Flavor and fragrance imitation compounding method based on taste feature fusion algorithm

Abstract

The invention discloses a flavor and fragrance compounding method based on a taste feature fusion algorithm, which comprises the following steps: (1) carrying out electronic tongue evaluation on flavors and fragrances, and establishing a taste evaluation method; (2) obtaining a taste feature matrix; (3) calculating the similarity of the taste features based on a correlation analysis method, and screening out fragrance raw materials having high correlation with the target essence; (4) obtaining a taste feature matrix of the high-correlation aromatic raw material under different concentration gradient conditions; (5) establishing an interpolation model of the concentration-taste value of the high-correlation aroma raw material; and (6) constructing an Euclidean distance function and carrying out minimum search in a continuous concentration interval to obtain the optimal concentration of the high-correlation aroma raw material closest to the taste sense of the target essence. The method is objective, scientific and rapid.

Description

Technical Field

[0001] This invention belongs to the field of flavor and fragrance technology, specifically relating to a method for imitating and compounding flavor and fragrances by using the taste characteristics of flavor and fragrance and their fusion algorithm. Background Technology

[0002] Independent control of core flavoring and fragrance technologies is crucial for the tobacco industry to enhance product competitiveness and mitigate market risks. Therefore, major tobacco companies have consistently focused on improving their independent flavoring capabilities to ensure their products remain competitive in the face of increasingly sophisticated consumer sensory demands. Among these, advanced analytical, imitation, and compounding technologies for flavorings and fragrances are core factors supporting innovation in cigarette product style, sensory experience, improved smoking quality, and stable quality assurance.

[0003] Traditional fragrance and flavor replication relies heavily on the sensory evaluation and experience of perfumers. However, these evaluations are susceptible to individual differences, physiological fluctuations, and environmental factors, leading to subjective biases that are difficult to replicate. Furthermore, the training of professional perfumers is demanding, time-consuming, and challenging, requiring multiple rounds of iterative optimization during formulation design. This results in lengthy, costly, and labor-intensive research and development. Consequently, modern techniques such as chromatography and mass spectrometry, which rely on precise component analysis, have emerged for fragrance replication and replication. While modern analytical techniques can analyze the volatile aroma components of a target compound to some extent, the qualitative and quantitative analysis of large molecular taste components remains a blind spot in chromatographic analysis. There is also a disconnect between the "component-perception" relationship, especially when dealing with complex natural fragrance raw materials. The interaction of taste and olfactory substances means that simple component addition cannot fully reflect the comprehensive sensory characteristics of the target compound. Therefore, replication and replication based solely on intrinsic components often results in significant differences in sensory quality.

[0004] To address the problems of existing technologies, this invention proposes a method for the imitation and compounding of flavorings and fragrances based on a fusion algorithm of flavor characteristics and similarity. Summary of the Invention

[0005] To address the shortcomings and deficiencies of existing technologies, this invention proposes a method for blending flavorings and fragrances based on a fusion algorithm that integrates the taste characteristics and similarity of flavorings and fragrances.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] (1) Conduct electronic tongue evaluation of flavorings and fragrances, and establish a taste evaluation method applicable to target flavorings and fragrance raw materials;

[0008] (2) Use taste evaluation methods to obtain the taste feature matrix of the target flavoring and flavor raw materials;

[0009] (3) Based on the correlation analysis method, calculate the similarity of taste characteristics between each flavoring raw material and the target flavoring, screen out flavoring raw materials with high correlation with the target flavoring, and quickly eliminate samples with significant differences.

[0010] (4) Use taste evaluation methods to obtain the taste characteristic matrix of highly correlated flavor raw materials under different concentration gradient conditions;

[0011] (5) Establish an interpolation model for the concentration of highly correlated flavoring raw materials and their taste values;

[0012] (6) In order to determine the optimal concentration that is closest to the taste sensory experience of the highly correlated flavor raw material and the target flavor, the optimal concentration prediction value is obtained by constructing a Euclidean distance function and performing a minimum search within a continuous concentration range.

[0013] (7) Professional perfumers verify the predicted value based on the optimal concentration.

[0014] Preferably, the optimization factors of the taste evaluation method established in step (1) include the type of blank solution, the dilution ratio, and the measurement temperature.

[0015] Preferably, the types of fragrance raw materials mentioned in step (2) are inferred through existing analytical techniques or judged by the experience of professional perfumers.

[0016] Preferably, the taste characteristics mentioned in step (2) include several of the following: sour, sweet, bitter, salty, umami, astringent, bitter aftertaste, astringent aftertaste, and umami aftertaste. Different weights can be assigned to different taste indicators in the matrix according to the sample requirements.

[0017] Preferably, the correlation analysis method mentioned in step (3) includes one or more statistical methods such as Pearson correlation coefficient; the high correlation screening principle is set to correlation r>0.8.

[0018] Preferably, the different concentration gradient conditions mentioned in step (4) are set to one or more concentrations from 0.01% to 100% according to the material characteristics.

[0019] Preferably, step (5) is to achieve taste prediction of flavoring raw materials in any concentration range. The interpolation model mentioned can be cubic spline interpolation or other interpolation functions. A continuous prediction model of the sample in each taste dimension can be established within the concentration range [0.01%, 1.00%].

[0020] Preferably, the numerical optimization of the function algorithm mentioned in step (6) can be achieved by using the golden section search and continuous parabolic interpolation to find the minimum Euclidean distance. The specific calculation steps are as follows:

[0021] (61) Construct the target flavor vector: The multidimensional flavor characteristics of the target flavor are: t = [sour, sweet, ..., bitter];

[0022] (62) Constructing a prediction vector function for highly correlated flavor ingredients: At any concentration c, the predicted value of the flavor ingredient's taste characteristics is: x(c)=[f1(c),f2(c),…,f i (c)], where f i (c) is the spline function for the i-th taste dimension;

[0023] (63) Define the Euclidean distance function: The sensory distance between the fragrance raw material and the target fragrance at concentration c is:

[0024] (64) Applying the golden section search algorithm:

[0025] (641) Search interval: c min ,c max = [0.01%, 1.00%];

[0026] (642) Each time the concentration point c is tested j Calculate D(c) j );

[0027] (643) In the search iteration, the interval is gradually narrowed to find the minimum point.

[0028] The advantages of this invention are:

[0029] 1. Compared with traditional human sensory evaluation techniques, the method of this invention uses an electronic tongue sensor to digitally evaluate the taste characteristics of target flavorings and flavoring raw materials, and deeply mines the taste contour similarity and dynamic concentration optimization model statistical algorithm to achieve objective, scientific, and rapid matching of flavoring characteristics and prediction of optimal addition concentration. This provides digital technical support for the selection of raw materials, blending ratio and sensory characteristic optimization of flavoring compounding, significantly reduces the workload of human sensory evaluation, improves work efficiency and effectively avoids the risk of subjective bias.

[0030] 2. Compared with the imitation compounding technology based on the content of ingredients, the method of the present invention starts from the sensory needs, takes the basic tastes such as sour, sweet, bitter, salty and umami as the target of flavor compounding, and aims at the sensory functionality and application applicability of flavors. At the same time, it fully considers the differences in taste characteristics of raw materials from different sources of the same type, and proposes a brand-new imitation compounding idea, which is more in line with the original intention of flavor compounding work.

[0031] 3. The candidate formulations proposed based on the dynamic concentration optimization model will greatly shorten the flavor formulation development cycle. Attached Figure Description

[0032] Figure 1This is a schematic diagram of the candidate fragrance concentration with the smallest European distance for the target fragrance in Example 1. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the scope of the invention.

[0034] Example 1: A replica compound of fruit flavoring G1.

[0035] The main characteristic aroma of Fruit Flavoring G1 is apricot-based, combined with other fruity notes. Its function is to enhance the fruity aroma and flavor of cigarettes, enrich the smoke's complexity, improve sweetness, and increase aroma concentration. Analysis of the volatile components of G1 flavoring suggests that the raw materials are natural extracts with a predominantly sweet and sour flavor profile, including jujube extract, fig extract, and plum extract. G1 flavoring compounding method:

[0036] (1) The taste values ​​of the target flavor G1 were obtained by electronic tongue detection: sour -2.42, sweet 1.48, bitter 1.07, salty 0.91, umami 1.37, astringent 0.28, bitter aftertaste 0.21, astringent aftertaste 0.16, and umami aftertaste 0.09.

[0037] (2) The taste values ​​of 30 different flavor raw materials, such as jujube, fig, plum, prickly pear, blackcurrant, etc., were detected by electronic tongue and compared by Pearson correlation analysis. Five flavor raw material candidates with high similarity to the target flavor G1 (similarity > 0.9) were screened out, as shown in Table 1.

[0038] Table 1. Similarity between candidate fragrance raw materials and target fragrances

[0039]

[0040] (3) The five selected flavor raw materials were subjected to gradient dilution (1%, 5%, 10%, 25%, 50%, 100%), and the taste characteristics of each raw material were detected at the six dilution concentrations.

[0041] (4) A highly correlated flavor raw material concentration-taste value interpolation model was established by using the cubic spline interpolation method to obtain continuous predicted values ​​of flavor raw materials in each taste dimension;

[0042] (5) Use the golden section search and continuous parabolic interpolation method to find the candidate fragrance concentration with the smallest Euclidean distance from the target fragrance.

[0043] For example, the 9-dimensional taste characteristics of the target flavor G1 are:

[0044] t=[-2.423,1.070,0.280,0.207,0.163,1.367,0.090,0.910,1.483]

[0045] The taste feature vector of flavoring ingredient #7 at a concentration of c = 10% is:

[0046] x=[-2.610,0.270,0.087,0.073,0.090,0.810,0.060,0.293,0.903]

[0047] Euclidean distance:

[0048]

[0049] D(c) was calculated from multiple concentrations: D(0.05)≈2.8; D(0.10)≈1.33; D(0.25)≈1.10; D(0.50)≈1.42; D(0.75)≈1.91… The minimum value appeared around 25%. Further refining the sampling points to 18%, 22%, 28%, etc., gradually approached the minimum value. Finally, the candidate fragrance concentration with the smallest distance from the target fragrance's European style was found. The results are as follows… Figure 1 As shown in Table 2;

[0050] Table 2. Predicted Optimal Concentrations of 5 Fragrance Raw Material Candidates

[0051]

[0052] (6) Based on the screening results and the predicted optimal concentration value, a basic formula was developed, and professional perfumers verified it based on the basic formula; the results are shown in Table 3.

[0053] Table 3 Predicted Optimal Concentration of Candidate Flavor Raw Materials

[0054]

[0055] The results in Table 3 show that the optimal concentration calculated by the method of the present invention is largely consistent with the actual concentration finally determined by the perfumer.

[0056] The foregoing has shown and described the main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of preferred examples. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed.

Claims

1. A method for imitating and compounding flavorings and fragrances based on a taste feature fusion algorithm, characterized in that, Includes the following steps: (1) Conduct electronic tongue evaluation of flavorings and fragrances, and establish a taste evaluation method applicable to target flavorings and fragrance raw materials; (2) Use taste evaluation methods to obtain the taste feature matrix of the target flavoring and flavor raw materials; (3) Based on the correlation analysis method, calculate the similarity of taste characteristics between each flavoring raw material and the target flavoring, and screen out flavoring raw materials that are highly correlated with the target flavoring; (4) Use taste evaluation methods to obtain the taste characteristic matrix of highly correlated flavor raw materials under different concentration gradient conditions; (5) Establish an interpolation model for the concentration of highly correlated flavoring raw materials and their taste values; (6) Construct a Euclidean distance function and perform a minimum search within a continuous concentration range to obtain the optimal concentration that is closest to the taste sensation of the highly correlated flavoring raw material and the target flavoring.

2. The method according to claim 1, characterized in that, The taste characteristics in step (2) include several of the following: sour, sweet, bitter, salty, umami, astringent, bitter aftertaste, astringent aftertaste, and umami aftertaste.

3. The method according to claim 1, characterized in that, The correlation analysis method in step (3) includes Pearson correlation analysis; high correlation is defined as correlation r > 0.

8.

4. The method according to claim 1, characterized in that, The different concentration gradients in step (4) are set to one or more concentrations ranging from 0.01% to 100% according to the material characteristics.

5. The method according to claim 1, characterized in that, The interpolation model in step (5) is cubic spline interpolation.

6. The method according to claim 1, characterized in that, The Euclidean distance function in step (6) is numerically optimized using the golden section search and continuous parabolic interpolation method to find the minimum Euclidean distance; the specific calculation steps are as follows: (61) Construct the target flavor vector: The multidimensional flavor characteristics of the target flavor are: t = [sour, sweet, ..., bitter]; (62) Constructing a prediction vector function for highly correlated flavor ingredients: At any concentration c, the predicted value of the flavor ingredient's taste characteristics is: x(c)=[f1(c),f2(c),…,f i (c)], where f i (c) is the spline function for the i-th taste dimension; (63) Define the Euclidean distance function: The sensory distance between the fragrance raw material and the target fragrance at concentration c is: (64) Applying the golden section search algorithm: (641) Search interval: c min ,c max = [0.01%, 1.00%]; (642) Each time the concentration point c is tested j Calculate D(c) j ); (643) In the search iteration, the interval is gradually narrowed to find the minimum point.

Images