Cigarette material type selection method and device of cigarette product and computer equipment

Abstract

The invention relates to a cigarette material type selection method and device for cigarette products and computer equipment, and the method comprises the steps: respectively training various types of cigarette material prediction models based on a historical data set of the cigarette products; sequentially inputting the obtained design parameters and default index parameters into each cigarette material prediction model for model selection prediction, and determining a model selection result of each cigarette material index based on a prediction result; wherein the model selection result of the previous cigarette material indexes is used for updating default index parameters so as to input a subsequently called cigarette material prediction model for model selection prediction; based on the model selection result of each cigarette material index, the model selection result of the cigarette product is obtained, intelligent and automatic cigarette material model selection is realized, the problems of low model selection efficiency and precision are solved, the experiment frequency and time cost are reduced, and the research and development and production cost is also reduced.

Description

Technical Field

[0001] This application relates to the technical field of cigarette product design, and particularly to a method, device, and computer equipment for selecting cigarette material for cigarette products. Background Art

[0002] In the design of cigarette products, the selection of cigarette materials is a crucial link, directly affecting product quality, user experience, and health risks.

[0003] Traditional methods mainly rely on the experience and subjective judgment of experts, with problems such as low efficiency and insufficient accuracy. Existing experimental testing methods, such as orthogonal design for multiple combination experiments to determine the optimal solution, are time-consuming and difficult to meet the complex and changing market demands.

[0004] Regarding the problems of low efficiency and poor accuracy in the selection of cigarette materials in related technologies, no effective solutions have been proposed yet. Summary of the Invention

[0005] In this embodiment, a method, device, and computer equipment for selecting cigarette materials for cigarette products are provided to solve the problems of low efficiency and poor accuracy in the selection of cigarette materials in related technologies.

[0006] In a first aspect, in this embodiment, a method for selecting cigarette materials for cigarette products is provided. The method includes:

[0007] Obtain a historical data set of cigarette products, where the historical data set includes relevant data on cigarette material indicators, smoke indicators, and cut tobacco formula indicators;

[0008] Based on the historical data set, train corresponding cigarette material prediction models for different types of the cigarette material indicators respectively;

[0009] Input the obtained design parameters and default index parameters into each of the cigarette material prediction models in sequence for selection prediction, and determine the selection results of each of the cigarette material indicators based on the prediction results; among them, the selection results of the prior cigarette material indicators are used to update the default index parameters for input into the subsequent called cigarette material prediction models for selection prediction;

[0010] Based on the selection results of each of the cigarette material indicators, obtain the selection result of the cigarette product.

[0011] In some of the embodiments, based on the historical data set, training corresponding cigarette material prediction models for different types of the cigarette material indicators respectively includes:

[0012] Determine a to-be-trained index among the cigarette material indexes, use the to-be-trained index as a label, and use the cigarette material indexes, the smoke indexes, and the cut tobacco formula indexes other than the to-be-trained index as features to train a cigarette material prediction model corresponding to the current to-be-trained index;

[0013] Replace the to-be-trained index, and repeat the above training process until all categories of cigarette material prediction models are trained.

[0014] In some embodiments, input the obtained design parameters and default index parameters into each of the cigarette material prediction models in sequence for model selection prediction, and determine the model selection results of each of the cigarette material indexes based on the prediction results, including:

[0015] Obtain design parameters and default index parameters; the default index parameters include the default parameter values of each of the cigarette material indexes and the cut tobacco formula indexes; the design parameters include the smoke design value and the cigarette design value;

[0016] Determine a to-be-model-selected index among the cigarette material indexes; use the cigarette material prediction model corresponding to the to-be-model-selected index as the target model; extract target parameters from the default index parameters, where the target parameters include the default parameter values of the non-to-be-model-selected indexes and the cut tobacco formula indexes; input the design parameters and the target parameters into the target model for model selection prediction to obtain a target prediction result;

[0017] Based on the target prediction result, determine the model selection result of the to-be-model-selected index;

[0018] Based on the model selection result of the to-be-model-selected index, update the default index parameters, replace the to-be-model-selected index, and repeat the above model selection process based on the updated default index parameters until the model selection results of all the cigarette material indexes are obtained.

[0019] In some embodiments, based on the target prediction result, determine the model selection result of the to-be-model-selected index, including:

[0020] Calculate the similarity between the target prediction result and each existing material in the preset material list;

[0021] Based on the similarity, determine the model selection result of the to-be-model-selected index among the existing materials.

[0022] In some embodiments, based on the model selection results of each of the cigarette material indexes, obtain the model selection result of the cigarette product, including:

[0023] Obtain a preset smoke model;

[0024] Input the default parameter values of the tobacco blend formula, the design parameters, and the selection results of each cigarette material index into a flue gas model to predict the flue gas prediction data of the cigarette product;

[0025] Analyze the flue gas prediction data, and based on the analysis results, determine the selection result of the cigarette product or re-select the cigarette materials and then determine the selection result of the cigarette product.

[0026] In some embodiments, before obtaining the preset flue gas model, it further includes:

[0027] Use the cigarette material index and the tobacco blend formula index as features, and the flue gas index as a label to train a flue gas model.

[0028] In some embodiments, after obtaining the preset flue gas model, it further includes:

[0029] Obtain the importance ranking of each feature in the flue gas model, and based on the importance ranking, determine the selection order of each cigarette material index.

[0030] In some embodiments, the cigarette material prediction model and the flue gas model are trained based on the random forest regression algorithm.

[0031] In some embodiments, the cigarette material index includes cigarette paper, tipping paper, plug wrap paper, and filter rod;

[0032] The flue gas index includes tar content, nicotine content in the flue gas, and carbon monoxide content;

[0033] The design parameters include flue gas design values and cigarette design values.

[0034] In a second aspect, in this embodiment, a cigarette material selection device for a cigarette product is provided, and the device includes:

[0035] A data preparation module for obtaining a historical data set of a cigarette product, where the historical data set includes relevant data of cigarette material indexes, flue gas indexes, and tobacco blend formula indexes;

[0036] A material model training module for training corresponding cigarette material prediction models for different types of cigarette material indexes based on the historical data set;

[0037] The material selection module is used to sequentially input the obtained design parameters and default index parameters into each of the cigarette material prediction models for selection prediction, and determine the selection results of each cigarette material index based on the prediction results. Among them, the selection results of the prior cigarette material indexes are used to update the default index parameters for inputting into the cigarette material prediction models called subsequently for selection prediction.

[0038] The selection confirmation module is used to obtain the selection result of the cigarette product based on the selection results of each cigarette material index.

[0039] In a third aspect, the present application also provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the cigarette material selection method for the cigarette product described in the first aspect.

[0040] In a fourth aspect, the present application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the cigarette material selection method for the cigarette product described in the first aspect.

[0041] Compared with the related art, in the cigarette material selection method, device, and computer device provided in this embodiment, by obtaining the historical data set of the cigarette product, the historical data set includes relevant data of cigarette material indexes, smoke indexes, and cut tobacco formula indexes; based on the historical data set, corresponding cigarette material prediction models are respectively trained for different types of cigarette material indexes; the obtained design parameters and default index parameters are sequentially input into each of the cigarette material prediction models for selection prediction, and the selection results of each cigarette material index are determined based on the prediction results. Among them, the selection results of the prior cigarette material indexes are used to update the default index parameters for inputting into the cigarette material prediction models called subsequently for selection prediction; based on the selection results of each cigarette material index, the selection result of the cigarette product is obtained. This embodiment realizes the intelligence and automation of cigarette material selection based on model training and prediction, solves the problems of low selection efficiency and accuracy, reduces the number of experiments and time costs, and also reduces the R & D and production costs.

[0042] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more concise and understandable. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0044] Figure 1 It is a schematic flowchart of the method for selecting cigarette rod materials of a cigarette product in an embodiment of the present application;

[0045] Figure 2 It is a schematic flowchart of the selection of cigarette paper in an embodiment of the present application;

[0046] Figure 3 It is a schematic flowchart of the training and prediction of a flue gas model in an embodiment of the present application;

[0047] Figure 4 It is a schematic flowchart of the method for selecting cigarette rod materials of a cigarette product in a preferred embodiment of the present application;

[0048] Figure 5 It is a structural block diagram of a device for selecting cigarette rod materials of a cigarette product in an embodiment of the present application.

[0049] Reference numerals: 10, data preparation module; 20, material model training module; 30, material selection module; 40, selection confirmation module. Detailed implementation manners

[0050] To understand the purpose, technical solution and advantages of the present application more clearly, the present application will be described and illustrated below with reference to the accompanying drawings and embodiments.

[0051] Unless otherwise defined, technical terms or scientific terms involved in this application shall have the general meanings understood by those with ordinary skills in the technical field to which this application belongs. In this application, words such as "a", "an", "one kind", "the", "these", etc. do not indicate a limitation in quantity, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product or device including a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent in these processes, methods, products or devices. The terms "connect", "be connected", "couple" and other similar words involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly connected or indirectly connected. The "plurality" involved in this application means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, and B exists alone. Usually, the character " / " indicates that the objects associated before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0052] In this embodiment, a method for selecting cigarette rod materials for a cigarette product is provided. Figure 1 It is a flowchart of the method for selecting cigarette rod materials for the cigarette product in this embodiment, as Figure 1 shown, and this process includes the following steps:

[0053] Step S210, obtain the historical data set of the cigarette product, and the historical data set includes relevant data of cigarette rod material indicators, smoke indicators, and cut tobacco formula indicators.

[0054] Specifically, collect the detection data related to the cigarette product, perform preprocessing operations such as cleaning, denoising, and normalization on the collected data to ensure the data quality, and classify the preprocessed data according to various indicators to form a historical data set. Among them, the cigarette rod material indicators include but are not limited to cigarette paper, tipping paper, forming paper, and filter rod, and the smoke indicators include but are not limited to tar content, nicotine content in smoke, and carbon monoxide content.

[0055] Step S220, based on the historical data set, train corresponding cigarette rod material prediction models for different types of cigarette rod material indicators respectively.

[0056] Specifically, for cigarette paper indicators, a cigarette paper prediction model is trained; for tipping paper indicators, a tipping paper prediction model is trained, and the same applies to other cigarette material indicators. Therefore, the cigarette material prediction model includes a cigarette paper prediction model, a tipping paper prediction model, etc.

[0057] Step S230: Input the obtained design parameters and default indicator parameters into each cigarette material prediction model in sequence for selection prediction, and determine the selection results of each cigarette material indicator based on the prediction results. Among them, the selection results of the previous cigarette material indicators are used to update the default indicator parameters for input into the cigarette material prediction model called subsequently for selection prediction.

[0058] Specifically, the design parameters include but are not limited to flue gas design values and cigarette design values. Among them, the flue gas design values include tar, nicotine in the flue gas, carbon monoxide, etc., and the cigarette design values include length, circumference, etc. The default indicator parameters are the default parameter values of one or more cigarette product indicators, and the cigarette product indicators include but are not limited to cigarette material indicators, flue gas indicators, and tobacco blend indicators. The initial default indicator parameters can be the empirical values provided by material developers. When using each cigarette material prediction model, it is carried out in the preset option order, and the selection result of the previous cigarette material indicator is used as the basis for the selection of the subsequent cigarette material indicator. For example, after determining the selection result of the tipping paper prediction model, the default parameter value of the tipping paper indicator is updated based on this selection result, and the updated default indicator parameters are used as the input data of the forming paper prediction model to predict the forming paper indicator.

[0059] Step S240: Obtain the selection result of the cigarette product based on the selection results of each cigarette material indicator.

[0060] Specifically, based on the selection results of each cigarette material indicator, the existing selection parameters are screened from the preset material list to determine the selection result of the cigarette product, or the selection results of each cigarette material indicator are directly used as the selection result of the cigarette product. The specific implementation method can be determined according to the usage scenario and accuracy requirements, and is not limited in this embodiment.

[0061] In this embodiment, by obtaining the historical data set and design parameters of the cigarette product, various cigarette material prediction models are trained based on the historical data set; based on the design parameters and default index parameters, the cigarette material prediction models are used for selection prediction in sequence, and the selection results of each cigarette material index are determined based on the prediction results; among them, the selection results of the previous cigarette material indexes are used to update the default index parameters so as to be input into the cigarette material prediction models called subsequently for selection prediction; finally, based on the selection results of each cigarette material index, the selection result of the cigarette product is obtained, realizing the intelligence and automation of cigarette material selection, solving the problems of low selection efficiency and accuracy, reducing the number of experiments and time costs, and also reducing the R & D and production costs. The automated process of this embodiment reduces human intervention, ensures the objectivity and consistency of the selection results, and improves work efficiency. And this method provides a scientific basis for cigarette design, helps manufacturers better understand material characteristics, and improves product quality.

[0062] In some of these embodiments, step S220, based on the historical data set, respectively trains corresponding cigarette material prediction models for different types of cigarette material indexes, including:

[0063] Step S211, determine a to-be-trained index among the cigarette material indexes, use the to-be-trained index as a label, and use the cigarette material indexes, smoke indexes, and cut tobacco formula indexes other than the to-be-trained index as features to train a cigarette material prediction model corresponding to the current to-be-trained index.

[0064] Step S212, replace the to-be-trained index, and repeat the above training process until all categories of cigarette material prediction models are trained.

[0065] Exemplarily, taking the cigarette paper index among the cigarette material indexes as a label and other indexes except the cigarette paper index as features, a cigarette paper prediction model is trained using the random forest regression algorithm. Other cigarette material prediction models are trained in the same way. In this embodiment, in addition to the random forest regression algorithm, machine learning algorithms such as gradient boosting decision trees or support vector machines can also be used, and there is no limitation on this.

[0066] In this embodiment, by taking a certain cigarette material index as a label and other indexes as features, and using a machine learning algorithm for training and model establishment, the prediction efficiency and accuracy of the specific parameters of the cigarette material are improved.

[0067] In some of these embodiments, step S230, input the obtained design parameters and default index parameters into each cigarette material prediction model for selection prediction in sequence, and determine the selection results of each cigarette material index based on the prediction results, including:

[0068] Step S231: Obtain design parameters and default index parameters. The default index parameters include the default parameter values of each cigarette material index and the cigarette tobacco blend index. The design parameters include the flue gas design values and the cigarette design values. Among them, the flue gas design values include the design values of indexes such as tar content, nicotine content in the flue gas, and carbon monoxide content. The cigarette design values include the design values such as cigarette length and cigarette circumference.

[0069] Step S232: Determine a to-be-selected index among the cigarette material indexes. Use the cigarette material prediction model corresponding to the to-be-selected index as the target model. Extract the target parameters from the default index parameters. The target parameters include the default parameter values of the non-to-be-selected indexes and the cigarette tobacco blend index. Input the design parameters and the target parameters into the target model for selection prediction to obtain the target prediction result.

[0070] Specifically, the non-to-be-selected indexes are the remaining material indexes of the cigarette material indexes except the to-be-selected index. Refer to Figure 2 , taking the index of cigarette paper as an example. When the cigarette paper is the to-be-selected index and the cigarette paper prediction model is the target model, extract the data of other indexes from the default index parameters, such as the default index parameters of tipping paper, forming paper, filter rod, and cigarette tobacco blend, as the target parameters. Input the default index parameters of tipping paper, forming paper, filter rod, and cigarette tobacco blend, as well as the flue gas design values and the cigarette design values, into the cigarette paper prediction model to obtain the target prediction result of the cigarette paper. Among them, the default index parameters are preset empirical values provided by material developers.

[0071] Step S233: Determine the selection result of the to-be-selected index based on the target prediction result.

[0072] Specifically, directly use the target prediction result as the selection result of the to-be-selected index, or determine the selection result of the to-be-selected index through the methods of the following Step S310 to Step S320.

[0073] Step S234: Update the default index parameters based on the selection result of the to-be-selected index, and replace the to-be-selected index. Repeat the above selection process based on the updated default index parameters until the selection results of all cigarette material indexes are obtained.

[0074] Steps S232 to S233 are the process of selecting a model for one of the cigarette material indicators. When there are multiple cigarette material indicators, after step S233, in the order of selection, the next cigarette material indicator is taken as the indicator to be selected, and the corresponding cigarette material prediction model is obtained. Steps S232 to S234 are repeatedly executed until all material selections are completed. Among them, after the selection of a certain material is completed, the selection result of the current cigarette material indicator is used to replace the default parameter of the current cigarette material indicator previously, and a new default parameter is obtained. When predicting with other subsequent cigarette material prediction models, the selection result of the current cigarette material indicator (i.e., the new default parameter of the current cigarette material indicator) will be used as a feature to input other cigarette material prediction models, so as to use the selection result of the previous cigarette material indicator as the input parameter of the subsequent cigarette material indicator.

[0075] In some of these embodiments, in step S233 above, based on the target prediction result, determining the selection result of the indicator to be selected includes:

[0076] Step S310, calculating the similarity between the target prediction result and each existing material in the preset material list.

[0077] Specifically, a material list is constructed to record the data of various available cigarette material indicators, and the parameter range of the cigarette materials is limited to facilitate procurement or customization. For example, the list of cigarette paper types is shown in Table 1.

[0078] Table 1:

[0079]

[0080] See Figure 2 , comparing the target prediction result of the cigarette paper with the types of each existing material in the preset material list, and finding the closest type in the material list as the target. Among them, the closest method can be calculated using cosine similarity, and the formula is as follows:

[0081]

[0082] Among them, the range of the cosine similarity value is [-1, 1]. Positive 1 indicates complete similarity and positive correlation, negative 1 indicates opposite direction and negative correlation, and 0 indicates perpendicular and orthogonal without correlation. A represents the vector of the current target prediction result, and B represents the vector of the existing material type. ║A║ represents the L2 norm of vector A, that is, the vector length; A·B represents the dot product of vector A and vector B.

[0083] Step S320, determining the selection result of the indicator to be selected based on the similarity among the existing materials.

[0084] Specifically, the existing materials whose similarity meets the material selection requirements are used as the selection results of the cigarette material indicators. Whether it meets the material selection requirements can be determined through threshold judgment.

[0085] In this embodiment, the predicted indicators are numerical, so the maximum cosine similarity is used to calculate the closest degree to each category, rather than directly performing classification prediction. Using the direct classification method will result in the inability to select such results when the predicted results are not in the existing category list. However, by applying the method of maximum cosine similarity, the existing category closest to the predicted value can be found, thus avoiding the problem of inability to select due to new classifications not being in the list. This method not only improves the accuracy of prediction but also enhances the adaptability and flexibility to unforeseen categories.

[0086] In some of these embodiments, in step S240 above, based on the selection results of each cigarette material indicator, the selection result of the cigarette product is obtained. Refer to Figure 3 , including:

[0087] Step S241, obtain a preset flue gas model. The flue gas model is used to predict the specific situation of the flue gas indicators of the cigarette product.

[0088] Step S242, input the default parameter values, design parameters of the tobacco blend, and the selection results of each cigarette material indicator into the flue gas model, and predict the flue gas prediction data of the cigarette product.

[0089] Step S243, analyze the flue gas prediction data, and determine the selection result of the cigarette product based on the analysis result or determine the selection result of the cigarette product after re-selecting the cigarette material.

[0090] Specifically, compare the flue gas prediction data output by the flue gas model with the flue gas design value, and calculate the evaluation index of the selection total model composed of the current prediction models of each cigarette material. If the evaluation index is low, the prediction model of the cigarette material can be adjusted and then re-selected.

[0091] In this embodiment, the quality of the selection is ensured by performing flue gas evaluation on the selection result of the cigarette product.

[0092] In some of these embodiments, refer to Figure 3 , before obtaining the flue gas model, it further includes: using the cigarette material indicators and the tobacco blend indicators as features and the flue gas indicators as labels to train the flue gas model.

[0093] Specifically, train with the random forest regression algorithm, and use the flue gas indicators in the historical dataset as labels and other indicators as features to specifically train the flue gas model to improve the flue gas prediction accuracy of the cigarette product.

[0094] In some of these embodiments, after obtaining the preset flue gas model, it further includes: obtaining the importance ranking of each feature in the flue gas model, and determining the selection order of each cigarette material index based on the importance ranking.

[0095] Specifically, during the training process of the flue gas model, an importance score for each feature is generated, and the selection is made in descending order of importance.

[0096] For example, the top ten indicators in terms of importance ranking as shown in Table 2 are obtained, so as to determine the corresponding selection material order as tipping paper, plug wrap paper, cigarette paper, and filter rod.

[0097] Table 2:

[0098] Serial number Value Index Belonging material 1 0.4963 Perforated tipping paper air permeability Filter rod tipping paper 2 0.1062 Forming paper air permeability Forming paper 3 0.0982 Number of punching rows Filter rod tipping paper 4 0.0702 Width of hole belt Filter rod tipping paper 5 0.0274 Potassium ion content Cigarette paper 6 0.0268 Cut stem weight Cut tobacco 7 0.0249 Net consumption of cut tobacco per cigarette Cut tobacco 8 0.0195 Expanded cut tobacco weight Cut tobacco 9 0.0184 Open draw resistance Filter rod tipping paper 10 0.0156 Main leaf group Cut tobacco

[0099] In this embodiment, since there are interactions between multiple cigarette material selections and they jointly affect the flue gas indicators of cigarette products, the priority of each cigarette material is determined through the flue gas model, improving the selection accuracy.

[0100] The following describes and illustrates this embodiment through preferred embodiments. Refer to Figure 4 , this preferred embodiment provides a method for selecting cigarette materials for a cigarette product, and the method includes the following steps:

[0101] S1. Data collection: Collect design data and detection data related to cigarette products.

[0102] S2. Data processing: Clean the collected data; the cigarette materials mainly include three papers and one rod, so the data set is divided into cigarette material indicators (cigarette paper, tipping paper, plug wrap paper, filter rod), flue gas indicators (tar content, nicotine content in flue gas, and carbon monoxide content), and cut tobacco formula indicators.

[0103] S3. Model training:

[0104] S3.1 Establish a flue gas prediction model: Using the above data set, taking the cigarette material indicators and other indicators as features and the flue gas indicators as labels, and adopting the random forest regression algorithm, after training, establish a flue gas prediction model, hereinafter referred to as the flue gas model.

[0105] S3.2 Determine the selection order of cigarette materials: Perform importance ranking of features in the flue gas model to determine the selection order of each indicator of cigarette materials.

[0106] S3.3 Establish a cigarette material prediction model:

[0107] Taking the indicators other than the cigarette paper in the cigarette material indicators as labels and the other indicators of the data set as features, and using the random forest regression algorithm, after training, establish a cigarette paper prediction model, hereinafter referred to as the cigarette paper model.

[0108] Taking the external index of the tipping paper among the cigarette material indexes as the label, and the other indexes of the data set as the features, using the random forest regression algorithm, after training, a tipping paper prediction model is established, hereinafter referred to as the tipping paper model.

[0109] Taking the external index of the plug wrap among the cigarette material indexes as the label, and the other indexes of the data set as the features, using the random forest regression algorithm, after training, a plug wrap prediction model is established, hereinafter referred to as the plug wrap model.

[0110] Taking the external index of the filter rod among the cigarette material indexes as the label, and the other indexes of the data set as the features, using the random forest regression algorithm, after training, a filter rod prediction model is established, hereinafter referred to as the filter rod model.

[0111] One of the above indexes refers to a set of data in the data set.

[0112] S4. Conduct model selection:

[0113] S4.1 According to the design requirements, determine the flue gas design values (tar content, nicotine content in the flue gas, and carbon monoxide content), and the cigarette design values (length, circumference, etc.) as the design parameters.

[0114] S4.2 Prepare the existing type data of cigarette materials: List the lists of various materials currently available for selection (cigarette paper, tipping paper, plug wrap, and filter rod) for model selection.

[0115] S4.3 Conduct model selection (taking the tipping paper model selection as an example):

[0116] S4.3.1 Import the tipping paper model. Taking the design parameters, the default parameters of the cigarette material indexes (excluding the tipping paper index parameters), and the flue gas indexes as the input items, input them into the tipping paper model, and output the tipping paper prediction data. Among them, the so-called default parameters are the empirical values provided by the index developers.

[0117] S4.3.2 Compare the tipping paper prediction data with the existing tipping paper types, search for the most similar type, and select this type as the target. Cosine similarity is used when searching for the most similar type.

[0118] S4.3.3 Replace the specific indexes of the existing list of tipping paper with the original default tipping paper parameters, representing the input items, and conduct the model selection of the next material.

[0119] S4.3.4 Conduct the model selection of other materials, that is, repeat the process from S6.3.1 to S6.3.3 until the model selection of all materials is completed.

[0120] S5. Evaluation and Verification: Based on the above cigarette material selection results, using them as input items, input into the smoke model, output the smoke prediction data, compare the difference between the smoke prediction data and the smoke design value, and finally give the evaluation index of this selection total model. If the evaluation index is low, it is necessary to adjust the selection parameters, select again, and obtain the selection results of the cigarette products. The input items in this step include cigarette material parameters (i.e., the results after selection), design parameters, and tobacco blend parameters.

[0121] Through steps such as data collection, data processing, model training, selection, and evaluation and verification, this preferred embodiment realizes the intelligence and automation of material selection. Compared with the traditional method that relies on experience and intuition, the present invention significantly improves the selection efficiency and accuracy, reduces the number of experiments and time costs, and lowers the R & D and production costs. The automated process reduces human intervention, ensures the objectivity and consistency of the selection results, and improves work efficiency. This method provides a scientific basis for cigarette design, helps manufacturers better understand material characteristics, and thus make choices that better meet the design requirements, improving product quality and market competitiveness.

[0122] It should be noted that the steps shown in the above process or the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0123] In this embodiment, a cigarette material selection device for cigarette products is also provided. This device is used to implement the above embodiments and preferred embodiments, and those that have been described will not be repeated. The following terms such as "module", "unit", "sub-unit", etc. can be a combination of software and / or hardware that can achieve a predetermined function. Although the device described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0124] Figure 5 is the structural block diagram of the cigarette material selection device for cigarette products in this embodiment, as Figure 5 shown, this device includes: a data preparation module 10, a material model training module 20, a material selection module 30, and a selection confirmation module 40.

[0125] The data preparation module 10 is used to obtain the historical data set of cigarette products, and the historical data set includes relevant data on cigarette material indicators, smoke indicators, and tobacco blend indicators.

[0126] The material model training module 20 is used to respectively train corresponding cigarette material prediction models for different types of cigarette material indicators based on the historical data set.

[0127] The material selection module 30 is used to sequentially input the obtained design parameters and default index parameters into each cigarette material prediction model for selection prediction, and determine the selection results of each cigarette material index based on the prediction results. Among them, the selection results of the prior cigarette material indexes are used to update the default index parameters for input into the cigarette material prediction model called subsequently for selection prediction.

[0128] The selection confirmation module 40 is used to obtain the selection result of the cigarette product based on the selection results of each cigarette material index.

[0129] In some of the embodiments, the material model training module 20 is further used to determine a to-be-trained index among the cigarette material indexes, use the to-be-trained index as a label, use the cigarette material indexes, smoke indexes, and cut tobacco formula indexes other than the to-be-trained index as features, and train to obtain a cigarette material prediction model corresponding to the current to-be-trained index. Replace the to-be-trained index, and repeat the above training process until all categories of cigarette material prediction models are trained.

[0130] In some of the embodiments, the material selection module 30 is further used to obtain design parameters and default index parameters. The default index parameters include the default parameter values of each cigarette material index and cut tobacco formula index. The design parameters include the smoke design value and cigarette design value. Determine a to-be-selected index among the cigarette material indexes. Use the cigarette material prediction model corresponding to the to-be-selected index as the target model. Extract target parameters from the default index parameters. The target parameters include the default parameter values of the non-to-be-selected indexes and cut tobacco formula index. Input the design parameters and target parameters into the target model for selection prediction to obtain a target prediction result. Based on the target prediction result, determine the selection result of the to-be-selected index. Based on the selection result of the to-be-selected index, update the default index parameters, replace the to-be-selected index, and repeat the above selection process based on the updated default index parameters until the selection results of all cigarette material indexes are obtained.

[0131] In some of the embodiments, the material selection module 30 is further used to calculate the similarity between the target prediction result and each existing material in the preset material list. Determine the selection result of the to-be-selected index based on the similarity among the existing materials.

[0132] In some of the embodiments, the selection confirmation module 40 is further used to obtain a preset smoke model. Input the default parameter value of the cut tobacco formula, design parameters, and the selection results of each cigarette material index into the smoke model to predict and obtain the smoke prediction data of the cigarette product. Analyze the smoke prediction data, and determine the selection result of the cigarette product based on the analysis result or re-perform the cigarette material selection to determine the selection result of the cigarette product.

[0133] In some of these embodiments, it further includes: a flue gas model training module 50, which is used to use cigarette material indexes and cut tobacco formula indexes as features and flue gas indexes as labels to train a flue gas model.

[0134] In some of these embodiments, a material selection module 30 is used to, after obtaining a preset flue gas model, obtain the importance ranking of each feature in the flue gas model, and determine the selection order of each cigarette material index based on the importance ranking.

[0135] In some of these embodiments, the cigarette material prediction model and the flue gas model are trained based on the random forest regression algorithm.

[0136] In some of these embodiments, the cigarette material indexes include cigarette paper, tipping paper, forming paper, and filter rod; the flue gas indexes include tar content, nicotine content in the flue gas, and carbon monoxide content; the design parameters include flue gas design values and cigarette design values.

[0137] It should be noted that the above-mentioned various modules can be functional modules or program modules, and can be implemented either by software or by hardware. For the modules implemented by hardware, the above-mentioned various modules can be located in the same processor; or the above-mentioned various modules can also be located in different processors in any combined form.

[0138] In this embodiment, a computer device is further provided, including a memory and a processor. A computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0139] Optionally, the above-mentioned computer device may further include a transmission device and input / output devices. Among them, the transmission device is connected to the above-mentioned processor, and the input / output devices are connected to the above-mentioned processor.

[0140] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementation manners, and will not be repeated in this embodiment.

[0141] In addition, in combination with the cigarette material selection method of the cigarette product provided in the above embodiments, a storage medium can also be provided to implement it in this embodiment. A computer program is stored on the storage medium; when the computer program is executed by a processor, it implements any one of the cigarette material selection methods of the cigarette product in the above embodiments.

[0142] It should be understood that the specific embodiments described herein are for the purpose of explaining this application and not for limiting it. All other embodiments obtained by those of ordinary skill in the art based on the embodiments provided in this application without creative efforts shall fall within the protection scope of this application.

[0143] Obviously, the accompanying drawings are only some examples or embodiments of this application. For those of ordinary skill in the art, this application can also be applied to other similar situations based on these drawings without creative efforts. Additionally, it can be understood that although the work done during the development process here may be complex and time-consuming, for those of ordinary skill in the art, certain design, manufacturing, or production changes based on the technical content disclosed in this application are only routine technical means and should not be regarded as insufficient disclosure of this application.

[0144] The term "embodiment" in this application means that the specific features, structures, or characteristics described in connection with the embodiment may be included in at least one embodiment of this application. The phrase appears in various positions in the specification and does not necessarily mean the same embodiment, nor does it mean being independent or alternative to other embodiments and mutually exclusive. Those of ordinary skill in the art can clearly or implicitly understand that the embodiments described in this application can be combined with other embodiments without conflict.

[0145] The above-described embodiments only represent several implementation manners of this application. Their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of patent protection. It should be noted that for those of ordinary skill in the art, without departing from the concept of this application, several modifications and improvements can still be made, and these all belong to the protection scope of this application. Therefore, the protection scope of this application shall be subject to the appended claims.

Claims

1. A method for selecting cigarette rod materials for a cigarette product, characterized in that, The method includes: Obtaining a historical data set of cigarette products, where the historical data set includes relevant data on cigarette material indicators, smoke indicators, and tobacco blend indicators; Based on the historical data set, for different types of the cigarette material indicators, respectively training corresponding cigarette material prediction models; Sequentially inputting the obtained design parameters and default index parameters into each of the cigarette material prediction models for selection prediction, and determining the selection results of each of the cigarette material indicators based on the prediction results; wherein, the selection result of the prior cigarette material indicator is used to update the default index parameters for inputting into the cigarette material prediction model called subsequently for selection prediction; Based on the selection results of each of the cigarette material indicators, obtaining the selection result of the cigarette product.

2. The method for selecting cigarette rod materials of the cigarette product according to claim 1, wherein Based on the historical data set, for different types of the cigarette material indicators, respectively training corresponding cigarette material prediction models, including: Determining a to-be-trained indicator among the cigarette material indicators, using the to-be-trained indicator as a label, and using the cigarette material indicators other than the to-be-trained indicator, the smoke indicators, and the tobacco blend indicators as features, and training to obtain a cigarette material prediction model corresponding to the current to-be-trained indicator; Replacing the to-be-trained indicator, and repeating the above training process until all categories of the cigarette material prediction models are trained.

3. The method for selecting cigarette rod materials of the cigarette product according to claim 1, characterized in that Sequentially inputting the obtained design parameters and default index parameters into each of the cigarette material prediction models for selection prediction, and determining the selection results of each of the cigarette material indicators based on the prediction results, including: Obtaining design parameters and default index parameters; the default index parameters include the default parameter values of each of the cigarette material indicators and the tobacco blend indicators; the design parameters include a smoke design value and a cigarette design value; Determining a to-be-selected indicator among the cigarette material indicators; using the cigarette material prediction model corresponding to the to-be-selected indicator as a target model; extracting target parameters from the default index parameters, where the target parameters include the default parameter values of the non-to-be-selected indicators and the tobacco blend indicators; inputting the design parameters and the target parameters into the target model for selection prediction to obtain a target prediction result; Based on the target prediction result, determining the selection result of the to-be-selected indicator; Based on the selection result of the to-be-selected indicator, updating the default index parameters, and replacing the to-be-selected indicator, and repeating the above selection process based on the updated default index parameters until the selection results of all the cigarette material indicators are obtained.

4. The method for selecting cigarette rod materials of the cigarette product according to claim 3, characterized in that, Based on the target prediction result, determining the selection result of the to-be-selected indicator, including: Calculating the similarity between the target prediction result and each existing material in a preset material list; Based on the similarity, determining the selection result of the to-be-selected indicator among the existing materials.

5. The method for selecting cigarette rod materials of the cigarette product according to claim 1, characterized in that Based on the selection results of each of the cigarette material indicators, obtaining the selection result of the cigarette product, including: Obtaining a preset smoke model; Input the default parameter values of the tobacco blend formula, the design parameters, and the selection results of each cigarette material index into a flue gas model to predict the flue gas prediction data of the cigarette product; Analyze the flue gas prediction data and determine the selection result of the cigarette product based on the analysis result or determine the selection result of the cigarette product after re-selecting the cigarette materials.

6. The method for selecting cigarette rod materials of the cigarette product according to claim 5, wherein Before obtaining the preset flue gas model, it further includes: Using the cigarette material index and the tobacco blend formula index as features and the flue gas index as a label to train a flue gas model.

7. The method for selecting cigarette rod materials of the cigarette product according to claim 5, characterized in that, After obtaining the preset flue gas model, it further includes: Obtain the importance ranking of each feature in the flue gas model and determine the selection order of each cigarette material index based on the importance ranking.

8. The method for selecting cigarette rod materials of the cigarette product according to claim 5, characterized in that, The cigarette material prediction model and the flue gas model are trained based on the random forest regression algorithm.

9. The method for selecting cigarette rod materials of the cigarette product according to claim 1, characterized in that, The cigarette material index includes cigarette paper, tipping paper, plug wrap paper, and filter rod; The flue gas index includes tar content, nicotine content in the flue gas, and carbon monoxide content.

10. A cigarette rod material selection device for a cigarette product, characterized in that, The device includes: A data preparation module for obtaining a historical data set of cigarette products, where the historical data set includes relevant data of cigarette material indexes, flue gas indexes, and tobacco blend formula indexes; A material model training module for training corresponding cigarette material prediction models for different types of cigarette material indexes respectively based on the historical data set; A material selection module for inputting the obtained design parameters and default index parameters into each cigarette material prediction model in sequence for selection prediction, and determining the selection result of each cigarette material index based on the prediction result; among them, the selection result of the previous cigarette material index is used to update the default index parameters for inputting into the subsequent called cigarette material prediction model for selection prediction; A selection confirmation module for obtaining the selection result of the cigarette product based on the selection results of each cigarette material index.

Images