Tobacco leaf sweetness characteristic typicality digital characterization method

Abstract

The invention discloses a tobacco leaf sweetness characteristic typicality digital characterization method, which comprises the following steps: S1, respectively collecting tobacco leaf samples with different sweetness styles, and measuring sweetness scores of the tobacco leaf samples; s2, tobacco leaf carbohydrate substances are extracted; s3, dissolving tobacco leaf carbohydrates, and carrying out thermal cracking-gas chromatography-mass spectrometry analysis to obtain a carbohydrate cracking product map; s4, adopting a machine learning method to establish a quantitative evaluation model between the tobacco leaf carbohydrate splitting product atlas and the sweetness score; and S5, carrying out typicality digital characterization on the sweetness characteristics of the tobacco leaf sample through the quantitative evaluation model. The method not only provides an objective and accurate means for the sweetness availability of the tobacco leaves, but also realizes digital description, overcomes the subjectivity and uncertainty of judgment mainly depending on subjective cognition of smoke panel test personnel in the prior art, and lays a foundation for digital use of tobacco leaf raw materials.

Description

Technical Field

[0001] The invention relates to a method for digitally representing typical sweetness characteristics of tobacco leaves, and belongs to the technical field of tobacco leaf sweetness evaluation. Background Art

[0002] Tobacco leaves, as the primary part of the tobacco plant, are the basic raw material for various tobacco products. Tobacco belongs to the genus Nicotiana in the Solanaceae family, and its scientific name is Nicotiana tabacum. Due to its wide range of cultivation, tobacco leaves from different producing areas vary in chemical composition due to differences in climate, soil conditions, and cultivation practices. This affects the usability of the leaves and results in different smoking characteristics.

[0003] Sugars are one of the main chemical components of tobacco leaves, and are generally the most abundant. Sugars in tobacco leaves primarily include small, water-soluble sugars such as glucose, fructose, and sucrose, as well as polysaccharides such as starch, cellulose, and pectin. The term "total sugars" in tobacco refers to water-soluble sugars. The composition and content of sugars in tobacco leaves significantly influence the sweetness and usability of the leaves. However, to date, it has been difficult to quantitatively describe the sugar content and usability of different tobacco styles and types. The total sugar and reducing sugar content of tobacco leaves are typically measured to characterize some of the leaf's quality, but this differs from the final usability of the leaves. Consequently, inferior tobacco leaves are often passed off as genuine, causing unstable product quality and direct economic losses for cigarette companies. Traditional sensory evaluation methods struggle to objectively and scientifically assess the quality characteristics of tobacco raw materials on a large scale.

[0004] Furthermore, many methods have traditionally been used to extract tobacco leaf sugar components, but these extraction and separation methods present numerous challenges. Extraction of tobacco leaf sugar components using leaching methods requires high extraction temperatures, takes a long time, and is inefficient. Microwave-assisted extraction of tobacco leaf sugar components can more effectively dissolve substances compared to leaching methods. However, this technique suffers from microwave selectivity, allowing more polar substances to receive more microwave energy, making it difficult to dissolve less polar substances. Enzyme-assisted extraction of tobacco leaf sugar components is affected by factors such as temperature, pH, substrate concentration, enzyme concentration, and inhibitors. Under these conditions, adding too much or too little enzyme can prolong enzymatic hydrolysis time or result in enzyme waste. Ultrahigh pressure extraction (UHP) yields higher extraction rates for tobacco leaf sugar components than other extraction methods, but it can also lead to the dissolution of other impurities, which can affect the purity of tobacco leaf sugar components. Summary of the Invention

[0005] Based on the above, the present invention provides a method for typical digital characterization of the sweetness characteristics of tobacco leaves, establishes a quantitative evaluation model based on the spectrum of tobacco leaf sugar substance cracking products and the sweetness score, and performs typical digital characterization of the sweetness characteristics of tobacco leaves according to the quantitative evaluation model to overcome the shortcomings of the existing technology.

[0006] The technical solution of the present invention is: a method for digitally characterizing the typical sweetness characteristics of tobacco leaves, comprising:

[0007] S1 collected tobacco samples with different sweetness styles and measured their sweetness scores;

[0008] S2 extracts sugars from tobacco leaves;

[0009] S3 dissolves the sugar substances in the tobacco leaves and performs thermal cracking-gas chromatography-mass spectrometry analysis to obtain a spectrum of the sugar substance cracking products;

[0010] S4 uses machine learning methods to establish a quantitative evaluation model between the spectrum of tobacco leaf sugar pyrolysis products and sweetness scores;

[0011] S5 uses a quantitative evaluation model to perform a typical digital characterization of the sweetness characteristics of tobacco samples.

[0012] Preferably, in step S2, tobacco sugars are extracted using a low co-solvent, and the specific method is as follows:

[0013] Choline chloride and 1,2-propylene glycol were mixed at a molar ratio of 1:2, mixed evenly at 80° C., and heated until the solution became transparent to obtain a deep eutectic solvent solution;

[0014] Weigh the tobacco leaf sample into a centrifuge tube, add a low eutectic solvent solution at a solid-liquid ratio of 1:40, and perform ultrasonic-assisted extraction for 60 minutes. The extract is centrifuged at 4°C for 10 minutes (8000 rpm). The filtrate is filtered, a small amount of co-crystallizing agent is added, and vacuum freeze-drying is performed to obtain solid tobacco leaf sugar substances.

[0015] Preferably, in step S3, the specific method of pyrolysis-gas chromatography-mass spectrometry analysis is as follows:

[0016] The tobacco sugars obtained in step S2 were placed in a pyrolysis sample boat, which was then placed in a pyrolysis reactor. GC-MS was then started, with a pyrolysis temperature of 500°C and a residence time of 0.2 min.

[0017] GC-MS conditions: A HP-5MS gas chromatography column (60 m × 250 μm × 0.25 μm) was used, high-purity helium (99.999%) was used as the carrier gas, the column flow rate was 1.0 ml / min, the injection volume was 1 μL, and the pyrolysis volatile products were injected with a split ratio of 10:1. The GC-MS interface temperature was maintained at 280°C, and the heating program was as follows: initial temperature 40°C, hold for 2 min, increase to 140°C at a rate of 5°C / min, hold for 2 min, then increase to 280°C at a rate of 5°C / min, hold for 5 min.

[0018] Mass spectrometer settings: electron impact ionization source, electron energy 70 eV, ion source temperature 230 °C, quadrupole temperature 230 °C, scan range 30-550 amu.

[0019] Preferably, in step S4, the carbohydrate decomposition product spectrum is divided into a model training set and a model test set in a ratio of 7:3, and the training set is used to establish a quantitative evaluation model between the tobacco leaf carbohydrate decomposition product spectrum and the sweetness score using the BP artificial neural network method, and the effectiveness of the quantitative evaluation model is tested.

[0020] Preferably, 70% of the carbohydrate pyrolysis product spectra are extracted according to the KS algorithm as the model test set, and the remaining samples are used as the model training set.

[0021] Preferably, the quantitative evaluation model is tested for effectiveness, specifically including:

[0022] For the model training set, the self-validation method and cross-validation method were used respectively, and the atlas data of each sample was substituted into the quantitative evaluation model for validation and cross-validation to check the error of the model;

[0023] For the model test set, the atlas data of each sample was substituted into the quantitative evaluation model to analyze its sweetness score. The results obtained by the quantitative evaluation model were compared with the actual scores of the samples to verify the error of the model.

[0024] Beneficial effects of the present invention: The present invention provides a typical digital characterization method for the sweetness characteristics of tobacco leaves. First, the sweetness score of the sample is measured, and then the sample is subjected to sugar extraction to eliminate interference from other substances. Then, pyrolysis-gas chromatography-mass spectrometry analysis is performed to obtain a spectrum of sugar pyrolysis products. A quantitative evaluation model is then established based on the spectrum and the sweetness score. The typical digital characterization of the sweetness characteristics of tobacco leaves is performed according to the parameters in the quantitative evaluation model. This not only provides an objective and accurate means for the sweetness availability of tobacco leaves, but also realizes digital description, overcoming the subjectivity and uncertainty of the prior art that mainly relies on the subjective cognition of smoking evaluators for judgment, and laying the foundation for the digital use of tobacco leaf raw materials. In addition, the present invention uses a low eutectic solvent to extract sugar substances, which can effectively eliminate interference from other substances. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flow chart of the method for digitally characterizing the typical sweetness characteristics of tobacco leaves;

[0026] Figure 2 It is the BP neural network topology;

[0027] Figure 3 is the total ion current (TIC) of the pyrolysis products at different pyrolysis residence times;

[0028] Figure 4 This is a comparison chart of the predicted sweetness score and the actual sweetness score. DETAILED DESCRIPTION

[0029] To make the above-mentioned objects, features, and advantages of the present invention more readily apparent, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings. The following description sets forth numerous specific details to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways than those described herein, and those skilled in the art may make similar modifications without departing from the scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0030] refer to Figure 1 In this embodiment, a method for digitally characterizing the typical sweetness characteristics of tobacco leaves includes:

[0031] S1 collected tobacco samples with different sweetness styles and measured their sweetness scores;

[0032] The sensory evaluation method is used to have professional smoking testers evaluate and score the sweetness of tobacco samples.

[0033] S2 extracts sugars from tobacco leaves;

[0034] The sugar substances in tobacco leaves are extracted by low co-solvents. The specific method is as follows:

[0035] Choline chloride and 1,2-propylene glycol were mixed at a molar ratio of 1:2, mixed evenly at 80° C., and heated until the solution became transparent to obtain a deep eutectic solvent solution;

[0036] Weigh the tobacco leaf sample into a centrifuge tube, add a low eutectic solvent solution at a solid-liquid ratio of 1:40, and perform ultrasonic-assisted extraction for 60 minutes. The extract is centrifuged at 4°C for 10 minutes (8000 rpm). The filtrate is filtered, a small amount of co-crystallizing agent is added, and vacuum freeze-drying is performed to obtain solid tobacco leaf sugar substances.

[0037] S3 dissolves the sugar substances in the tobacco leaves and performs thermal cracking-gas chromatography-mass spectrometry analysis to obtain a spectrum of the sugar substance cracking products;

[0038] The specific method is as follows:

[0039] Weigh 1 mg of the solid tobacco sugars from step S2 into a pyrolysis sample boat. Stuff the sample boat with quartz wool to prevent spillage when the crucible falls freely into the pyrolysis furnace. Place the sample boat in the pyrolysis chamber and start the GC-MS at 500°C for a residence time of 0.2 min.

[0040] GC-MS conditions: A HP-5MS gas chromatography column (60 m × 250 μm × 0.25 μm) was used, with high-purity helium (99.999%) as the carrier gas. The column flow rate was 1.0 ml / min, and the injection volume was 1 μL. The pyrolysis volatile products were injected with a split ratio of 10:1. The GC-MS interface temperature was maintained at 280°C. The temperature program was as follows: initial temperature at 40°C, hold for 2 min, increase to 140°C at a rate of 5°C / min, hold for 2 min, then increase to 280°C at a rate of 5°C / min, hold for 5 min.

[0041] Mass spectrometer settings: electron impact ionization source, electron energy 70 eV, ion source temperature 230 °C, quadrupole temperature 230 °C, scan range 30-550 amu.

[0042] Obtain the thermal decomposition product spectrum of the sample.

[0043] S4 uses machine learning methods to establish a quantitative evaluation model between the spectrum of tobacco leaf sugar pyrolysis products and sweetness scores;

[0044] The specific method is as follows:

[0045] First, the saccharide cleavage product spectra were normalized using the z-score method for all sample data;

[0046] Secondly, 70% of the carbohydrate pyrolysis product spectra were extracted according to the KS algorithm as the model test set, and the remaining samples were used as the model training set;

[0047] Then, using the training set, the BP artificial neural network method was used to establish a quantitative evaluation model between the spectrum of tobacco leaf sugar decomposition products and the sweetness score, and the effectiveness of the quantitative evaluation model was tested. The BP neural network has the ability of self-learning and self-organizing nonlinear mapping, and is suitable for modeling problems with unclear knowledge background, complex information, and unclear reasoning rules. The BP neural network is a multi-layer feedforward neural network. The main feature of this network is that the signal is transmitted forward and the error is propagated backward. In the forward transmission, the input signal is processed layer by layer from the input layer through the hidden layer to the output layer. The state value of the neurons in each layer affects the state of the neurons in the next layer. If the output layer does not obtain the expected output, it switches to back propagation and adjusts the network weights and thresholds according to the prediction error, so that the predicted output of the BP neural network continues to approach the expected output.

[0048] The topological structure of BP neural network is as follows Figure 2 As shown in the figure, X1, X2, ..., X n Is the input value of BP neural network (X1=[x1,x2,…,x n ]), Y1, Y2, …, Y m is the predicted value of BP neural network (Y1=[y1,y2,…,y m ]),ω ij and ω jk are the network weights between the input layer and hidden layer neurons and between the hidden layer and output layer neurons, respectively. The thresholds of the hidden layer and output layer neurons are θ j and θ k As can be seen from the figure, the BP neural network can be regarded as a nonlinear function. The network receives data X, calculates and passes it to the hidden layer through the network weight and threshold, and then transforms it through the activation function and passes it to the output layer to obtain the actual output z (that is, the predicted value Y). The output of the hidden layer neurons and the output layer neurons is:

[0049]

[0050] Then, the network calculates the error by comparing the output value Z with the expected value E. The error information is passed back layer by layer based on the error feedback, and the error contribution of each layer of neurons is calculated. Then, the gradient descent method is used to iteratively adjust the weights and thresholds of each layer of neurons to reduce them to an acceptable range. The relevant calculations are as follows:

[0051] The error between the network's output and the expected value is:

[0052]

[0053] The error E affects the weight ω between the hidden layer and the output layer. ki The partial derivative of is:

[0054]

[0055] The error E affects the weight ω between the hidden layer and the input layer. ij The partial derivative of is:

[0056]

[0057] The modified formula of weight obtained from the above two formulas is:

[0058]

[0059] Where η 1 and η 2 is the learning step size of the hidden layer and the output layer.

[0060] Similarly, the threshold correction formula is:

[0061]

[0062] For the model training set, the spectral data of carbohydrate cleavage products and the sweetness score data were substituted into the input and output layers, respectively. Gradient descent was then used to iteratively adjust the weights and thresholds of each neuron layer to an acceptable range, thus generating a quantitative evaluation model. For the model test set, the spectral data of each sample was substituted into the quantitative evaluation model to calculate its sweetness score. The results obtained by the quantitative evaluation model were compared with the actual scores of the samples to verify the model's effectiveness.

[0063] The following is a detailed description of the research process of extracting sugars from tobacco leaves and analyzing the pyrolysis products of sugars in the present invention:

[0064] 1. Extraction of tobacco sugars using low co-solvents

[0065] Choline chloride and betaine were used as hydrogen bond acceptors (HBAs), and citric acid, oxalic acid, lactic acid, 1,2-propylene glycol, 1,4-butanediol, formic acid, acetic acid, and ethylene glycol were used as hydrogen bond donors (HBDs) in a molar ratio of 1:2. The mixture was heated in hot water at 80°C until completely dissolved and a homogeneous transparent DES solution was formed. The detailed information of the DESs is shown in Table 1.

[0066] Table 1 Detailed information of DES

[0067]

[0068]

[0069] The effects of the above 10 low eutectic solvents in extracting tobacco leaf sugar components were compared. The results showed that the highest extraction rate was 23.1% for group 2 choline chloride-1,2-propylene glycol.

[0070] 2. Pyrolysis-Gas Chromatography-Mass Spectrometry Analysis

[0071] 1. Selection of thermal cracking temperature

[0072] In the thermal pyrolysis analysis of sample sugar components, the choice of pyrolysis temperature has a significant impact on the research. Excessively high temperatures will cause the sugar components to break down into smaller molecules, while prolonged pyrolysis times may lead to secondary reactions within the furnace, affecting the authenticity of the experimental results. Excessively low temperatures will make it difficult to break down the larger molecules in the sugar components, resulting in fewer products and a poor reflection of the original sample's composition.

[0073] Referring to samples of the same type, 400℃ is a critical temperature. When the temperature is lower than 400℃, only smaller molecular substances in the sugar component are cracked. When the temperature is higher than 400℃, complex substances in the sugar component begin to decompose. Therefore, 400℃, 500℃ and 600℃ are selected for thermal cracking temperature.

[0074] In this study, pyrolysis was performed at 400°C, 500°C, and 600°C for 0.2 min under the aforementioned instrumental operating conditions. The products were analyzed to identify the optimal temperature. Pyrolysis products at different temperatures were separated and identified using Py-GC-MS, and peak counts were calculated.

[0075] Table 2 Number of pyrolysis product peaks at different temperatures

[0076]

[0077] There are 58 peaks at 500°C and 77 peaks at 600°C. Because sugars continue to undergo intramolecular decomposition reactions such as dehydration at temperatures above 500°C, producing a large amount of complex substances that affect the original state of the sample, 500°C was selected as the optimal reaction temperature for thermal decomposition.

[0078] 2. Optimize residence time

[0079] If the thermal cracking time is too long, it may cause secondary reaction of the substance in the furnace, making the thermal cracking effect unrealistic; if the thermal cracking time is too short, it may lead to insufficient thermal cracking of the sample, which cannot truly reflect the material composition of the original sample. In order to select the appropriate thermal cracking residence time, this experiment selected 0.1min, 0.2min, and 0.3min thermal cracking times, and the cracking temperature was 500℃ for the experiment. The chromatograms were compared, and the peaks were almost overlapping, but it was still found that the peaks of the 0.2min chromatogram were sharper and the separation was better than those of the 0.1min and 0.3min chromatograms. Figure 3 Peaks 1 and 2 are detected, so a retention time of 0.2 min is selected.

[0080] 3. Spectrum of carbohydrate pyrolysis products

[0081] Pyrolysis product analysis was performed on sugar extracts from multiple samples of two varieties, Yun 87 and Xiangyan No. 7. Chromatographic peak area data for 105 pyrolysis products were obtained for each sample. These samples were then subjected to sensory evaluation and their sweetness scores were recorded.

[0082] Multiple samples were divided into several training sets and several test sets according to the KS algorithm. The chromatographic peak area of the pyrolysis product of the training set samples was input as the X value into the BP neural network input layer, and the sensory sweetness score was used as the output layer. The network training was carried out with the goal of minimizing the prediction error through interactive verification to obtain the BP neural network model. The chromatographic peak area of the pyrolysis product of the test set samples was input as the X value into the established BP neural network model, and its sweetness score was calculated and compared with the actual sensory sweetness score to verify the prediction accuracy of the model. The results are as follows: Figure 4 As shown, the root mean square error of the prediction is: 0.0287.

[0083] The above-described embodiment merely represents one embodiment of the present invention. While the description is relatively specific and detailed, it should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art may make various modifications and improvements without departing from the spirit of the present invention, and these modifications and improvements fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for digitally characterizing the typicality of tobacco leaf sweetness characteristics, characterized in that: include: S1 collected tobacco samples with different sweetness styles and measured their sweetness scores; S2 extracts sugars from tobacco leaves; S3 dissolves the sugar substances in the tobacco leaves and performs thermal cracking-gas chromatography-mass spectrometry analysis to obtain a spectrum of the sugar substance cracking products; S4 uses machine learning methods to establish a quantitative evaluation model between the spectrum of tobacco leaf sugar pyrolysis products and sweetness scores; S5 uses a quantitative evaluation model to perform a typical digital characterization of the sweetness characteristics of tobacco samples.

2. The method for digitally characterizing the typicality of tobacco leaf sweetness characteristics according to claim 1, characterized in that: In step S2, tobacco sugars are extracted using a low co-solvent, and the specific method is as follows: Choline chloride and 1,2-propylene glycol were mixed at a molar ratio of 1:2, mixed evenly at 80° C., and heated until the solution became transparent to obtain a deep eutectic solvent solution; Weigh the tobacco leaf sample into a centrifuge tube, add a low eutectic solvent solution at a solid-liquid ratio of 1:40, and perform ultrasonic-assisted extraction for 60 minutes. The extract is centrifuged at 4°C for 10 minutes (8000 rpm). The filtrate is filtered, a small amount of co-crystallizing agent is added, and vacuum freeze-drying is performed to obtain solid tobacco leaf sugar substances.

3. The method for digitally characterizing the typicality of tobacco leaf sweetness characteristics according to claim 1, characterized in that: In step S3, the specific method of pyrolysis-gas chromatography-mass spectrometry analysis is as follows: The tobacco sugars obtained in step S2 were placed in a pyrolysis sample boat, which was then placed in a pyrolysis reactor. GC-MS was then started, with a pyrolysis temperature of 500°C and a residence time of 0.2 min. GC-MS conditions: A HP-5MS gas chromatography column (60 m × 250 μm × 0.25 μm) was used, high-purity helium (99.999%) was used as the carrier gas, the column flow rate was 1.0 ml / min, the injection volume was 1 μL, and the pyrolysis volatile products were injected with a split ratio of 10:

1. The GC-MS interface temperature was maintained at 280°C, and the heating program was as follows: initial temperature 40°C, hold for 2 min, increase to 140°C at a rate of 5°C / min, hold for 2 min, then increase to 280°C at a rate of 5°C / min, hold for 5 min. Mass spectrometer settings: electron impact ionization source, electron energy 70 eV, ion source temperature 230 °C, quadrupole temperature 230 °C, scan range 30-550 amu.

4. The method for digitally characterizing the typicality of tobacco leaf sweetness characteristics according to claim 1, characterized in that: In step S4, the carbohydrate decomposition product spectrum is divided into a model training set and a model test set in a ratio of 7:

3. The training set is used to establish a quantitative evaluation model between the tobacco leaf carbohydrate decomposition product spectrum and the sweetness score using the BP artificial neural network method, and the effectiveness of the quantitative evaluation model is tested.

5. The method for digitally characterizing the typicality of tobacco leaf sweetness characteristics according to claim 4, characterized in that: According to the KS algorithm, 70% of the carbohydrate pyrolysis product spectra were extracted as the model test set, and the remaining samples were used as the model training set.

6. The method for digitally characterizing the typicality of tobacco leaf sweetness characteristics according to claim 4, characterized in that: Conduct validity tests on the quantitative evaluation model, including: For the model training set, the self-validation method and cross-validation method were used respectively, and the atlas data of each sample was substituted into the quantitative evaluation model for validation and cross-validation to check the error of the model; For the model test set, the atlas data of each sample was substituted into the quantitative evaluation model to analyze its sweetness score. The results obtained by the quantitative evaluation model were compared with the actual scores of the samples to verify the error of the model.

Images