Method and device for improving formula of tobacco essence

Abstract

The invention provides a method for improving a tobacco essence formula, which comprises the following steps of: coding an initial formula based on the type and the preparation amount of an essence raw material in the initial formula to obtain a first formula characteristic; performing similarity calculation on the first formula feature and second formula features of the multiple empirical formulas to obtain formula similarities between the initial formula and the multiple empirical formulas; based on the multiple formula similarities, respective evaluation scores of the initial formula in multiple evaluation dimensions are determined; based on the respective evaluation scores of the initial formula in the plurality of evaluation dimensions, the types and the preparation amounts of the incense raw materials in the initial formula are adjusted, and the types and the preparation amounts of the improved incense raw materials are obtained; and determining an improved formula based on the types and the preparation amounts of the improved perfume raw materials.

Description

Technical Field

[0001] This disclosure relates to the field of cigarette product design, and more specifically, to a method and apparatus for improving a tobacco flavoring formulation. Background Technology

[0002] Cigarette flavoring is one of the core aspects of cigarette product design. Traditional flavoring relies primarily on experienced flavorists who, through sensory evaluation and long-term practical experience, comprehensively assess the characteristics, compatibility, and style of flavoring ingredients. The industry has accumulated a wealth of experiential formulas and flavoring ingredient documentation; however, methods for improving cigarette flavoring formulations based on traditional flavoring generally suffer from the following problems:

[0003] Information on fragrance raw materials is not structured: the characteristics of fragrance raw materials are mostly described in text, such as "mainly balsamic fragrance, supplemented with sweet fragrance and herbal fragrance, which has the function of harmonizing fragrance and increasing fragrance volume", which is difficult to use directly for calculation or statistical analysis.

[0004] Lack of a unified indicator system: Different companies and perfumers use different terminology, focus, and scoring habits, making it difficult to compare formulas and raw materials on the same coordinate system.

[0005] The evaluation dimensions have low quantification: Currently, empirical "good / average / poor" or qualitative descriptions are commonly used, and there is a lack of unified numerical scales for "fragrance intensity" and "degree of enrichment", which is not conducive to computer-aided analysis and model training.

[0006] Difficulty in integrating with intelligent algorithms: Although attempts have emerged in recent years such as digital perfumery systems and large language model-assisted design, there is a lack of a stable and assimilable "fragrance-functional indicator system", which makes it difficult for the model to understand the formula style and output controllable results.

[0007] Therefore, there is an urgent need for a method that does not rely on a specific system, but starts from the "method" level, maps the description of flavoring raw materials and formulation composition to a unified evaluation dimension to achieve quantitative scoring, and improves tobacco flavoring formulations based on quantitative scoring. Summary of the Invention

[0008] In view of this, the present disclosure provides an improved method and apparatus for formulating tobacco flavorings.

[0009] One aspect of this disclosure provides a method for improving a tobacco flavoring formulation, comprising: encoding an initial formulation based on the types and quantities of flavoring ingredients in the initial formulation to obtain a first formulation feature; calculating the similarity between the first formulation feature and second formulation features of multiple empirical formulations to obtain formulation similarity between the initial formulation and the multiple empirical formulations; determining the evaluation scores of the initial formulation on multiple evaluation dimensions based on the multiple formulation similarities; adjusting the types and quantities of flavoring ingredients in the initial formulation based on the evaluation scores of the initial formulation on multiple evaluation dimensions to obtain improved types and quantities of flavoring ingredients; and determining an improved formulation based on the improved types and quantities of flavoring ingredients.

[0010] According to embodiments of this disclosure, the similarity calculation of the first formulation feature with the second formulation features of a plurality of empirical formulations is performed to obtain the formulation similarity between the initial formulation and the plurality of empirical formulations, including: for each second formulation feature of an empirical formulation, the cosine similarity calculation of the second formulation feature and the first formulation feature is performed to obtain the first similarity; the first similarity is processed by non-negative similarity to obtain the formulation similarity between the initial formulation and the empirical formulations.

[0011] According to embodiments of this disclosure, the first formulation feature and a plurality of second formulation features are normalized to obtain normalized first formulation features and a plurality of normalized second formulation features; wherein, for each second formulation feature of an empirical formulation, the cosine similarity between the second formulation feature and the first formulation feature is calculated to obtain a first similarity, including: calculating the cosine similarity between the normalized second formulation feature of each empirical formulation and the normalized first formulation feature to obtain a first similarity.

[0012] According to embodiments of this disclosure, a non-negative similarity processing is performed on the first similarity to obtain the formula similarity between the initial formula and the empirical formula, including: performing non-negative similarity processing on the first similarity to obtain a second similarity; and performing exponentiation processing with the product of the second similarity and the amplification factor as the exponent and the natural constant as the base to obtain the formula similarity between the initial formula and the empirical formula, wherein the amplification factor is a predefined real number.

[0013] According to embodiments of this disclosure, determining the evaluation scores of an initial formula on multiple evaluation dimensions based on multiple formula similarities includes: determining the similarity weights of multiple empirical formulas based on the ratios of the multiple formula similarities to the cumulative similarity, wherein the cumulative similarity represents the sum of the multiple formula similarities; and performing a weighted summation of the similarity weights of the multiple empirical formulas and the preset evaluation scores of the multiple empirical formulas on multiple evaluation dimensions to obtain the evaluation scores of the initial formula on multiple evaluation dimensions.

[0014] According to embodiments of this disclosure, determining the evaluation score of an initial formula on multiple evaluation dimensions based on multiple formula similarities includes: determining the initial evaluation score of the initial formula on multiple evaluation dimensions based on multiple formula similarities; for the initial evaluation score of each evaluation dimension, if the initial evaluation score is greater than a first preset value or less than a second preset value, determining the evaluation score of the initial formula on the evaluation dimension based on the first preset value or the second preset value.

[0015] According to embodiments of this disclosure, the initial formula is encoded based on the types and quantities of fragrance ingredients in the initial formula to obtain a first formula feature, including: obtaining the respective quantities and contribution values ​​of the various fragrance ingredients in the initial formula, wherein the contribution values ​​are quantitatively assigned by a perfumery expert based on multiple evaluation dimensions; determining the respective weights of the various fragrance ingredients based on the ratios of the multiple quantities to the total amount used in the initial formula; and performing a weighted summation of the weights of the multiple quantities and the contribution values ​​of the various fragrance ingredients in the initial formula to obtain the first formula feature.

[0016] According to embodiments of this disclosure, based on the evaluation scores of the initial formula across multiple evaluation dimensions, adjusting the types and quantities of fragrance raw materials in the initial formula to obtain improved types and quantities of fragrance raw materials includes: obtaining preset target evaluation scores across multiple evaluation dimensions; using the multiple target evaluation scores as target values ​​and the evaluation scores of the initial formula across multiple evaluation dimensions as result values, performing error sum-of-squares calculations to obtain evaluation differences; using minimizing the evaluation differences as the objective function, adjusting the types and quantities of fragrance raw materials in the initial formula based on a linear programming algorithm or a nonlinear programming algorithm to obtain improved types and quantities of fragrance raw materials.

[0017] According to embodiments of this disclosure, if the total configuration cost based on the improved fragrance raw material types and their formulation amounts does not meet the preset cost conditions, the improved formula is further improved.

[0018] Another aspect of this disclosure provides an apparatus for improving tobacco flavoring formulations, comprising: a formulation feature acquisition module, used to encode the initial formulation based on the types and quantities of flavoring raw materials in the initial formulation to obtain a first formulation feature; a formulation similarity determination module, used to calculate the similarity between the first formulation feature and second formulation features of multiple empirical formulations to obtain the formulation similarity between the initial formulation and the multiple empirical formulations; a formulation evaluation module, used to determine the evaluation score of the initial formulation on multiple evaluation dimensions based on the multiple formulation similarities; a formulation improvement module, used to adjust the types and quantities of flavoring raw materials in the initial formulation based on the evaluation scores of the initial formulation on multiple evaluation dimensions to obtain improved types and quantities of flavoring raw materials; and a formulation determination module, used to determine an improved formulation based on the improved types and quantities of flavoring raw materials.

[0019] According to embodiments of this disclosure, by encoding the initial formula of cigarettes, formula features that can be quantified are obtained, and the evaluation scores of the formula features on multiple evaluation dimensions are determined based on similarity calculation. Through specific digital evaluation scores, quantitative and accurate cigarette improvement formulas can be obtained while meeting cost constraints and minimizing evaluation score errors, thereby improving the efficiency and accuracy of cigarette flavoring. Attached Figure Description

[0020] The above and other objects, features and advantages of this disclosure will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0021] Figure 1 A flowchart is shown of an improved method for formulating tobacco flavorings according to an embodiment of the present disclosure.

[0022] Figure 2 A flowchart of a method for determining formulation similarity according to an embodiment of the present disclosure is shown.

[0023] Figure 3 A flowchart of a method for determining formulation similarity according to another embodiment of this disclosure is shown.

[0024] Figure 4 A flowchart of an evaluation score determination method according to an embodiment of the present disclosure is shown.

[0025] Figure 5 A flowchart of a method for determining evaluation scores according to another embodiment of this disclosure is shown.

[0026] Figure 6 A flowchart of a method for determining a first formulation feature according to an embodiment of the present disclosure is shown.

[0027] Figure 7A flowchart illustrating a method for determining the types and quantities of fragrance raw materials according to embodiments of the present disclosure is shown.

[0028] Figure 8 A block diagram of an improved apparatus for a tobacco flavoring formulation according to an embodiment of the present disclosure is shown. Detailed Implementation

[0029] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation. However, it will be apparent that one or more embodiments may be practiced without these specific details. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concepts of the present disclosure.

[0030] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0031] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0032] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0033] In the embodiments disclosed herein, the collection, updating, analysis, processing, use, transmission, provision, disclosure, and storage of data (e.g., including but not limited to user personal information) comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. In particular, necessary measures have been taken to prevent unauthorized access to user personal information data and to safeguard user personal information security, network security, and national security.

[0034] In the embodiments disclosed herein, user authorization or consent is obtained before acquiring or collecting user personal information.

[0035] Traditional flavoring relies heavily on experienced perfumers who, through sensory evaluation and long-term practical experience, comprehensively judge the characteristics, compatibility, and style of flavoring ingredients to determine the appropriate quantities. However, this method is highly dependent on the perfumer's subjective experience, lacks a unified indicator system, and is difficult to integrate with intelligent algorithms, hindering precise adjustments and improvements to tobacco flavoring formulas. Therefore, this paper proposes quantifying the various descriptive evaluation dimensions of flavoring ingredients and adaptively processing the quantified feature values ​​to obtain evaluation scores easily integrated with intelligent algorithms. Through intelligent algorithm planning, more precise cigarette improvement formulas can be obtained while meeting cost constraints.

[0036] Figure 1 A flowchart is shown of an improved method for formulating tobacco flavorings according to an embodiment of the present disclosure.

[0037] like Figure 1 As shown, the method for improving the formulation of tobacco flavoring includes operations S110~S150.

[0038] In operation S110, based on the types and quantities of fragrance raw materials in the initial formula, the initial formula is coded to obtain the first formula characteristics.

[0039] An initial formulation is a pre-obtained tobacco flavoring formula that requires improvement. It contains information on the types of various flavoring ingredients and the dosage of each ingredient. The initial formulation can be empirical data from various tobacco industry technical centers, or experimental data from various tobacco industry technical centers.

[0040] Fragrance raw materials can be natural fragrances and extracts, synthetic fragrance monomers, reactive fragrances, humectants, or processing aids, etc. For example, the types of fragrance raw materials include, but are not limited to: 2-(5- or 6-)methoxy-3-methylpyrazine, 2-ethylpyridine, furfural, benzaldehyde, ethyl acetate, phenylethyl phenylacetate, phenylacetic acid, 4-ethylguaiacol, acetophenone, methylcyclopentenolone, vanillin, cinnamon oil, angelica root oil, vanilla bean tincture, Peruvian extract, alcohol, and 1,2-propanediol, etc.

[0041] In the embodiments of this disclosure, the process of encoding the initial formula is a quantitative evaluation process of cigarettes based on the initial formula. Cigarettes based on the initial formula have multiple evaluation dimensions, and the encoding process involves quantifying and assigning values ​​to each evaluation dimension. The first formula feature is the specific evaluation value of the cigarette based on the initial formula across multiple evaluation dimensions.

[0042] Evaluation dimensions can be obtained from the company's raw material manual or industry literature. In the embodiments of this disclosure, evaluation dimensions include, but are not limited to, balsamic aroma, sweet aroma, baked aroma, milky aroma, herbal aroma, woody aroma, fruity aroma, spicy aroma, sour aroma, green aroma, harmonious aroma, increased richness, increased aroma quantity, increased transparency, increased concentration, improved delicacy, improved softness, masking off-flavors, and enhanced aroma quality, etc.

[0043] In the embodiments of this disclosure, the values ​​assigned to the evaluation dimensions can be determined by flavoring experts, based on industry experience data, or determined by algorithmic processing of experience data. The assignment of values ​​to the evaluation dimensions needs to be done on a uniform scale, which can range from 0 to 10 or from 0 to 100. For example, for a certain formula of cigarettes, which includes five evaluation dimensions—pasty aroma, sweet aroma, harmonious aroma, increased richness, and increased aroma quantity—with a scale range of 0 to 10, the padding aroma can be assigned a value of 8, the sweet aroma a value of 7, the harmonious aroma a value of 6, the increased richness a value of 5, and the increased aroma quantity a value of 4.

[0044] In operation S120, the similarity between the first formula feature and the second formula feature of multiple empirical formulas is calculated to obtain the formula similarity between the initial formula and the multiple empirical formulas.

[0045] Empirical formulations are known tobacco flavoring formulations used as references, and can be tobacco flavoring formulations stored in the databases of various tobacco industry technical centers. In embodiments of this disclosure, the second formulation feature of the empirical formulation is obtained in the same manner as the first formulation feature.

[0046] Similarity calculation is a method to measure the degree of similarity between two or more objects. Similarity calculation methods include, but are not limited to: cosine similarity, Euclidean distance, Pearson correlation coefficient, etc.

[0047] In operation S130, based on the similarity of multiple formulations, the evaluation scores of the initial formulation on multiple evaluation dimensions are determined.

[0048] In the embodiments of this disclosure, the number of formula similarities is consistent with the number of empirical formulas. Each empirical formula has corresponding specific evaluation values ​​across multiple evaluation dimensions. The method for obtaining these specific evaluation values ​​is the same as the method for obtaining the specific values ​​of the initial formula across multiple evaluation dimensions described above, and will not be repeated here. By processing the calculations of multiple formula similarities and the specific evaluation values ​​corresponding to multiple formulas, the evaluation scores of the initial formula across multiple evaluation dimensions can be obtained.

[0049] In operation S140, based on the evaluation scores of the initial formula on multiple evaluation dimensions, the types and quantities of fragrance raw materials in the initial formula are adjusted to obtain the improved types and quantities of fragrance raw materials.

[0050] To ensure the formula achieves the desired performance indicators, each evaluation dimension has a corresponding target evaluation score. The initial formula's evaluation score in any evaluation dimension should be as close as possible to the target evaluation score for that dimension. Based on this goal, specific linear or nonlinear programming algorithms can be used to continuously adjust the types and quantities of fragrance ingredients in the initial formula to obtain improved types and quantities of fragrance ingredients.

[0051] In operation S150, an improved formula is determined based on the improved types and quantities of fragrance raw materials.

[0052] Through the embodiments of this disclosure, quantitative formula characteristics can be obtained, thereby obtaining specific digital evaluation scores through the quantitative formula characteristics. Based on the evaluation scores, the tobacco flavoring formula can be continuously adjusted to obtain quantitative and accurate cigarette improvement formulas, thereby improving the efficiency and accuracy of cigarette flavoring.

[0053] Figure 2 A flowchart of a method for determining formulation similarity according to an embodiment of the present disclosure is shown.

[0054] like Figure 2 As shown, the method for determining the similarity of the formulations includes operations S210~S220.

[0055] In operation S210, for each empirical formula's second formula feature, the cosine similarity between the second formula feature and the first formula feature is calculated to obtain the first similarity.

[0056] In operation S220, the first similarity is processed by non-negative similarity to obtain the formula similarity between the initial formula and the empirical formula.

[0057] Cosine similarity is an index that measures the difference in direction between two vectors. It assesses their similarity by calculating the cosine of the angle between the two vectors. In embodiments of this disclosure, the first formula feature and the second formula feature can be represented as vectors. Applying non-negative similarity processing to the first similarity means assigning a value of zero to the first similarity if the value is less than zero.

[0058] Through the embodiments of this disclosure, the interference of outliers can be eliminated and a more accurate formulation similarity can be obtained.

[0059] According to embodiments of this disclosure, the first formulation feature and multiple second formulation features are normalized to obtain normalized first formulation features and multiple normalized second formulation features. Specifically, for each second formulation feature of an empirical formulation, a cosine similarity calculation is performed between the second formulation feature and the first formulation feature to obtain a first similarity. This includes: performing a cosine similarity calculation between the normalized second formulation feature and the normalized first formulation feature of each empirical formulation to obtain a first similarity.

[0060] Normalization is the process of scaling data proportionally to bring it into a specific range, primarily used to eliminate the influence of different dimensions between features. In the embodiments of this disclosure, normalization methods include, but are not limited to, extreme value normalization, mean-standard deviation normalization, robust normalization, L2 normalization, etc.

[0061] Through the embodiments of this disclosure, the influence of scoring differences caused by different evaluation dimensions can be eliminated, and a more accurate formula similarity can be obtained.

[0062] Figure 3 A flowchart of a method for determining formulation similarity according to another embodiment of this disclosure is shown.

[0063] like Figure 3 As shown, the method for determining the similarity of the formulations includes operations S310~S320.

[0064] In operation S310, the first similarity is processed by non-negative similarity to obtain the second similarity.

[0065] In operation S320, the product of the second similarity and the amplification factor is used as the exponent, and the natural constant is used as the base to perform power operation to obtain the formula similarity between the initial formula and the empirical formula, where the amplification factor is a predefined real number.

[0066] In embodiments of this disclosure, the amplification factor can be any real number between 0.1 and 10. For example, when the second similarity is 0.5, an amplification factor of 3 is chosen, at which point the formula similarity is... .

[0067] The embodiments of this disclosure can enhance the influence of highly similar empirical formulations, resulting in more accurate formulation similarity.

[0068] Figure 4 A flowchart of an evaluation score determination method according to an embodiment of the present disclosure is shown.

[0069] like Figure 4 As shown, the method for determining the evaluation score includes operations S410~S420.

[0070] In operation S410, the similarity weights of multiple empirical formulas are determined based on the ratios of the similarity of multiple formulas to the cumulative similarity, where the cumulative similarity represents the sum of the similarities of multiple formulas.

[0071] In operation S420, the initial formula is obtained by weighting and summing the similarity weights of multiple empirical formulas and the preset evaluation scores of each empirical formula on multiple evaluation dimensions.

[0072] The methods for obtaining the preset evaluation scores of multiple empirical formulas across multiple evaluation dimensions have been described above and will not be repeated here. The preset evaluation score of the j-th empirical formula across multiple evaluation dimensions can be expressed as:

[0073] (1)

[0074] in, Let be the evaluation score of the j-th empirical formula on the M-th evaluation dimension.

[0075] The similarity weight of the j-th empirical formula can be based on the following formula:

[0076] (2)

[0077] in, Let be the similarity score of the j-th recipe.

[0078] The initial formula's evaluation scores across multiple evaluation dimensions can be expressed by the following formula:

[0079] (3)

[0080] Through the embodiments of this disclosure, a more accurate evaluation score for the initial formulation can be obtained by using the evaluation scores of multiple empirical formulations and formulation similarity.

[0081] Figure 5 A flowchart of a method for determining evaluation scores according to another embodiment of this disclosure is shown.

[0082] like Figure 5 As shown, the method for determining the evaluation score includes operations S510~S520.

[0083] In operation S510, based on the similarity of multiple formulations, the initial evaluation scores of the initial formulations on multiple evaluation dimensions are determined.

[0084] In operation S520, for the initial evaluation score of each evaluation dimension, if the initial evaluation score is greater than the first preset value or less than the second preset value, the evaluation score of the initial formula on the evaluation dimension is determined based on the first preset value or the second preset value.

[0085] The methods for obtaining the initial evaluation scores of the initial formula across multiple evaluation dimensions are as described above and will not be repeated here. The first and second preset values ​​can be any real numbers greater than 0, and their purpose is to eliminate outliers in the initial evaluation scores and avoid their influence.

[0086] In the embodiments of this disclosure, if the initial evaluation score is greater than a first preset value, the first preset value is used as the updated value of the initial evaluation score; if the initial evaluation score is less than a second preset value, the first preset value is used as the updated value of the initial evaluation score. For example, the initial evaluation scores of the initial formula on multiple evaluation dimensions are 0, 2, 4, 6, 7, 8, 13, and 20, respectively; the first preset value is 1; and the second preset value is 9. Since 0 is less than the first preset value, and 13 and 20 are greater than the second preset value, the initial evaluation score 0 is modified to 1, and the initial evaluation scores 13 and 20 are modified to 9, resulting in initial evaluation scores of 1, 2, 4, 6, 7, 8, 9, and 9 on multiple evaluation dimensions.

[0087] Through the embodiments of this disclosure, the influence of abnormal evaluation scores can be avoided, and a more accurate evaluation score for the initial formulation can be obtained.

[0088] Figure 6 A flowchart of a method for determining a first formulation feature according to an embodiment of the present disclosure is shown.

[0089] like Figure 6 As shown, the method for determining the first formulation characteristic includes operations S610 to S630.

[0090] In operation S610, the configuration amount of each of the various fragrance raw materials in the initial formula and the contribution value of each of the various fragrance raw materials are obtained. The contribution value is a quantitative assignment of fragrance raw materials by the perfumery expert based on multiple evaluation dimensions.

[0091] In operation S620, the configuration weights of various fragrance raw materials are determined based on the ratio of multiple configuration quantities to the total amount of the initial formula.

[0092] In operation S630, the first formula feature is obtained by weighting and summing the weights of multiple configuration quantities and the contribution values ​​of various fragrance raw materials in the initial formula.

[0093] In the embodiments of this disclosure, the total amount of the initial formulation is:

[0094] (4)

[0095] in Let be the amount of the i-th raw material in the initial formula.

[0096] The allocation weight of the i-th fragrance raw material is:

[0097] (5)

[0098] The first formula is characterized by:

[0099] (6)

[0100] in, This represents the contribution value of the i-th fragrance ingredient in the initial formula.

[0101] For example, the initial formula includes two fragrance raw materials: benzaldehyde and ethyl acetate. Based on the evaluation dimension of balm fragrance, the perfumer assigns a contribution value of 2 to benzaldehyde and a contribution value of 3 to ethyl acetate. The amount of benzaldehyde is 0.05 and the amount of ethyl acetate is 0.6. The first formula characteristic in the evaluation dimension of balm fragrance is 2×0.05+3×0.6=1.9.

[0102] Figure 7 A flowchart illustrating a method for determining the types and quantities of fragrance raw materials according to embodiments of the present disclosure is shown.

[0103] like Figure 7 As shown, the improved method for determining the types and quantities of fragrance raw materials includes operations S710~S730.

[0104] By operating the S710, preset target evaluation scores are obtained on multiple evaluation dimensions.

[0105] In operation S720, multiple target evaluation scores are used as target values, and the evaluation scores of the initial formula on multiple evaluation dimensions are used as result values. The sum of squared errors is then used to calculate the evaluation difference.

[0106] In operation S730, the objective function is to minimize the evaluation difference. Based on linear programming or nonlinear programming algorithms, the types and quantities of fragrance raw materials in the initial formula are adjusted to obtain the improved types and quantities of fragrance raw materials.

[0107] The preset target evaluation scores for multiple evaluation dimensions can be determined based on empirical data from the tobacco industry. For example, the preset target evaluation score for the aroma / scent evaluation dimension is... The preset target evaluation score for the sweet aroma evaluation dimension is: The initial formula's evaluation scores in the balsamic and sweet aroma dimensions are x and y, respectively. By continuously adjusting the proportions of the fragrance ingredients in the initial formula using linear or nonlinear programming algorithms, the values ​​of x and y are changed, resulting in a final product. The types and quantities of fragrance raw materials corresponding to the minimum value are used as the improved types and quantities of fragrance raw materials.

[0108] According to embodiments of this disclosure, if the total configuration cost based on the improved fragrance raw material types and their formulation amounts does not meet the preset cost conditions, the improved formula is further improved.

[0109] The total configuration cost is derived from the price of each fragrance ingredient and the quantity of each ingredient in the formula. The preset cost is a constant representing the price and can be obtained based on empirical data. If the total configuration cost exceeds the preset cost, the formula needs to be revised.

[0110] For example, take an empirical formula containing a variety of fragrance ingredients, and the preparation amount of each fragrance ingredient is shown in Table 1.

[0111] Table 1: Composition of Formula A

[0112]

[0113] Based on the preparation amount of Formula A, the amount of "Peruvian extract" was adjusted from 2.0 g to 3.0 g, while the other preparation amounts remained unchanged. The adjusted formula composition is shown in Table 2.

[0114] Table 2: Composition of Formula B

[0115]

[0116] Based on formula A, the aromatic ingredient "Angelica root oil" was removed, while the proportions of the remaining aromatic ingredients remained the same as in formula A. The adjusted formula composition is shown in Table 3.

[0117] Table 3: Composition of Formula C

[0118]

[0119] According to the embodiments of this disclosure, formulations A, B, and C are scored on multiple evaluation dimensions to obtain evaluation scores for the three formulations on multiple dimensions. The evaluation score results are shown in Table 4.

[0120] Table 4: Evaluation scores for the three formulations

[0121]

[0122] As shown in Table 4, increasing the amount of Peruvian extract enhances the aroma and sweetness of the formula, consistent with its characteristic of being "primarily aroma-based, supplemented by sweet and resinous notes." Similarly, removing Angelica root oil reduces the herbal aroma of the formula, which is also consistent with its description of being "primarily herbal, supplemented by sweet and spicy notes."

[0123] According to the embodiments of this disclosure, formulations A, B, and C are scored on multiple evaluation dimensions to obtain evaluation scores for the three formulations on multiple other dimensions. The evaluation score results are shown in Table 5.

[0124] Table 5: Evaluation scores for the three formulations

[0125]

[0126] As shown in Table 5, increasing the amount of Peruvian extract enhances the richness and aroma of the formulation, consistent with its characteristic of "enhancing aroma quality and increasing aroma quantity." Similarly, removing Angelica root oil reduces the masking value of off-flavors in the formulation, consistent with its description of "improving mellowness."

[0127] Figure 8 A block diagram of an improved apparatus for a tobacco flavoring formulation according to an embodiment of the present disclosure is shown.

[0128] like Figure 8 As shown, the tobacco flavoring formulation improvement device 800 includes a formulation feature acquisition module 810, a formulation similarity determination module 820, a formulation evaluation module 830, a formulation improvement module 840, and a formulation determination module 850.

[0129] The formula feature acquisition module 810 is used to encode the initial formula based on the types and quantities of fragrance raw materials in the initial formula to obtain the first formula feature.

[0130] The formula similarity determination module 820 is used to calculate the similarity between the first formula feature and the second formula feature of multiple empirical formulas, respectively, to obtain the formula similarity between the initial formula and the multiple empirical formulas.

[0131] The formula evaluation module 830 is used to determine the evaluation scores of the initial formula on multiple evaluation dimensions based on the similarity of multiple formulas.

[0132] The formula improvement module 840 is used to adjust the types and quantities of fragrance raw materials in the initial formula based on the evaluation scores of the initial formula on multiple evaluation dimensions, so as to obtain the improved types and quantities of fragrance raw materials.

[0133] Formula determination module 850 is used to determine an improved formula based on the improved types of fragrance raw materials and their preparation amounts.

[0134] Those skilled in the art will understand that the features described in the various embodiments of this disclosure can be combined and / or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. In particular, the features described in the various embodiments of this disclosure can be combined and / or combined in various ways without departing from the spirit and teachings of this disclosure. All such combinations and / or combinations fall within the scope of this disclosure.

[0135] The embodiments of this disclosure have been described above. However, these embodiments are for illustrative purposes only and are not intended to limit the scope of this disclosure. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this disclosure, and all such substitutions and modifications should fall within the scope of this disclosure.

Claims

1. An improved method for formulating tobacco flavorings, characterized in that, The method includes: Based on the types and quantities of fragrance raw materials in the initial formula, the initial formula is coded to obtain the first formula characteristics; The similarity between the first formula feature and the second formula feature of multiple empirical formulas is calculated to obtain the formula similarity between the initial formula and the multiple empirical formulas. Based on the similarity of multiple formulas, the evaluation scores of the initial formula on multiple evaluation dimensions are determined. Based on the evaluation scores of the initial formula on multiple evaluation dimensions, the types and quantities of fragrance raw materials in the initial formula are adjusted to obtain the improved types and quantities of fragrance raw materials. Based on the improved types and quantities of fragrance raw materials, an improved formula was determined.

2. The method of claim 1, wherein, The similarity between the first formulation feature and the second formulation features of multiple empirical formulations is calculated to obtain the formulation similarity between the initial formulation and the multiple empirical formulations, including: For each empirical formula, the second formula feature is compared with the first formula feature by cosine similarity calculation to obtain the first similarity. The first similarity is processed by non-negative similarity to obtain the formula similarity between the initial formula and the empirical formula.

3. The method of claim 2, wherein, The method further includes: The first formulation feature and the multiple second formulation features are normalized respectively to obtain the normalized first formulation feature and the multiple normalized second formulation features; Specifically, for each empirical formula's second formula feature, calculating the cosine similarity between the second formula feature and the first formula feature to obtain a first similarity includes: The first similarity is obtained by calculating the cosine similarity between the normalized second formula feature and the normalized first formula feature of each empirical formula.

4. The method of claim 2, wherein, The step of performing non-negative similarity processing on the first similarity to obtain the formulation similarity between the initial formulation and the empirical formulation includes: The first similarity is processed by non-negative similarity to obtain the second similarity; Using the product of the second similarity and the amplification factor as the exponent and the natural constant as the base, a power operation is performed to obtain the formula similarity between the initial formula and the empirical formula, wherein the amplification factor is a predefined real number.

5. The method of claim 1, wherein, Based on the similarity of multiple formulations, the evaluation scores of the initial formulation on multiple evaluation dimensions are determined, including: Based on the ratio of the similarity of each of the multiple formulas to the cumulative similarity, the similarity weight of the multiple empirical formulas is determined, wherein the cumulative similarity represents the sum of the similarities of the multiple formulas; The initial formula is obtained by weighting and summing the similarity weights of the multiple empirical formulas and the preset evaluation scores of each of the multiple empirical formulas on multiple evaluation dimensions.

6. The method of claim 1, wherein, Based on the similarity of multiple formulations, the evaluation scores of the initial formulation on multiple evaluation dimensions are determined, including: Based on the similarity of multiple formulas, the initial evaluation scores of the initial formula on multiple evaluation dimensions are determined. For each evaluation dimension, if the initial evaluation score is greater than a first preset value or less than a second preset value, the initial formula's evaluation score on that evaluation dimension is determined based on the first preset value or the second preset value.

7. The method of claim 1, wherein, Based on the types and quantities of fragrance raw materials in the initial formula, the initial formula is coded to obtain the first formula characteristics, including: The formulation amounts and contribution values ​​of various fragrance ingredients in the initial formula are obtained, wherein the contribution values ​​are quantitatively assigned to the fragrance ingredients by a perfumery expert based on multiple evaluation dimensions. Based on the ratio of the multiple configuration quantities to the total amount of the initial formula, the configuration weights of the various fragrance raw materials are determined; The first formula feature is obtained by weighting and summing the weights of the multiple configuration quantities and the contribution values ​​of the various fragrance raw materials in the initial formula.

8. The method of claim 1, wherein, Based on the evaluation scores of the initial formula across multiple evaluation dimensions, the types and quantities of fragrance raw materials in the initial formula are adjusted to obtain improved types and quantities of fragrance raw materials, including: Obtain preset target evaluation scores across multiple evaluation dimensions; The evaluation difference is obtained by taking the multiple target evaluation scores as target values ​​and the evaluation scores of the initial formula on multiple evaluation dimensions as result values, and performing error sum of squares calculation. Using the minimization of the evaluation difference as the objective function, and based on linear programming or nonlinear programming algorithms, the types and quantities of fragrance raw materials in the initial formula are adjusted to obtain the improved types and quantities of fragrance raw materials.

9. The method of claim 1, wherein, If the total configuration cost based on the improved types and quantities of fragrance raw materials does not meet the preset cost conditions, the improved formula will continue to be improved.

10. In a smoking flavor formulation, the improvement wherein, The device includes: The formula feature acquisition module is used to encode the initial formula based on the types and quantities of fragrance raw materials in the initial formula to obtain the first formula feature; The formula similarity determination module is used to calculate the similarity between the first formula feature and the second formula feature of multiple empirical formulas, respectively, to obtain the formula similarity between the initial formula and the multiple empirical formulas. The formula evaluation module is used to determine the evaluation scores of the initial formula on multiple evaluation dimensions based on the similarity of multiple formulas. The formula improvement module is used to adjust the types and quantities of fragrance raw materials in the initial formula based on the evaluation scores of the initial formula on multiple evaluation dimensions, so as to obtain the improved types and quantities of fragrance raw materials. The formula determination module is used to determine the improved formula based on the types and quantities of the improved fragrance raw materials.

Images