Black tea digital blending method based on leading edge solution set multi-objective optimization algorithm

Abstract

The invention discloses a digital black tea blending method based on a leading edge solution set multi-objective optimization algorithm, and the method is based on the tea appearance score, the tea soup taste score, the tea soup color score, the tea soup fragrance score and the market price of blending raw materials. The PlatEMO multi-objective optimization framework and the GDE3 algorithm are combined to be used for digital blending of the black tea, the problems that in the traditional tea blending process, artificial sensory evaluation is high in subjectivity, poor in consistency and unstable in product quality are solved, and meanwhile the accuracy and precision of the black tea blending result are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tea production and processing, and in particular to a black tea digital blending method based on a frontier solution set multi-objective optimization algorithm and an application thereof. Background Art

[0002] Tea originated in China. As a natural and healthy drink, it is deeply loved and respected by people all over the world. Tea blending is a tea processing technology, which is mostly used by commercial tea processing companies and is a complex technology. The so-called tea blending is to combine products with certain commonalities but different shapes and qualities, and select their shortcomings to improve their shape, make their colors uniform, enhance their fragrance, or enrich their flavors. Tea blending is a commonly used method to improve the quality of tea, stabilize the quality of tea, expand the supply, increase the quantity, and obtain higher economic benefits.

[0003] The blending of black tea is a highly professional skill. It is usually determined by experienced tea experts through sensory evaluation to determine the sensory characteristics of each raw material and determine the final blending plan. The blending of black tea is essentially a mixing problem and can be regarded as a classic optimization problem. Through the reasonable combination of characteristic raw materials of different qualities, the final blended sample that meets specific requirements and is cost-effective is formed.

[0004] The current tea industry is developing rapidly, with new formats emerging one after another, and blending technology has become a key core technology. However, the traditional model can no longer meet the needs of the rapid development of the industry, and a more systematic and scientific method is urgently needed to improve the efficiency and accuracy of blending technology. The prior art "A green tea flavor blending method based on a linear hypothesis algorithm" only considers the taste quality of the tea, but does not consider the appearance, aroma and other qualities of the tea; the prior art "Intelligent tea blending method and system" uses multiple sensor information to obtain tea information, but does not consider the price factor of the tea.

[0005] The core of tea blending lies in the blending of sensory factors such as "color, aroma, taste, and shape". At present, the tea blending industry urgently needs digital blending technology based on product quality characteristics to ensure that the quality of blended products meets the target requirements. Especially in the field of black tea blending, there is currently a lack of intelligent tea blending methods that can comprehensively consider the three indicators of tea appearance, taste, and cost. Summary of the invention

[0006] The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art and to provide a black tea digital blending method based on a frontier solution set multi-objective optimization algorithm.

[0007] The first object of the present invention is to provide a black tea digital blending method based on a frontier solution set multi-objective optimization algorithm.

[0008] The second object of the present invention is to provide a digital blending model for black tea based on the multi-objective optimization algorithm of the frontier solution set.

[0009] The third object of the present invention is to provide the application of the above digital blending model for black tea in blending black tea.

[0010] In order to achieve the above objects, the present invention is realized through the following solutions:

[0011] A digital blending method for black tea based on the multi-objective optimization algorithm of the frontier solution set includes the following steps:

[0012] S1. Obtain the index parameters of each blending raw material; the index parameters include the appearance score of tea leaves, the taste score of tea soup, the color score of tea soup, the aroma score of tea soup, and the market price.

[0013] S2. Use the PlatEMO multi-objective optimization framework to transform the tea blending problem into a multi-objective optimization problem that takes into account both quality and cost.

[0014] S3. Input the index parameters obtained in step S1 into the GDE3 algorithm for processing, output the operation result of the model, and obtain the Pareto frontier solution set.

[0015] S4. Combine the Pareto frontier solution set obtained in step S3 with subjective preferences to obtain a black tea blending plan.

[0016] The method of the present invention combines the PlatEMO multi-objective optimization framework and the GDE3 algorithm for the digital blending of black tea, solves the problems of strong subjectivity, poor consistency, and unstable product quality in the traditional tea blending process, and can quickly generate an optimal blending plan for black tea.

[0017] Preferably, in step S1, the index parameters of each blending raw material are obtained through sensory evaluation.

[0018] More preferably, the blending raw material is tea leaves.

[0019] More preferably, the standards for the sensory evaluation are GB / T 23776-2018 and GB / T 14487-2017.

[0020] In a specific embodiment of the present invention, obtaining the index parameters of each blending raw material specifically includes: a team composed of 10 black tea evaluation experts (5 males and 5 females, aged between 25 and 45 years old) scores and evaluates the blending raw materials according to the tea sensory evaluation method (GB / T23776-2018) and the Chinese tea sensory evaluation vocabulary (GB / T 14487-2017) to obtain the index parameters.

[0021] Preferably, the PlatEMO multi-objective optimization framework in step S2 includes the following steps:

[0022] S21. Convert the tea leaf appearance score, tea soup taste score, tea soup color score, and tea soup aroma score in the index parameters obtained in step S1 into quality evaluation factors respectively according to formula 1;

[0023] Formula 1:

[0024] where S i is the index parameter of the i-th blended raw material, is the average value of this index parameter among all blended raw materials, and σ s is the standard deviation of this index parameter among all blended raw materials;

[0025] Then calculate the quality deviation Q according to formula 2;

[0026] Formula 2:

[0027] where S i1 , S i2 , S i3 and S i4 are all calculated by formula 1, x i is the blending ratio of the i-th blended raw material, S i1 is the S i of the tea leaf appearance score, S i2 is the S i of the tea soup taste score, S i3 is the S i of the tea soup color score, S i4 is the Si of the tea soup aroma score, and Starget is this index parameter after z-score standardization;

[0028] S22. Calculate the cost P according to the market price in the index parameters obtained in step S1, combined with the blending ratios of each blended raw material;

[0029] S23. Compare the quality deviation objective function Q obtained in step S21 and the cost P obtained in step S22, and select the minimum value among them;

[0030] S24. Set the following constraint conditions: Q ≤ δ, a i ≤ x i ≤ b i ;

[0031] where δ is the quality deviation threshold, x i is the blending ratio of the i-th blended raw material, a i is the lower limit of the blending ratio of the i-th blended raw material, and bi is the upper limit of the blending ratio of the i-th blended raw material.

[0032] Preferably, the process of inputting the index parameters obtained in step S1 into the GDE3 algorithm in step S3 includes the following steps:

[0033] S31. Population initialization: Generate random solutions based on a preset population size as the initial population;

[0034] S32. Differential operation: Randomly select three random solutions x 1 , x 2 and x 3 in the initial population obtained in step S31 and perform differential calculation according to formula 3 to obtain a new solution v i ;

[0035] Formula 3: v i = x 1 + F(x 2 - x 3 ); where F is the scaling factor, and its value range is [0, 1];

[0036] S33. Crossover operation: Perform a crossover operation on the new solution v i obtained in step S32 and x i according to formula 4 to obtain μ i ;

[0037] Formula 4:

[0038] where CR is the crossover rate, rand(0, 1) is a random value from a uniform distribution in the interval [0, 1], i is the i-th solution vector, j is the j-th dimension, D is the number of dimensions of the solution vector, and x i is the blending ratio of the i-th blended raw material;

[0039] S34. Fitness calculation: For x i and μ i obtained in step S33, calculate the fitness F(x i ) and F(μ i ). If F(μ i ) < F(x i ), then use μ i to replace x i ;

[0040] S35. Repeat steps S31 to S34 until the maximum number of iterations is reached or the function value converges.

[0041] Preferably, the preset population size in step S31 is 100.

[0042] Preferably, the maximum number of iterations in step S35 is 10,000 times.

[0043] Preferably, the combination of the Pareto front solution set obtained in step S3 and the subjective preference in step S4 specifically includes the following steps:

[0044] S41. Set the following constraints: Q ≤ δ, a i ≤ x i ≤ b i , to obtain a subjective preference scheme;

[0045] where δ is the quality deviation threshold, x i is the blending ratio of the i-th blended raw material, a i is the lower limit of the blending ratio of the i-th blended raw material, and b i is the upper limit of the blending ratio of the i-th blended raw material;

[0046] S42. On the basis of the Pareto front solution set obtained in step S3, add the subjective preference scheme as the final optimization result of the algorithm.

[0047] The present invention also claims a black tea digital blending model based on a multi-objective optimization algorithm for the front solution set, including a data acquisition module, a blending module, and a result output module;

[0048] The data acquisition module is used to obtain the index parameters of the blended raw materials; the index parameters include the appearance score of tea leaves, the taste score of the tea soup, the color score of the tea soup, the aroma score of the tea soup, and the market price;

[0049] The blending module, based on the index parameters of the blended raw materials obtained by the data acquisition module, executes the black tea digital blending method described in any one of the above, to obtain a black tea blending scheme;

[0050] The result output module is used to output the black tea blending scheme obtained by the blending module.

[0051] Preferably, the blended raw material is tea leaves.

[0052] Preferably, the acquisition standard of the index parameters of the blended raw materials is GB / T 23776-2018 and GB / T 14487-2017.

[0053] The present invention also claims the application of the black tea digital blending model described in any one of the above in blending black tea.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention provides a digital blending method for black tea based on a multi-objective optimization algorithm for the frontier solution set, comprising the following steps: S1. Obtain the index parameters of each blending raw material; the index parameters include the appearance score of the tea leaves, the taste score of the tea soup, the color score of the tea soup, the aroma score of the tea soup, and the market price; S2. Use the PlatEMO multi-objective optimization framework to transform the tea blending problem into a multi-objective optimization problem that takes into account both quality and cost; S3. Input the index parameters obtained in step S1 into the GDE3 algorithm for processing, and output the operation result of the model; S4. Use the Pareto frontier solution set combined with subjective preferences to obtain the black tea blending plan. Based on the appearance score of the tea leaves, the taste score of the tea soup, the color score of the tea soup, the aroma score of the tea soup, and the market price of the blending raw materials, the method combines the PlatEMO multi-objective optimization framework and the GDE3 algorithm for the digital blending of black tea, which not only solves the problems of strong subjectivity, poor consistency, and unstable product quality in the traditional tea blending process, but also improves the accuracy and precision of the black tea blending result. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 FIG. is a schematic flow chart of a digital blending method for black tea based on a multi-objective optimization algorithm in Embodiment 1;

[0057] Figure 2 FIG. is a scatter plot of the Pareto frontier solution set in the tea blending process in Embodiment 2;

[0058] Figure 3 FIG. is a result diagram of the Pareto frontier solution set in the tea blending process in Embodiment 2;

[0059] Figure 4 FIG. is a result diagram of the solution set of the blending plan in Embodiment 2. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] The present invention will be further described in detail below with reference to the accompanying drawings of the specification and specific embodiments. The embodiments are only used to explain the present invention and are not intended to limit the scope of the present invention. The test methods used in the following embodiments are all conventional methods unless otherwise specified; the materials, reagents, etc. used are reagents and materials that can be obtained from commercial channels unless otherwise specified.

[0061] Embodiment 1 A Digital Blending Method for Black Tea Based on a Multi-Objective Optimization Algorithm

[0062] A digital blending method for black tea based on a multi-objective optimization algorithm, the schematic flow chart of which is as Figure 1 shown, comprising the following steps:

[0063] S1. Based on the Tea Sensory Evaluation Method (GD / T 23776-2018) and the Vocabulary of Chinese Tea Sensory Evaluation (GB / T 14487-2017), use sensory evaluation to score each blending raw material (tea leaves) to obtain the index parameters of each blending raw material, including the appearance score of the tea leaves of each blending raw material, the taste score of the tea soup, the color score of the tea soup, the aroma score of the tea soup, and the market price;

[0064] S2. Use the PlatEMO multi-objective optimization framework to transform the tea blending problem into a multi-objective optimization problem that takes into account both quality and cost, specifically including the following steps:

[0065] S21. Convert the appearance score of the tea leaves, the taste score of the tea soup, the color score of the tea soup, and the aroma score of the tea soup of each blending raw material in the index parameters obtained in step S1 into quality evaluation factors according to formula 1 respectively;

[0066] Formula 1:

[0067] where S i is the index parameter of the i-th blending raw material, is the average value of the index parameters among all blending raw materials, and σ s is the standard deviation of this index parameter among all blending raw materials;

[0068] Then calculate the quality deviation objective function Q according to formula 2;

[0069] Formula 2:

[0070] where S i1 、S i2 、S i3 and S i4 are all calculated by formula 1, x i is the blending ratio of the i-th blending raw material, S i1 is the S i of the appearance score of the tea leaves, S i2 is the S i of the taste score of the tea soup, S i3 is the S i of the color score of the tea soup, S i4 is the S i of the aroma score of the tea soup, S target is the index parameter after z-score standardization;

[0071] S22. According to the market price in the index parameters obtained in step S1, combined with the blending ratio of each blending raw material, calculate the cost P according to formula 5;

[0072] Formula 5: where $c_i$ is the market selling price of the $i$-th blended raw material, and $x_i$ is the blending ratio of the $i$-th blended raw material;

[0073] S23. Compare the quality deviation objective function $Q$ obtained in step S21 and the cost $P$ obtained in step S22, and select the minimum value among them;

[0074] S24. Set the following constraint conditions: $Q\leq\delta$, $a_i\leq x$ i $\leq b$ i ;

[0075] where $\delta$ is the quality deviation threshold, $x$ i is the blending ratio of the $i$-th blended raw material, $a$ i is the lower limit of the blending ratio of the $i$-th blended raw material, and $b$ i is the upper limit of the blending ratio of the $i$-th blended raw material;

[0076] S3. Input the index parameters obtained in step S1 into the GDE3 algorithm for processing, output the model operation results, and obtain the Pareto front solution set, which specifically includes the following steps:

[0077] S31. Population initialization: Generate random solutions based on a preset population size (100) as the initial population;

[0078] S32. Differential operation: Randomly select three random solutions $x$ 1 , $x$ 2 and $x$ 3 from the initial population in step S31, and perform differential calculation according to formula 3 to obtain a new solution $v$ i ;

[0079] Formula 3: $v$ i $=x$ 1 $+F(x$ 2 $-x$ 3 ); where $F$ is the scaling factor, and its value range is $[0,1]$;

[0080] S33. Crossover operation: Perform a crossover operation on the new solution $v_i$ obtained in step S32 and $x_i$ according to formula 4 to obtain $\mu_i$;

[0081] Formula 4:

[0082] where $CR$ is the crossover rate, $rand(0,1)$ is a random value from a uniform distribution in the interval $[0,1]$, $i$ is the $i$-th solution vector, $j$ is the $j$-th dimension, $D$ is the number of dimensions of the solution vector, and $x_i$ is the blending ratio of the $i$-th blended raw material;

[0083] S34. Fitness calculation: For \(x_i\) and \(\mu_i\) obtained in step S33, calculate the fitness values \(F(x_i)\) and \(F(\mu_i)\). If \(F(\mu_i)\lt F(x_i)\), then replace \(x_i\) with \(\mu_i\).

[0084] S35. Repeat steps S31 - S34 until the maximum number of iterations (10,000 times) is reached or the function value converges.

[0085] S4. Combine the Pareto front solution set obtained in step S3 with subjective preferences to obtain the black tea blending plan, which specifically includes the following steps:

[0086] S41. Set the following constraint conditions: Q ≤ δ, a i ≤ x i ≤ b i , and obtain the subjective preference plan.

[0087] where δ is the quality deviation threshold, \(x\) i is the blending ratio of the \(i\) - th blending raw material, \(a\) i is the lower limit of the blending ratio of the \(i\) - th blending raw material, and \(b\) i is the upper limit of the blending ratio of the \(i\) - th blending raw material.

[0088] S42. On the basis of the Pareto front solution set obtained in step S3, add the subjective preference plan as the final optimization result of the algorithm.

[0089] Example 2 Application of a Digital Blending Method for Black Tea Based on a Multi - objective Optimization Algorithm

[0090] I. Experimental Method

[0091] Take 8 different black teas (denoted as B1 - B8) from Yingde City, Guangdong Province as black tea blending raw materials. The prices of the 8 black teas are between 100 and 400 yuan per catty.

[0092] A team composed of 10 black tea evaluation experts (including 5 males and 5 females, aged between 25 and 45 years old) scored and evaluated the sensory quality (tea leaf appearance, tea soup taste, tea soup color, and tea soup aroma) of the 8 black tea blending raw materials according to the Tea Sensory Evaluation Method (GB / T 23776 - 2018) and the Chinese Tea Sensory Evaluation Vocabulary (GB / T 14487 - 2017); the sensory quality scoring results and prices of the 8 black tea blending raw materials are shown in Table 1.

[0093] Table 1 Sensory Quality Scoring Results and Prices of 8 Black Tea Blending Raw Materials

[0094]

[0095] Using the 8 kinds of black tea blending raw materials shown in Table 1, combining with the black tea digital blending method shown in Example 1, and respectively setting the subjective preference (target score) in step S4 to 87-92, a black tea blending scheme is obtained.

[0096] II. Experimental Results

[0097] In the embodiment of the present invention, the scatter plot of the Pareto front solutions in the tea blending process is as Figure 2 shown, and the result diagram of the Pareto front solution set is as Figure 3 shown; the result of the black tea blending scheme is shown in Table 2, and the result diagram of the blending scheme solution set is as Figure 4 shown.

[0098] Table 2 Results of the Black Tea Blending Scheme

[0099]

[0100] Note: The percentages of each tea in the table are the mass percentages of the tea in the blending scheme.

[0101] The results show that the manual review score is close to the target score, proving that the black tea digital blending method based on the multi-objective optimization algorithm is successful.

[0102] Example 3 A Black Tea Digital Blending Model Based on a Multi-Objective Optimization Algorithm

[0103] A black tea digital blending model based on a multi-objective optimization algorithm includes a data acquisition module, a blending module, and a result output module;

[0104] The data acquisition module is used to obtain the index parameters of the blending raw materials (tea leaves); among them, the index parameters of the blending raw materials include the appearance score of the tea leaves, the taste score of the tea soup, the color score of the tea soup, the aroma score of the tea soup, and the market price. The acquisition standards of each index parameter are the tea sensory evaluation method (GB / T 23776-2018) and the Chinese tea sensory evaluation vocabulary (GB / T14487-2017);

[0105] The blending module, based on the index parameters of the blending raw materials obtained by the data acquisition module, executes the black tea digital blending method shown in Example 1, and obtains a black tea blending method based on the target score;

[0106] The result output module is used to output the black tea blending scheme obtained by the blending module.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than limit the protection scope of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description and ideas. It is not necessary and impossible to list all implementation manners here. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A black tea digital blending method based on a frontier solution set multi-objective optimization algorithm, characterized in that: The following steps are involved: S1. Obtaining the index parameters of each blending raw material; the index parameters include tea appearance score, tea taste score, tea soup color score, tea aroma score and market price; S2. Use PlatEMO multi-objective optimization framework to transform the tea blending problem into a multi-objective optimization problem that takes both quality and cost into consideration; S3. Input the index parameters obtained in step S1 into the GDE3 algorithm for processing, output the model operation results, and obtain the Pareto frontier solution set; S4. Combine the Pareto frontier solution set obtained in step S3 with the subjective preference to obtain a black tea blending plan.

2. The digital blending method for black tea according to claim 1, characterized in that: In step S1, the index parameters of each blended raw material are obtained through sensory evaluation.

3. The digital blending method for black tea according to claim 2, characterized in that: The sensory evaluation reference standards are GB / T 23776-2018 and GB / T 14487-2017.

4. The digital blending method for black tea according to claim 1, characterized in that: The PlatEMO multi-objective optimization framework described in step S2 includes the following steps: S21. The tea appearance score, tea soup taste score, tea soup color score and tea soup aroma score in the index parameters obtained in step S1 are converted into quality evaluation factors according to formula 1; Formula 1: Among them, S i is the index parameter of the i-th blended raw material, is the average value of the index parameter in all blended raw materials, σ s is the standard deviation of the parameter in all blended ingredients; Then calculate the quality deviation Q according to formula 2; Formula 2: Among them, S i1 , S i2 , S i3 and S i4 All are calculated by formula 1, x i is the proportion of the i-th blending raw material, S i1 S for tea appearance i , S i2 S for tea taste i , S i3 S for tea soup color i , S i4 S for tea aroma i , S target This is the indicator parameter after z-score standardization; S22. Calculate the cost P based on the market price of the index parameter obtained in step S1 and the proportion of each blended raw material; S23. Compare the quality deviation objective function Q obtained in step S21 and the cost P obtained in step S22, and select the minimum value; S24. Set the following constraints: Q≤δ、a i ≤x i ≤ b i ; Where δ is the quality deviation threshold, x i is the proportion of the i-th blending raw material, a i is the lower limit of the proportion of the i-th blending raw material, b i is the upper limit of the blending ratio of the i-th blending raw material.

5. The digital blending method for black tea according to claim 1, characterized in that: In step S3, the process of inputting the index parameters obtained in step S1 into the GDE3 algorithm includes the following steps: S31. Population initialization: Generate random solutions as the initial population based on the preset population size; S32. Differential operation: Randomly select three random solutions x1, x2 and x3 from the initial population of step S31 and perform differential calculation according to formula 3 to obtain a new solution v i ; Formula 3: v i =x1+F(x2-x3); where F is the scaling factor, and its value range is [0,1]; S33. Crossover operation: The new solution v obtained in step S32 is i With x i Perform the crossover operation according to Formula 4 to obtain μ i ; Formula 4: Where CR is the crossover rate, rand(0,1) is a random value from a uniform distribution in the interval [0,1], i is the i-th solution vector, j is the j-th dimension, D is the number of dimensions of the solution vector, and x i is the blending ratio of the i-th blending raw material; S34. Fitness calculation: for x i and μ obtained in step S33 i , calculate the fitness F(x i ) and F(μ i ), F(μ i )<F(x i ), then use μ i Replace x i ; S35. Repeat steps S31 to S34 until the maximum number of iterations is reached or the function value converges.

6. The black tea digital blending method according to claim 1, characterized in that: The step S4 of combining the Pareto frontier solution set obtained in step S3 with the subjective preference specifically includes the following steps: S41. Set the following constraints: Q≤δ、a i ≤x i ≤ b i , get the subjective preference scheme; Where δ is the quality deviation threshold, x i is the proportion of the i-th blending raw material, a i is the lower limit of the proportion of the i-th blending raw material, b i is the upper limit of the blending ratio of the i-th blending raw material; S42. Based on the Pareto frontier solution set obtained in step S3, the subjective preference scheme is added as the final optimization result of the algorithm.

7. A black tea digital blending model based on a frontier solution set multi-objective optimization algorithm, characterized in that: It includes data acquisition module, assembly module and result output module; The data acquisition module is used to obtain the index parameters of the blended raw materials; the index parameters include the tea appearance score, tea soup taste score, tea soup color score, tea soup aroma score and market price; The blending module executes the black tea digital blending method according to claims 1 to 6 based on the index parameters of the blending raw materials obtained by the data acquisition module to obtain a black tea blending plan; The result output module is used to output the black tea blending scheme obtained by the blending module.

8. The digital blending model of black tea according to claim 7, characterized in that: The blending raw material is tea leaves.

9. The digital black tea blending model according to claim 7, characterized in that: The acquisition standards of the index parameters of the blended raw materials are GB / T 23776-2018 and GB / T 14487-2017.

10. Use of the digital black tea blending model according to any one of claims 7 to 9 in blending black tea.

Images