A method and apparatus for designing a sheet drying process
By using a model prediction system and adjusting process parameters, the problem that traditional thin-plate drying processes cannot simultaneously address the sensory and physical quality of cigarettes was solved, thus achieving personalized drying effect optimization for cigarette brands.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA TOBACCO HENAN IND CO LTD
- Filing Date
- 2023-06-09
- Publication Date
- 2026-05-19
AI Technical Summary
In existing technologies, traditional single-stage thin-plate drying processes are difficult to balance the sensory and physical quality of cigarette brands, and lack the design of thin-plate drying modes for different cigarette brands.
Based on the pattern prediction model, by inputting the physicochemical indicators of the flavorings of the target cigarette brand, the optimal single-stage or two-stage thin-plate drying process mode is predicted and determined, and the process parameters are adjusted to improve the targeting and drying effect.
This has enabled targeted improvement in the thin-plate drying process and drying effect for different cigarette brands, and optimized the balance between the sensory and physical quality of cigarettes.
Smart Images
Figure CN116687045B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of cigarette manufacturing technology, and more specifically, to a design method and apparatus for a thin-plate drying process. Background Technology
[0002] Thin-plate drying is a key process in the tobacco processing industry. Its functions are twofold: first, to remove some moisture from the tobacco leaves, improving their filling capacity and processing resistance; and second, to enhance the aroma and flavor of the cigarettes, improving sensory comfort and overall sensory quality. Production practice and extensive research have shown that low-temperature drying processes retain more of the tobacco's natural aroma, while high-temperature drying processes better improve the filling capacity and processing resistance of the tobacco leaves. Traditional single-stage thin-plate drying processes often present a trade-off in resolving this contradiction. Against this backdrop, a two-stage thin-plate drying process has emerged. The main characteristic of this technology is the establishment of two independently controlled drying zones within the thin-plate drying drum. The front section of the drum experiences high-temperature, rapid dehydration and expansion, improving the filling capacity of the tobacco leaves; the rear section experiences low-temperature, slow dehydration and aroma preservation, improving the sensory quality of the tobacco leaves. This achieves a balance and harmony between the sensory and physical quality of the tobacco leaves.
[0003] Existing research on variable temperature drying includes studies on kinetic characteristics, physical and chemical quality change characteristics, and control methods for segmented variable temperature drying in drums. However, there is limited research on the processing characteristics of thin-plate drying for different cigarette brands, and no relevant data on the design of thin-plate drying modes for different cigarette brands has been found. Summary of the Invention
[0004] This application provides a design method and apparatus for thin plate drying process. Based on the thin plate drying processing characteristics of different cigarette brands, a pattern prediction model is trained. The optimal thin plate drying mode suitable for the target cigarette brand is obtained by using the pattern prediction model, and the process parameters under the optimal mode are determined, thereby improving the targeting and drying effect of the thin plate drying process for each cigarette brand.
[0005] This application provides a design method for a thin plate drying process, including:
[0006] Input the target physicochemical indicators of flavorings and fragrances in the target cigarette brand into the pattern prediction model to obtain the optimal mode for thin plate drying of the target cigarette brand. The optimal mode is one of the single-stage thin plate drying process or the two-stage thin plate drying process.
[0007] Determine the optimal process parameters to develop a thin-plate drying solution for the target cigarette brand.
[0008] Preferably, training the pattern prediction model includes:
[0009] Determine the target physicochemical indicators of flavorings and fragrances, as well as the key evaluation elements for evaluating process modes;
[0010] A training set and a test set were constructed. The training set and the test set included the comprehensive evaluation results of different cigarette brands under different thin plate drying modes based on key evaluation elements, as well as the optimal mode for each cigarette brand. The values of the target physicochemical indicators corresponded to the cigarette brands.
[0011] The pattern prediction model is trained using the training and test sets.
[0012] Preferably, the key evaluation elements for evaluating the process mode include the sensory quality evaluation score of the finished cigarette and the tobacco filling value score, and the comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
[0013] Preferably, when constructing the training and test sets, the method for obtaining the score of the tobacco filling value includes:
[0014] At the process exit of the drum used for drying thin plates, multiple tobacco samples are prepared at preset time intervals.
[0015] Each tobacco sample was subjected to a filling test to obtain the filling test results;
[0016] The final filling value is obtained based on all filling test results and is used as the score for tobacco filling.
[0017] Preferably, when constructing the training and test sets, the method for obtaining the sensory quality evaluation scores of the finished cigarettes includes:
[0018] Prepare samples of tobacco shreds for machine rolling;
[0019] The prepared tobacco samples were balanced according to the tobacco storage process requirements and then rolled on a machine to obtain cigarettes.
[0020] According to the sensory quality evaluation requirements, the cigarettes were prepared into cigarette evaluation samples;
[0021] Multiple sensory quality evaluation results are obtained for the cigarette evaluation samples, and a sensory quality evaluation score is obtained based on the multiple sensory quality evaluation results.
[0022] This application also provides a design device for a thin plate drying process, including an optimal mode acquisition module and a process parameter determination module;
[0023] The optimal mode acquisition module is used to input the target physicochemical indicators of flavorings and fragrances in the target cigarette brand into the mode prediction model to obtain the optimal mode for thin plate drying of the target cigarette brand. The optimal mode is one of the single-stage thin plate drying process or the two-stage thin plate drying process.
[0024] The process parameter determination module is used to determine the process parameters under the optimal mode, thereby forming a thin-plate drying scheme for the target cigarette brand.
[0025] Preferably, the design device further includes a model acquisition module, which includes an indicator and feature determination module, a dataset construction module, and a model training module;
[0026] The indicator and element determination module is used to determine the target physicochemical indicators of flavorings and fragrances, as well as the key evaluation elements for evaluating process modes.
[0027] The dataset construction module is used to build training and test sets. The training and test sets include the comprehensive evaluation results of different cigarette brands under different thin plate drying modes based on key evaluation factors, as well as the optimal mode for each cigarette brand. The values of the target physicochemical indicators correspond to the cigarette brands.
[0028] The model training module is used to train the pattern prediction model using the training set and the test set.
[0029] Preferably, the key evaluation elements for evaluating the process mode include the sensory quality evaluation score of the finished cigarette and the tobacco filling value score, and the comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
[0030] Preferably, when the dataset construction module obtains the score of the tobacco filling value, it first receives the filling detection results obtained by performing filling detection on multiple tobacco samples, and then obtains the final filling value based on all the filling detection results, which is used as the score of the tobacco filling value.
[0031] The tobacco samples were prepared at the process exit of the drum used for drying thin plates, at preset time intervals.
[0032] Preferably, the dataset construction module obtains the sensory quality evaluation scores of the finished cigarettes based on the sensory quality evaluation results of multiple cigarette evaluation samples.
[0033] In preparing cigarette evaluation samples, the process begins by preparing machine-rolled tobacco samples, then balancing the prepared tobacco samples according to the tobacco storage process requirements, and finally rolling them into cigarettes. The cigarettes are then prepared into cigarette evaluation samples according to the sensory quality evaluation requirements.
[0034] Other features and advantages of this application will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description
[0035] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments of the present application and, together with their description, serve to explain the principles of the present application.
[0036] Figure 1 A flowchart illustrating the design method for the thin plate drying process provided in this application;
[0037] Figure 2 A flowchart of the training pattern prediction model provided for this application;
[0038] Figure 3 A comparison chart of the patterns and corresponding predicted probabilities of brand E obtained using the test set and the actual results provided in this application;
[0039] Figure 4 A structural diagram of the apparatus for the thin plate drying process provided in this application. Detailed Implementation
[0040] Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values of the components and steps set forth in these embodiments do not limit the scope of the present application.
[0041] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.
[0042] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the specification.
[0043] In all the examples shown and discussed herein, any specific values should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.
[0044] This application provides a design method and apparatus for thin plate drying process. Based on the thin plate drying processing characteristics of different cigarette brands, a pattern prediction model is trained. The optimal thin plate drying mode suitable for the target cigarette brand is obtained by using the pattern prediction model, and the process parameters under the optimal mode are determined, thereby improving the targeting and drying effect of the thin plate drying process for each cigarette brand.
[0045] like Figure 1 As shown, the design method for the thin plate drying process provided in this application includes:
[0046] S110: Input the target physicochemical indicators of flavorings and fragrances in the target cigarette brand into the pattern prediction model to obtain the optimal mode for thin-plate drying processing of the target cigarette brand.
[0047] The input data for the pattern prediction model is the target physicochemical indicators of flavorings and fragrances in the target cigarette brand. The model obtains the probability of the target cigarette brand adopting each thin-plate drying mode, and the thin-plate drying mode corresponding to the maximum probability value is taken as the optimal mode.
[0048] It should be noted that the model prediction includes two thin plate drying processes: a single-stage thin plate drying process and a two-stage thin plate drying process, with each process including at least one mode. The optimal mode output by the model is one of the single-stage or two-stage thin plate drying processes.
[0049] It should be noted that in the two-stage thin plate drying process, only the cylinder wall temperature of one stage (the front or rear stage) can be adjusted, and the hot air velocity of the entire drum can also be adjusted. Specifically, the cylinder wall temperature of the front stage is 125℃-145℃, and the cylinder wall temperature of the rear stage is 120℃-135℃, with a hot air velocity of 0.4-0.8 m / s. As an example, in the model prediction system, the two-stage thin plate drying process includes two modes: Mode 1, where the cylinder wall temperature of the front stage is higher than that of the rear stage, the hot air velocity of the drum is fixed, and the cylinder wall temperature of the rear stage is fixed; the cylinder wall temperature of the front stage is adjusted during the thin plate drying process. Mode 2, where the cylinder wall temperature of the front stage is higher than that of the rear stage, the cylinder wall temperatures of both the front and rear stages are fixed, and the hot air velocity of the drum is adjusted during the thin plate drying process. The single-stage thin plate drying process includes one mode (denoted as Mode 3), where the cylinder wall temperature of the drum is fixed, and the hot air velocity of the drum is adjusted during the thin plate drying process.
[0050] S120: Determine the process parameters under the optimal mode to form a thin-plate drying solution for the target cigarette brand.
[0051] The optimal process parameters can be determined using a trained neural network model or existing methods. Once the process parameters are determined, the thin plate drying process can be implemented according to these parameters.
[0052] Specifically, such as Figure 2 As shown, training a pattern prediction model includes the following steps:
[0053] S210: Determine the target physicochemical indicators for flavorings and fragrances, as well as the key evaluation elements used to evaluate the process mode. The values of the target physicochemical indicators correspond to the cigarette brand, and the key evaluation elements are used to calculate the comprehensive evaluation result.
[0054] In the tobacco product processing, flavorings and fragrances primarily serve to enhance or modify the style or improve the quality of tobacco products, improving tobacco aroma, increasing toughness, enhancing moisture retention, improving combustibility, and reducing breakage. The main physicochemical properties of flavorings and fragrances include relative density, refractive index, acid value, and total volatile components in the tobacco liquor. Therefore, as an example, the target physicochemical properties of flavorings and fragrances include relative density, refractive index, acid value, and total volatile components in the tobacco liquor.
[0055] Under different tobacco drying process conditions (mainly reflected in the drying temperature and residence time of the tobacco in the drying equipment), the total amount of volatile components in the tobacco liquid retained in the tobacco product will vary significantly, directly affecting the sensory smoking quality and tobacco breakage of the cigarette product. Changes in other physicochemical indicators have a smaller impact on sensory smoking quality and tobacco breakage. Therefore, the total amount of volatile components in the tobacco liquid is a key physicochemical indicator determining the sensory quality and breakage of the cigarette product, and sensory quality is an important indicator for distinguishing different cigarette brands. Based on this, preferably, the target physicochemical indicator for flavorings and fragrances is the total amount of volatile components in the tobacco liquid. In this application, different total amounts of volatile components in the tobacco liquid correspond to different cigarette brands.
[0056] As an example, the key evaluation elements for evaluating the process mode include the sensory quality evaluation score and the tobacco filling value score of the finished cigarette, and the comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
[0057] S220: Construct training and testing sets, which include comprehensive evaluation results of different cigarette brands under different thin-plate drying modes based on key evaluation factors, as well as the optimal mode for that cigarette brand. The values of the target physicochemical indicators correspond to the cigarette brand.
[0058] Specifically, MATLAB design software is used to create an experimental plan for the pattern prediction model, and then training and test sets are constructed.
[0059] When constructing the training and testing sets, for each cigarette brand, the operation is performed after the parameter indicators have stabilized under each thin plate drying mode to obtain key evaluation elements. The comprehensive evaluation result is calculated based on the key evaluation elements. Finally, the thin plate drying mode corresponding to the maximum value of the comprehensive evaluation result obtained by the cigarette brand under all thin plate drying modes is taken as the best mode for the cigarette brand.
[0060] As an example, when constructing the training and test sets, the method for obtaining the tobacco filling value score for each thin-plate drying mode includes:
[0061] P1: At the process exit of the roller used for drying thin plates, multiple tobacco samples are prepared at preset time intervals (e.g., 5 minutes).
[0062] P2: Perform a filling test on each tobacco sample to obtain the filling test results.
[0063] P3: The final filling value is obtained based on all filling test results and is used as the score of tobacco filling value, denoted as X2.
[0064] As an example, the average value of all fill detection results is used as the final fill value.
[0065] As another embodiment, after obtaining the average value of all filling test results, the difference between this average value and the standard value is calculated as the change, and then the final filling value is determined based on the change. As an example, the baseline score is 50 points; 0 < change < 0.2 cm. 3 For a value of / g, add 20 points to the filler value; for a change ≥0.2cm 3 When / g, add 40 points to the fill value score. -0.2cm 3 When / g < change < 0, deduct 20 points from the score for the fill-in-the-blank value; when the change ≤ -0.2cm 3 When the value is / g, the score for the filler value is reduced by 40 points. As shown in Table 1.
[0066] Table 1 Experimental Plan for Two-Stage Thin Plate Drying Process
[0067]
[0068] As an example, when constructing the training and test sets, the method for obtaining the sensory quality evaluation score of the finished cigarettes under each thin-plate drying mode includes:
[0069] Q1: Prepare samples of tobacco shreds for rolling on a machine.
[0070] Q2: The prepared tobacco sample is balanced according to the tobacco storage process requirements and then rolled on a machine to obtain cigarettes.
[0071] Q3: Prepare cigarette evaluation samples according to the sensory quality evaluation requirements.
[0072] Q4: Obtain multiple sensory quality evaluation results for the cigarette evaluation samples, and obtain the sensory quality evaluation score based on the multiple sensory quality evaluation results, denoted as X1.
[0073] As an example, when conducting sensory quality evaluation, there should be no fewer than 15 sensory evaluators, and the sensory evaluators should conduct independent blind evaluations and scoring.
[0074] As an example, the average score of all sensory judges' evaluations is used as the sensory quality evaluation score.
[0075] As another embodiment, after obtaining the average score of all sensory judges, the difference between this average and the standard value is calculated as the variation, and then the final fill value is determined based on the variation. For example, the baseline score is 50 points; when 0 < variation < 0.5 points, the sensory quality evaluation score is increased by 20 points; when the variation ≥ 0.5 points, the sensory quality evaluation score is increased by 40 points. When -0.5 points < variation < 0, the sensory quality evaluation score is decreased by 20 points; when the variation ≤ -0.5 points, the sensory quality evaluation score is decreased by 40 points. See Table 1.
[0076] After obtaining the sensory quality evaluation score and the tobacco filling value score, the comprehensive evaluation result is calculated by weighted average and denoted as Y.
[0077] As an example, as shown in Table 1, with sensory quality weighted at 70% and tobacco filling value weighted at 30%, the overall evaluation result is as follows:
[0078] Y = X1 * 70% + X2 * 30%
[0079] After obtaining the training and test sets, ANOVA analysis of variance was performed on the sensory quality evaluation scores, tobacco filling scores, and comprehensive evaluation results. The significance test results showed that the p-values were all less than 0.5, indicating that the model was effective.
[0080] In both the training and test sets, for each cigarette brand, the best pattern for that brand is marked as 1, and other patterns are marked as 0.
[0081] S230: Train the pattern prediction model using the training and test sets.
[0082] Specifically, in the example shown in Table 1, brands A and B are used as the training set, and brands D, E, and F are used as the test set. The total evaporation of liquid from all brands in the training set, the comprehensive evaluation results of all thin-plate drying modes, and the best mode for each brand are used as inputs to train the mode prediction model.
[0083] As an example, the pattern prediction model is trained as a BP neural network.
[0084] After the pattern prediction model is trained, the input data of the test set is imported into the pattern prediction model to predict the probability of each brand under each thin plate drying mode. Figure 3The test results for brand E are shown. The lower curve represents the actual results, and the upper curve represents the predicted results. In practice, brand E is more suitable for mode 2, corresponding to type "2" on the horizontal axis (marked as 1, i.e., the vertical axis is 1) in the graph, while mode 1 (corresponding to type "1" on the horizontal axis) and mode 3 (corresponding to type "3" on the horizontal axis) are marked as 0. The model's predicted probability for mode 2 is around 1.5, closer to 1, while the predicted probabilities for modes 1 and 3 are closer to 0. This prediction result matches the actual situation, therefore it is feasible and can be generalized for analysis and application.
[0085] Based on the above-described design method for thin plate drying process, this application also provides a design apparatus for thin plate drying process. For example... Figure 4 As shown, the design device for the thin plate drying process includes an optimal mode acquisition module 410 and a process parameter determination module 420.
[0086] The optimal mode acquisition module 410 is used to input the target physicochemical indicators of the flavorings and fragrances of the target cigarette brand into the mode prediction model to obtain the optimal mode for thin plate drying of the target cigarette brand. The optimal mode is one of the single-stage thin plate drying process or the two-stage thin plate drying process.
[0087] The process parameter determination module 420 is used to determine the process parameters in the optimal mode to form a thin plate drying scheme for the target cigarette brand.
[0088] Preferably, the design device further includes a model acquisition module 430, which includes an indicator and element determination module 4301, a dataset construction module 4302, and a model training module 4303.
[0089] The indicator and element determination module 4301 is used to determine the target physicochemical indicators of flavorings and fragrances, as well as the key evaluation elements for evaluating process modes.
[0090] The dataset construction module 4302 is used to construct training and testing sets. The training and testing sets include the comprehensive evaluation results of different cigarette brands under different thin plate drying modes based on key evaluation factors, as well as the optimal mode for each cigarette brand. The values of the target physicochemical indicators correspond to the cigarette brands.
[0091] The model training module 4303 is used to train the pattern prediction model using the training set and the test set.
[0092] Preferably, the key evaluation elements for evaluating the process mode include the sensory quality evaluation score of the finished cigarette and the tobacco filling value score, and the comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
[0093] Based on this, when the dataset construction module 4302 obtains the score of the tobacco filling value, it first receives the filling detection results obtained by performing filling detection on multiple tobacco samples, and then obtains the final filling value based on all the filling detection results, which is used as the score of the tobacco filling value.
[0094] The tobacco samples were prepared at the process exit of the drum used for drying thin plates, at preset time intervals.
[0095] When the dataset construction module 4302 obtains the sensory quality evaluation score of the finished cigarette, it does so based on the sensory quality evaluation results of multiple cigarette evaluation samples.
[0096] In preparing cigarette evaluation samples, the process begins by preparing machine-rolled tobacco samples, then balancing the prepared tobacco samples according to the tobacco storage process requirements, and finally rolling them into cigarettes. The cigarettes are then prepared into cigarette evaluation samples according to the sensory quality evaluation requirements.
[0097] This application can select suitable thin-plate drying process modes for cigarette brands with different thin-plate drying characteristics, providing a rapid and optimal method for the development of new cigarette products and the transformation of old products.
[0098] While specific embodiments of this application have been described in detail by way of examples, those skilled in the art should understand that the above examples are for illustrative purposes only and are not intended to limit the scope of this application. Those skilled in the art should understand that modifications can be made to the above embodiments without departing from the scope and spirit of this application. The scope of this application is defined by the appended claims.
Claims
1. A design method for a thin plate drying process, characterized in that, include: The target physicochemical indicators of flavorings and fragrances in the target cigarette brand are input into the pattern prediction model to obtain the optimal mode for thin plate drying of the target cigarette brand. The optimal mode is one of a single-stage thin plate drying process or a two-stage thin plate drying process. Determine the process parameters under the optimal mode to form a thin-plate drying scheme for the target cigarette brand; The process of training the pattern prediction model includes: determining the target physicochemical indicators of flavorings and fragrances, as well as key evaluation elements for evaluating process modes; constructing a training set and a test set, wherein the training set and the test set include the comprehensive evaluation results obtained by different cigarette brands under different thin-plate drying modes based on the key evaluation elements and the optimal mode for each cigarette brand, wherein the values of the target physicochemical indicators correspond to the cigarette brands; and training the pattern prediction model using the training set and the test set. The key evaluation elements used to evaluate the process mode include the sensory quality evaluation score of the finished cigarette and the tobacco filling value score. The comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
2. The design method for the thin plate drying process according to claim 1, characterized in that, When constructing the training set and the test set, the method for obtaining the tobacco filling value score includes: At the process exit of the drum used for drying thin plates, multiple tobacco samples are prepared at preset time intervals. Each of the tobacco samples was subjected to a filling test to obtain the filling test results; The final filling value is obtained based on all filling test results and is used as the score for the tobacco filling value.
3. The design method for the thin plate drying process according to claim 1, characterized in that, The method for obtaining the sensory quality evaluation score of the finished cigarettes when constructing the training set and the test set includes: Prepare samples of tobacco shreds for machine rolling; The prepared tobacco samples were balanced according to the tobacco storage process requirements and then rolled on a machine to obtain cigarettes. According to the sensory quality evaluation requirements, the cigarettes were prepared into cigarette evaluation samples; Multiple sensory quality evaluation results are obtained for the cigarette evaluation sample, and a sensory quality evaluation score is obtained based on the multiple sensory quality evaluation results.
4. A design apparatus for a thin plate drying process, characterized in that, Includes an optimal mode acquisition module and a process parameter determination module; The optimal mode acquisition module is used to input the target physicochemical indicators of the flavorings and fragrances in the target cigarette brand into the mode prediction model to obtain the optimal mode for thin plate drying processing of the target cigarette brand. The optimal mode is one of a single-stage thin plate drying process or a two-stage thin plate drying process. The process parameter determination module is used to determine the process parameters under the optimal mode to form a thin-plate drying scheme for the target cigarette brand. It also includes a model acquisition module, which includes an indicator and feature determination module, a dataset construction module, and a model training module; The indicator and element determination module is used to determine the target physicochemical indicators of flavorings and fragrances, as well as the key evaluation elements for evaluating process modes. The dataset construction module is used to construct training and testing sets. The training and testing sets include comprehensive evaluation results obtained by different cigarette brands under different thin-plate drying modes based on the key evaluation elements, as well as the optimal mode for each cigarette brand. The values of the target physicochemical indicators correspond to the cigarette brands. The model training module is used to train the pattern prediction model using the training set and the test set; The key evaluation elements used to evaluate the process mode include the sensory quality evaluation score of the finished cigarette and the tobacco filling value score. The comprehensive evaluation result is obtained based on the sensory quality evaluation score and the tobacco filling value score.
5. The design apparatus for the thin plate drying process according to claim 4, characterized in that, When the dataset construction module obtains the score of the tobacco filling value, it first receives the filling detection results obtained by performing filling detection on multiple tobacco samples, and then obtains the final filling value based on all the filling detection results, which is used as the score of the tobacco filling value. The tobacco sample is prepared at the process exit of the drum used for drying thin plates, at preset time intervals.
6. The apparatus for the thin plate drying process according to claim 4, characterized in that, The dataset construction module obtains the sensory quality evaluation scores of finished cigarettes based on multiple sensory quality evaluation results of cigarette evaluation samples. In preparing the cigarette evaluation sample, firstly, a sample of rolled tobacco is prepared, then the prepared tobacco sample is balanced according to the tobacco storage process requirements, and rolled on a machine to obtain a cigarette. Finally, the cigarette is prepared into the cigarette evaluation sample according to the sensory quality evaluation requirements.