samples are selected based on the near-infrared spectra of 889 kinds of stock aged tobacco leaves, combined with the preference for the use of raw materials and the Kennard-Stone algorithm, which is referred to as the KS algorithm in the embodiment of the present invention, and a conventional smoke index release prediction model of tobacco leaf raw materials based on chemical composition is established, so as to realize the rapid and effective analysis of the near-infrared spectrum smoke release of large quantities of tobacco leaf raw materials; on this basis, a tobacco leaf group formula smoke release prediction model is established according to the conventional smoke index release prediction model of tobacco leaf raw materials based on chemical composition, and the conventional smoke index release of the tobacco leaf group formula is predicted according to the tobacco leaf group formula smoke release prediction model, thereby effectively solving the smoke prediction of the leaf group formula after the random combination of various tobacco leaves in the digital formula, and providing data support for the digital design of cigarettes. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0026] Figure 1 A flow chart of a method for predicting smoke release of a cigarette leaf group formula according to an embodiment of the present invention;

[0027] Figure 2 A schematic diagram of a device for predicting smoke release of a cigarette leaf group formula according to an embodiment of the present invention;

[0028] Figure 3 is a near infrared spectrum of a tobacco sample according to an embodiment of the present invention;

[0029] Figure 4 The principal component projection distribution diagram of 889 alcoholized tobacco slices and representative samples in the embodiment of the present invention;

[0030] Figure 5It is a linear regression diagram of the prediction model of tar release per unit of tobacco shreds burned according to an embodiment of the present invention;

[0031] Figure 6 It is a linear regression diagram of the prediction model of nicotine release per unit of tobacco shreds burned according to an embodiment of the present invention;

[0032] Figure 7 It is a linear regression diagram of the prediction model of CO release per unit of tobacco shred combustion according to an embodiment of the present invention;

[0033] Figure 8 is a residual graph of tar release amount according to an embodiment of the present invention;

[0034] Fig. 9 This is a residual graph of nicotine release in an embodiment of the present invention;

[0035] Fig.10 This is a residual diagram of CO release in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0037] Method Embodiment

[0038] According to an embodiment of the present invention, a method for predicting the smoke release of a cigarette leaf group formula is provided. Figure 1 The flowchart of the method for predicting the smoke release of a cigarette leaf group blend according to the embodiment of the invention is as follows: Figure 1 As shown, the method for predicting the smoke release of a cigarette leaf group formula in an embodiment of the present invention specifically includes:

[0039] S1. obtaining original tobacco leaves, and processing the original tobacco leaves to obtain tobacco leaf samples;

[0040] S1 specifically includes: collecting 889 tobacco leaf raw materials from the stock of Company A. Manually remove the tobacco stems from the extracted samples, dry them naturally, control the moisture content between 6% and 8%, crush the samples with a crusher, pass them through a 60-mesh sieve, put the samples into self-sealing bags, exhaust all the air, and seal them.

[0041] S2. Using a spectrum collector to obtain spectral data of the tobacco leaf sample, and using a prediction model of a tobacco near-infrared big data platform to obtain a first preset type of chemical component;

[0042] S2 specifically includes: collecting sample spectral data under the conditions of ambient temperature (22±2)℃ and relative humidity (60±5)%. Spectral acquisition instrument parameters: scanning range is 10 000cm -1 ~4 000cm -1 , resolution 8cm -1 ; The spectrum was scanned 64 times; the light spot of the integrating sphere fell within 2 / 3 of the center of the sampling cup, the near-infrared spectra of all samples were collected twice (with re-sampling each time), and the two spectra should pass the consistency test (spectral similarity greater than 0.9999), and the average spectrum was taken as the near-infrared spectrum of the sample.

[0043] Based on the near-infrared spectrum of tobacco leaves and the prediction model of the tobacco near-infrared big data platform, 70 chemical components of tobacco leaves were obtained, including conventional chemical components, dichloromethane extract, solanesol, neophytadiene, pH, inorganic anions and cations, polyphenols, non-volatile acids and higher fatty acids, amino acids, Amadori compounds, etc.

[0044] Figure 3 is a near infrared spectrum of a tobacco sample of an embodiment of the present invention, according to Figure 3 It can be seen that the near-infrared absorption spectrum of tobacco is between 4000 and 9000 cm -1 The regional signal is strong and the information is rich. Since the absorption peaks of CH, NH, OH, and CO bonds are different in different spectral regions, different spectral ranges are selected for different chemical components during modeling, and the modeling spectral region with high effective information is optimized. The interference of baseline drift and random noise is eliminated by spectral derivation and smoothing, so as to more carefully reflect the spectral characteristics of the sample and improve the spectral resolution and sensitivity.

[0045] S3, after sorting the spectral data of the tobacco leaf samples by using the KS algorithm, selecting a second preset number of tobacco leaf samples as the data set to be analyzed;

[0046] S3 specifically includes: The influence of the optimization of the calibration set samples of tobacco leaf raw materials on the conventional flue gas near-infrared model is mainly reflected in the adaptability, prediction performance and simplicity of the model. If the sample information of the calibration set samples is incomplete and the representativeness is insufficient, the prediction results of the model will be greatly biased, and in extreme cases, it will even be impossible to predict unknown samples. This paper uses the KS algorithm to screen representative samples of near-infrared spectra.

[0047] The near infrared spectra of 889 kinds of alcoholized tobacco leaves were collected, and the representativeness of the near infrared spectra was ranked using the KS algorithm. Combined with the representative ranking of the near infrared spectra, 150 samples were determined as representative samples for the final standard cigarettes. The principal component analysis of the near infrared spectra was performed, and the principal component projection distribution diagram of 889 tobacco leaf raw materials and representative samples is shown in Figure 4 It can be seen from the figure that the selected 150 representative samples basically cover the regional range of all samples, and the projection points are evenly distributed in the entire sample main component area, which can represent the main spectral characteristics and measurement range of 889 samples.

[0048] The representativeness of the near-infrared spectra was ranked in combination with the KS algorithm. According to the representativeness of the near-infrared spectra, the final calibration set samples, i.e., the single-material cigarette samples for standard cigarettes, were determined.

[0049] Formulate process quality requirements for each process of tobacco production for representative samples, process representative single-material cigarettes according to the requirements, and produce tobacco samples; the moisture and filling values ​​of all tobacco samples are required to meet the requirements for cigarette rolling. It is required to roll standard cigarettes on the same machine, with the design parameters of (30+54)mm×24.5mm and the length of the tipping paper of 36mm. Control the hardness of each cigarette sample to be equivalent to that of normal cigarettes (66.0%±8.0%), and keep the parameters of standard cigarette auxiliary materials consistent. The specific parameters are shown in Table 1 below.

[0050] Table 1 Parameters of standard cigarette auxiliary materials

[0051]

[0052] According to the requirements of GB / T16447-2004, GB / T16450-2004, GB / T23356-2009, GB / T19609-2004, GB / T23203.1-2013 and GB / T23355-2009, the CO, total particulate matter, tar, moisture and nicotine release of cigarette samples are captured and tested.

[0053] The calculation method for the average weight of net burned tobacco (m) of a standard cigarette is as follows:

[0054] Assuming that the average weight of a standard cigarette is m0g, and the average weight of the remaining auxiliary materials after removing the tobacco is m1g, the net content of tobacco in the standard cigarette is (m0-m1)g. According to the national standard requirements for routine smoke measurement, the actual burning length of a standard cigarette is 45mm, so the average weight of the net burned tobacco in a standard cigarette is m=(m0-m1)×45 / 54g.

[0055] The calculation method for the release of conventional components per unit mass of burned tobacco (Q) is as follows:

[0056] Assuming the measured value of the release of conventional components of a standard cigarette is Q1, the release of conventional components per unit mass of burned tobacco is:

[0057] Q=Q1 / m

[0058] Where Q1 is the measured value of tar, nicotine and CO in a standard cigarette.

[0059] S4, establishing a prediction model for the release of conventional smoke indicators of tobacco leaf raw materials based on chemical components based on the first preset type of chemical components and the data set to be analyzed, wherein the conventional smoke indicators include: tar, nicotine and CO;

[0060] The quantitative prediction model of the conventional emission of tobacco leaf raw materials based on chemical composition specifically includes: (1) screening of modeling chemical components. Through the addition method, 7 derived indicators (inorganic cations, inorganic anions, total polyphenols, total non-volatile acids, total higher fatty acids, total amino acids and total Amadori) are generated and merged with the first 70 indicators. According to the correlation between the chemical composition of tobacco leaves and the conventional indicators of flue gas, the variables with the strongest correlation are first introduced. In subsequent rounds, the method of gradually introducing chemical components is adopted based on the partial F test value constructed based on the regression square sum (threshold value higher than 3.84 is introduced, and lower than 2.71 is eliminated). First, check whether there are variables that need to be eliminated, and then check whether there are variables that can be introduced. When there are no variables that need to be eliminated or introduced, the screening ends. The chemical components involved in the quantitative prediction model of the conventional emission of tobacco leaf raw materials based on chemical composition are finally determined through screening as shown in Table 2.

[0061] Among them, the seven derivative indicators are as follows:

[0062] 1. Inorganic cation: K + , Ca 2+ ,Mg 2+ Total content;

[0063] 2. Inorganic anions: Cl - , SO4 2- ,PO4 3- Total content;

[0064] 3. Total amount of polyphenols: the sum of the contents of neochlorogenic acid, chlorogenic acid, cryptochlorogenic acid, scopoletin, and rutin;

[0065] 4. Total non-volatile acid content: the sum of oxalic acid, malonic acid, succinic acid, malic acid, citric acid and vanillic acid;

[0066] 5. Total amount of higher fatty acids: the sum of myristic acid, hexadecanoic acid, linoleic acid, oleic acid + linolenic acid, octadecanoic acid, and eicosanoic acid;

[0067] 6. Total amino acid content: the sum of aspartic acid, threonine, serine, asparagine, glutamic acid, glutamine, glycine, alanine, valine, cystine, methionine, isoleucine, leucine, tyrosine, phenylalanine, 4-aminobutyric acid, lysine, histidine, tryptophan, arginine and proline;

[0068] 7. Total Amadori: Glucosamine (Glu-An), 1-deoxy-1-L-aminobutyric acid-D-fructose (Fru-Amb), 1-deoxy-1-L-histidine-D-fructose (Fru-His), 1-deoxy-1-L-proline-D-fructose (Fru-Pro), 1-deoxy-1-L-valine-D-fructose (Fru-Val), 1-deoxy-1-L-threonine-D-fructose (Fru-Thr), 1-deoxy-1-L-glycine-D-fructose (Fru-Gly), 1-deoxy-1-L-alanine-D-fructose (Fru-Ala), 1-deoxy-1-L-asparagine-D-fructose The total content of 1-deoxy-1-L-aspartate-D-fructose (Fru-Asp), 1-deoxy-1-L-glutamine-D-fructose (Fru-Gln), 1-deoxy-1-L-glutamate-D-fructose (Fru-Glu), 1-deoxy-1-L-isoleucine-D-fructose (Fru-Ile), 1-deoxy-1-L-leucine-D-fructose (Fru-Leu), 1-deoxy-1-L-tyrosine-D-fructose (Fru-Tyr), 1-deoxy-1-L-phenylalanine-D-fructose (Fru-Phe), and 1-deoxy-1-L-tryptophan-D-fructose (Fru-Trp) was 1.

[0069] Table 2 Specific chemical components involved in the prediction models of various conventional flue gas indicators

[0070]

[0071]

[0072] (2) Construction of a conventional quantitative prediction model for flue gas based on chemical composition. In order to improve the accuracy and generalization ability of the model, an integrated machine learning method was used to construct the model, including random forest regression, support vector regression and partial least squares regression (PLS) modeling methods. The modeling parameters were optimized based on the principle of maximizing the linear regression correlation coefficient between the model prediction value and the actual value and minimizing the residual. In the random forest method (Treebagger), the regression tree was used as a single learner and 100 trees were set. This method constructed 100 differential models based on regression trees through sampling with replacement and variable perturbation, and the average value was used as the model output result. In the support vector regression (fitrsvm), the data was first center-normalized, and then a nonlinear regression model was established based on the Gaussian kernel function. Before using partial least squares regression, we first used the specific chemical components screened in the first step to center-normalize the data. Through the partial least squares regression method, a regression model was established, which can further mine the potential information in the data and improve the prediction performance.

[0073] (3) The average prediction value of the three methods is taken as the final prediction result.

[0074] The linear regression determination coefficient (R) of the predicted value and the actual value in the full modeling and 5-fold validation 2 ) See Table 2 and Figure 5 , Figure 6 , Figure 7 , the residual statistics are shown in Table 3 and Figure 8 , Fig. 9 and Fig.10 .

[0075] Table 2 Full modeling and 5-fold R 2 Statistics

[0076] index <![CDATA[Full Modeling R 2 > <![CDATA[50% discount 2 > tar 0.956 0.827 Nicotine 0.974 0.914 CO 0.952 0.809

[0077] Table 3 Statistics of absolute residuals between predicted values ​​and actual values ​​of flue gas routine full modeling

[0078] index average value SD Minimum Maximum ≤1.0 ≤1.5 ≤2.0 tar 0.50 0.46 0.00 3.00 89％ 97％ 98％ CO 0.38 0.34 0.00 1.84 95％ 99％ 100％ index average value SD Minimum Maximum ≤0.1 ≤0.3 ≤0.5 Nicotine 0.07 0.09 0.00 0.52 82％ 97％ 99％

[0079] According to Table 2 and Figure 5 , Figure 6 , Figure 7 The results showed that the R values ​​of the model predictions and actual values ​​of tar, nicotine and CO were significantly higher when the 5-fold validation was used. 2 The values ​​of R and B were 0.827, 0.914 and 0.809, respectively, indicating that the prediction models of the three conventional smoke indicators are relatively stable and reliable. 2 The values ​​can reach 0.956, 0.974 and 0.952 respectively, which can realize accurate prediction of three conventional flue gas indicators.

[0080] According to Table 3 and Figure 8 , Fig. 9 and Fig.10 The results showed that the mean residual values ​​of tar, nicotine and CO were 0.50, 0.07 and 0.38 respectively, the standard deviations of residual values ​​were 0.46, 0.09 and 0.34 respectively, and the maximum residual values ​​were 3.00, 0.52 and 1.84 respectively; the 100% sample residual value of tar was within 1.5 mg / cigarette, the 100% sample residual value of nicotine was within 0.5 mg / cigarette, and the 100% sample residual value of CO was within 1.5 mg / cigarette, meeting the requirements for subsequent formula design.

[0081] S5, establishing a smoke emission prediction model for a tobacco leaf group formula according to the conventional smoke emission prediction model for tobacco leaf raw materials based on chemical composition;

[0082] The above-mentioned conventional smoke index release prediction model for tobacco raw materials based on chemical composition can predict the smoke release of a single tobacco raw material. By linearly superimposing the tobacco raw materials contained in the cigarette leaf group formula to generate a cigarette leaf group formula smoke release prediction model, the smoke release of the cigarette leaf group formula can be predicted. Then, with the help of this prediction model, it can be quickly determined whether the smoke release of the digitally recommended leaf group formula is within a reasonable range.

[0083] Tar prediction values ​​for tobacco leaf group blends:

[0084] Predicted nicotine values ​​for tobacco leaf group blends:

[0085] CO prediction values ​​for tobacco leaf group blends:

[0086] Among them, A1, A2, A3, ..., A n The proportion of each tobacco leaf raw material in the leaf group formula, satisfying A1+A2+A3+…+A n =1; m0 is the average weight of a standard cigarette, m1 is the average weight of the remaining auxiliary materials after removing the tobacco, m = (m0-m1), m is the average weight of the net burned tobacco of a standard cigarette, y 焦油1 , …, y 焦油n Indicates the tar release per unit weight of each tobacco leaf raw material in the leaf group formula, y 烟碱1 , …, y 烟碱n Indicates the nicotine release per unit weight of each tobacco leaf raw material in the leaf group formula, y CO1 , …, y COn It represents the CO release per unit weight of each tobacco raw material in the leaf group formula, and n represents the total number of each tobacco raw material in the leaf group formula.

[0087] S6, predicting the conventional smoke index release of the cigarette leaf group formula according to the cigarette leaf group formula smoke release prediction model;

[0088] In order to verify the predictive effect of the model in leaf group formulas, we designed two leaf group formulas, both using standard cigarette auxiliary materials.

[0089] Two specific examples are now used to demonstrate the effectiveness of the prediction model for smoke emission of the tobacco leaf group formula in the embodiment of the present invention. Tables 4 and 5 are the leaf group formulas of the two specific examples and the conventional smoke prediction values ​​of each component single material cigarette. Using the conventional smoke prediction data of tobacco leaf raw materials, combined with the prediction model for smoke emission of the tobacco leaf group formula provided in the embodiment of the present invention, the conventional smoke prediction value of the leaf group formula under standard auxiliary materials can be obtained. The prediction value and the results of the determination by the industry standard method are shown in the following table.

[0090] According to Table 4, first, the smoke release amount of each tobacco raw material in the leaf group formula is obtained by the tobacco raw material smoke release prediction model provided in the embodiment of the present invention (Y-1-Y-20), and then, the tar, nicotine and CO predicted release amounts of the cigarette stick are obtained by the cigarette leaf group formula smoke release prediction model provided in the embodiment of the present invention, which are 9.64 mg / stick, 0.93 mg / stick and 10.95 mg / stick. Compared with the actual measured values ​​of the same industry standards, the relative deviations of the tar, nicotine and CO predicted release amounts of the cigarette sticks predicted by the cigarette leaf group formula smoke release prediction model provided in the embodiment of the present invention are 2.44%, 4.49% and 2.53%, respectively.

[0091] According to Table 5, the smoke release of each tobacco raw material in the leaf group formula (PY-1-PY-16) is first obtained by the tobacco raw material smoke release prediction model provided in the embodiment of the present invention, and then the tar, nicotine and CO predicted release of the cigarette stick are obtained by the tobacco leaf group formula smoke release prediction model provided in the embodiment of the present invention, which are 8.77 mg / stick, 0.86 mg / stick and 10.31 mg / stick. Compared with the actual measured values ​​of the same industry standard, the relative deviations of the tar, nicotine and CO predicted release of the cigarette stick predicted by the tobacco leaf group formula smoke release prediction model provided in the embodiment of the present invention are -2.12%, -3.37% and -2.09% respectively.

[0092] In summary, the prediction results of the conventional smoke emission of the tobacco leaf group formula smoke emission prediction model provided by the embodiment of the present invention are basically consistent with the national standard test method, with an error of <5%, which verifies the accuracy and reliability of the prediction model.

[0093] Table 4 Leaf group formula of specific example 1 and conventional flue gas prediction value of each component single material smoke

[0094]

[0095]

[0096] Table 5 Leaf group formula of specific example 2 and conventional flue gas prediction value of each component single material smoke

[0097]

[0098] By adopting the embodiments of the present invention, the following beneficial effects are achieved:

[0099] By adopting the embodiment of the present invention, firstly, the advantage of near-infrared rapid detection is utilized to directly scan the tobacco dust sample, and the sample pre-treatment is simple; then, based on the near-infrared spectra of 889 kinds of aged tobacco leaves in stock, representative samples are selected in combination with the preference for the use of raw materials and the KS algorithm, and a smoke tar, nicotine and CO release prediction model based on the chemical composition of tobacco leaf raw materials is established, so as to realize the rapid and effective analysis of the near-infrared spectrum smoke release of a large number of tobacco leaf raw materials; on this basis, a smoke release prediction model for a cigarette leaf group formula is established according to the conventional smoke index release prediction model for tobacco leaf raw materials, and the conventional smoke index release of the cigarette leaf group formula is predicted according to the smoke release prediction model for the cigarette leaf group formula, thereby effectively solving the smoke prediction of the leaf group formula after the random combination of various tobacco leaves in the digital formula, and providing data support for the digital design of cigarettes.

[0100] Device Example 1

[0101] According to an embodiment of the present invention, a device for predicting the smoke release of a cigarette leaf group formula is provided. Figure 2 Schematic diagram of a device for predicting smoke release of a tobacco leaf group blend according to an embodiment of the invention. Figure 2 As shown, the device for predicting the smoke release amount of a cigarette leaf group formula according to the embodiment of the present invention specifically comprises:

[0102] The sample acquisition module 20 acquires raw tobacco leaves and processes the raw tobacco leaves to acquire tobacco leaf samples;

[0103] The chemical component acquisition module 21 uses a spectrum collector to acquire spectral data of the tobacco leaf sample, and uses a prediction model of a tobacco near-infrared big data platform to acquire chemical components of a first preset type;

[0104] The to-be-analyzed data set module 22 selects a second preset number of tobacco leaf samples as the to-be-analyzed data set after sorting the spectral data of the tobacco leaf samples using the KS algorithm;

[0105] The tobacco leaf raw material release model module 23 establishes a tobacco leaf raw material conventional smoke index release prediction model based on chemical components based on the first preset type of chemical components and the data set to be analyzed, wherein the conventional smoke indexes include: tar, nicotine and CO;

[0106] The leaf group formula release model module 24 establishes a tobacco leaf group formula smoke release prediction model based on the conventional smoke index release prediction model of the tobacco leaf raw material based on chemical composition;

[0107] The model prediction module 25 predicts the conventional smoke index release of the tobacco leaf group formula according to the tobacco leaf group formula smoke release prediction model.

[0108] Device Example 2

[0109] According to an embodiment of the present invention, there is provided an electronic device, including:

[0110] processor; and,

[0111] A memory is arranged to store computer executable instructions, which when executed cause the processor to perform the steps of the above method embodiments.

[0112] Device Example 3

[0113] According to an embodiment of the present invention, a storage medium is provided for storing computer executable instructions, wherein the computer executable instructions implement the steps of the method embodiment when executed.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting the smoke release of a tobacco leaf group blend, characterized in that include: S1. obtaining original tobacco leaves, and processing the original tobacco leaves to obtain tobacco leaf samples; S2. Using a spectrum collector to obtain spectral data of the tobacco leaf sample, and using a prediction model of a tobacco near-infrared big data platform to obtain a first preset type of chemical component; S3, after sorting the spectral data of the tobacco leaf samples by using the KS algorithm, selecting a second preset number of tobacco leaf samples as the data set to be analyzed; S4, establishing a prediction model for the release of conventional smoke indicators of tobacco leaf raw materials based on chemical components based on the first preset type of chemical components and the data set to be analyzed, wherein the conventional smoke indicators include: tar, nicotine and CO; S5, establishing a smoke emission prediction model for a tobacco leaf group formula according to the conventional smoke emission prediction model for tobacco leaf raw materials based on chemical composition; S6. Predicting the conventional smoke index release of the tobacco leaf group formula according to the tobacco leaf group formula smoke release prediction model.

2. The method according to claim 1, characterized in that The S4 specifically includes: Obtaining 7 derived indicators by a summation method, and combining the 7 derived indicators with the chemical components of the first preset type to obtain the chemical indicators to be modeled; According to the correlation between the chemical index to be modeled and the conventional flue gas index, the chemical components in the chemical index to be modeled that have the strongest correlation with the conventional flue gas index are first introduced, and then the chemical components in the chemical index to be modeled are introduced by screening based on the partial F test value constructed based on the regression square sum. The specific chemical components in the prediction model of each conventional flue gas index are finally determined through screening.

3. The method according to claim 1, characterized in that: The conventional smoke index release prediction model of tobacco raw materials based on chemical composition adopts an integrated machine learning method to construct the model, and the modeling parameters are optimized based on the principle of maximizing the linear regression correlation coefficient between the model prediction value and the actual value and minimizing the residual. The integrated machine learning method adopted includes: random forest regression, support vector regression and partial least squares regression modeling methods, and finally the average value of the three methods of random forest regression, support vector regression and partial least squares regression modeling is taken as the prediction result.

4. The method according to claim 1, characterized in that The tobacco leaf group formula smoke release prediction model specifically includes: The tar release prediction model is obtained by formula 1: The nicotine release prediction model is obtained by formula 2: The CO release prediction model is obtained by formula 3: Among them, A1, A2, A3, ..., A n The proportion of each tobacco leaf raw material in the leaf group formula, satisfying A1+A2+A3+…+A n =1; m0 is the average weight of a standard cigarette, m1 is the average weight of the remaining auxiliary materials after removing the tobacco, m = (m0-m1), m is the average weight of the net burned tobacco of a standard cigarette, y 焦油1 , …, y 焦油n Indicates the tar release per unit weight of each tobacco leaf raw material in the leaf group formula, y 烟碱1 , …, y 烟碱n Indicates the nicotine release per unit weight of each tobacco leaf raw material in the leaf group formula, y CO1 , …, y COn It represents the CO release per unit weight of each tobacco raw material in the leaf group formula, and n represents the total number of each tobacco raw material in the leaf group formula.

5. A device for predicting the smoke release of a tobacco leaf group, characterized in that include: A sample acquisition module is used to acquire original tobacco leaves and process the original tobacco leaves to acquire tobacco leaf samples; A chemical component acquisition module, which uses a spectrum collector to acquire spectral data of the tobacco leaf sample, and uses a prediction model of a tobacco near-infrared big data platform to acquire a first preset type of chemical component; The module for the data set to be analyzed uses a KS algorithm to sort the spectral data of the tobacco leaf samples, and then selects a second preset number of tobacco leaf samples as the data set to be analyzed; A tobacco leaf raw material release model module is used to establish a tobacco leaf raw material conventional smoke index release prediction model based on chemical components of the first preset type and the data set to be analyzed, wherein the conventional smoke indexes include: tar, nicotine and CO; A leaf group formula release model module is used to establish a tobacco leaf group formula smoke release prediction model based on the conventional smoke index release prediction model of tobacco leaf raw materials based on chemical composition; The model prediction module predicts the conventional smoke index release of the cigarette leaf group formula according to the cigarette leaf group formula smoke release prediction model.

6. The device according to claim 5, characterized in that The tobacco leaf raw material release amount model module is specifically used for: Obtaining 7 derived indicators by a summation method, and combining the 7 derived indicators with the chemical components of the first preset type to obtain the chemical indicators to be modeled; According to the correlation between the chemical index to be modeled and the conventional flue gas index, the chemical components in the chemical index to be modeled that have the strongest correlation with the conventional flue gas index are first introduced, and then the chemical components in the chemical index to be modeled are introduced by screening based on the partial F test value constructed based on the regression square sum. The specific chemical components in the prediction model of each conventional flue gas index are finally determined through screening.

7. The device according to claim 5, characterized in that The conventional smoke index release prediction model of tobacco raw materials adopts an integrated machine learning method to construct the model, and the modeling parameters are optimized based on the principle of maximizing the linear regression correlation coefficient between the model prediction value and the actual value and minimizing the residual. The integrated machine learning method adopted includes: random forest regression, support vector regression and partial least squares regression modeling methods.

8. The device according to claim 5, characterized in that The tobacco leaf group formula smoke release prediction model specifically includes: The tar release prediction model is obtained by formula 1: The nicotine release prediction model is obtained by formula 2: The CO release prediction model is obtained by formula 3: Among them, A1, A2, A3, ..., A n The proportion of each tobacco leaf raw material in the leaf group formula, satisfying A1+A2+A3+…+A n =1; m0 is the average weight of a standard cigarette, m1 is the average weight of the remaining auxiliary materials after removing the tobacco, m = (m0-m1), m is the average weight of the net burned tobacco of a standard cigarette, y 焦油1 , …, y 焦油n Indicates the tar release per unit weight of each tobacco leaf raw material in the leaf group formula, y 烟碱1 , …, y 烟碱n Indicates the nicotine release per unit weight of each tobacco leaf raw material in the leaf group formula, y CO1 , …, y COn It represents the CO release per unit weight of each tobacco raw material in the leaf group formula, and n represents the total number of each tobacco raw material in the leaf group formula.

9. An electronic device, comprising: processor; as well as, A memory arranged to store computer executable instructions, which when executed cause the processor to perform the steps of the method for predicting the smoke delivery of a cigarette leaf group blend as claimed in any one of claims 1 to 4.

10. A storage medium for storing computer executable instructions, wherein the computer executable instructions, when executed, implement the steps of the method for predicting the smoke release of a cigarette leaf group blend as claimed in any one of claims 1 to 4.

Images