Tobacco product raw material producing area prediction method and device and prediction model training method

Abstract

The invention relates to a method and device for predicting producing areas of tobacco product raw materials and a training method of a prediction model, and relates to the technical field of tobacco production. The method comprises the steps that thermogravimetric analysis data and chemical component data of tobacco product raw materials are obtained, and the thermogravimetric analysis data comprise a plurality of pyrolysis temperatures in the pyrolysis process of the tobacco product raw materials and the mass change rate, corresponding to each pyrolysis temperature, of the tobacco product raw materials; the chemical component data comprises the content of each chemical component in various chemical components in the tobacco product raw materials; determining characteristic data comprising a characteristic temperature in the plurality of pyrolysis temperatures and a mass change rate corresponding to the characteristic temperature from the thermogravimetric analysis data; based on at least one parameter of the characteristic temperature and the quality change rate corresponding to the characteristic temperature and the chemical component data, utilizing a trained prediction model to predict the production place of the tobacco product raw material. According to the technical scheme, the accuracy of tobacco product raw material producing area prediction can be improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of tobacco production, and particularly relates to a method, a device, and a training method for a prediction model for predicting the origin of raw materials for tobacco products. Background Art

[0002] In the manufacturing process of tobacco products, the quality and characteristics of raw materials for tobacco products (such as tobacco leaves) are key factors determining the sensory quality and combustion performance of tobacco products. Due to the diversity of natural conditions such as ecological climate and soil in tobacco-growing regions and the differences in the accumulation degree of planting technical experience in different regions, raw materials for tobacco products from different regions show obvious differences in quality and characteristics (such as flavor styles). For example, raw materials for tobacco products produced in Yunnan belong to the light flavor style, those produced in Guizhou belong to the middle flavor style, and those produced in Henan present the strong flavor style. In order to facilitate the subsequent industrial use of raw materials for tobacco products (such as the design of tobacco product formulas), it is necessary to accurately distinguish their origins. Summary of the Invention

[0003] To solve the above problems, embodiments of the present disclosure provide a method and a device for predicting the origin of raw materials for tobacco products based on thermogravimetric analysis data and chemical composition data of the raw materials for tobacco products, and propose a training method for a prediction model for performing such a prediction.

[0004] According to some embodiments of the present disclosure, a method for predicting the origin of raw materials for tobacco products is provided, including: obtaining thermogravimetric analysis data and chemical composition data of the raw materials for tobacco products, where the thermogravimetric analysis data includes multiple pyrolysis temperatures of the pyrolysis process of the raw materials for tobacco products and the mass change rate of the raw materials for tobacco products corresponding to each of the multiple pyrolysis temperatures, and the chemical composition data includes the content of each of multiple chemical components in the raw materials for tobacco products; determining characteristic data from the thermogravimetric analysis data, where the characteristic data includes characteristic temperatures among the multiple pyrolysis temperatures and the mass change rate corresponding to the characteristic temperatures; and predicting the origin of the raw materials for tobacco products by using a trained prediction model based on at least one parameter among the characteristic temperatures and the mass change rate corresponding to the characteristic temperatures and the chemical composition data.

[0005] In some embodiments, the multiple pyrolysis temperatures form a temperature range, and the mass change rate corresponding to the characteristic temperature includes one or more extreme values of the mass change rate of the raw materials for tobacco products within the temperature range.

[0006] In some embodiments, the plurality of pyrolysis temperatures form adjacent first and second temperature ranges, the mass change rate corresponding to the characteristic temperature includes a median value between a first extreme value of the mass change rate of the tobacco product raw material within the first temperature range and a second extreme value of the mass change rate of the tobacco product raw material within the second temperature range, the characteristic temperature includes a first temperature corresponding to the median value in the thermogravimetric analysis data, the temperature corresponding to the first extreme value is less than the temperature corresponding to the second extreme value, and the first temperature is greater than the temperature corresponding to the first extreme value and less than the temperature corresponding to the second extreme value.

[0007] In some embodiments, the characteristic temperature includes a second temperature corresponding to the median value in the thermogravimetric analysis data, and the second temperature is less than the temperature corresponding to the first extreme value or greater than the temperature corresponding to the second extreme value.

[0008] In some embodiments, the characteristic temperature includes a starting temperature and / or an ending temperature among the plurality of pyrolysis temperatures.

[0009] In some embodiments, the chemical composition data further includes a ratio of the contents of different chemical components among the plurality of chemical components in the tobacco product raw material.

[0010] In some embodiments, the trained prediction model is trained in the following manner: Obtain the thermogravimetric analysis data and chemical composition data of each tobacco product raw material sample from a plurality of tobacco product raw material samples from different origins. The thermogravimetric analysis data of each tobacco product raw material sample includes a plurality of sample pyrolysis temperatures of the pyrolysis process of each tobacco product raw material sample and the mass change rate corresponding to each sample pyrolysis temperature among the plurality of sample pyrolysis temperatures. The chemical composition data of each tobacco product raw material sample includes the content of each chemical component among the plurality of chemical components in each tobacco product raw material sample. Determine the characteristic data of each tobacco product raw material sample from the thermogravimetric analysis data of each tobacco product raw material sample. The characteristic data of each tobacco product raw material sample includes a sample characteristic temperature among the plurality of sample pyrolysis temperatures and the mass change rate corresponding to the sample characteristic temperature. Use at least one parameter among the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature and the chemical composition data of each tobacco product raw material sample as input and the predicted origin of each tobacco product raw material sample as output to train the prediction model until the training end condition is met.

[0011] In some embodiments, the trained prediction model includes a random forest model.

[0012] In some embodiments, the multiple chemical components include at least one of total nitrogen, nicotine, total sugar, reducing sugar, potassium, and chlorine.

[0013] According to some other embodiments of the present disclosure, there is provided a method for training a prediction model for predicting the origin of tobacco product raw materials. The training method includes: obtaining, for each tobacco product raw material sample among a plurality of tobacco product raw material samples from a plurality of different origins, thermogravimetric analysis data and chemical composition data. The thermogravimetric analysis data of each tobacco product raw material sample includes a plurality of sample pyrolysis temperatures of the pyrolysis process of each tobacco product raw material sample and the mass change rate corresponding to each sample pyrolysis temperature among the plurality of sample pyrolysis temperatures. The chemical composition data of each tobacco product raw material sample includes the content of each chemical component among the multiple chemical components in each tobacco product raw material sample; determining, from the thermogravimetric analysis data of each tobacco product raw material sample, the characteristic data of each tobacco product raw material sample. The characteristic data of each tobacco product raw material sample includes the sample characteristic temperature among the plurality of sample pyrolysis temperatures and the mass change rate corresponding to the sample characteristic temperature; using at least one parameter among the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature and the chemical composition data of each tobacco product raw material sample as input and the predicted origin of each tobacco product raw material sample as output to train the prediction model until the training end condition is satisfied.

[0014] According to some further embodiments of the present disclosure, there is provided a device for predicting the origin of tobacco product raw materials, including: an acquisition module configured to acquire thermogravimetric analysis data and chemical composition data of the tobacco product raw materials. The thermogravimetric analysis data includes a plurality of pyrolysis temperatures of the pyrolysis process of the tobacco product raw materials and the mass change rate of the tobacco product raw materials corresponding to each pyrolysis temperature among the plurality of pyrolysis temperatures. The chemical composition data includes the content of each chemical component among the multiple chemical components in the tobacco product raw materials; a determination module configured to determine characteristic data from the thermogravimetric analysis data. The characteristic data includes the characteristic temperature among the plurality of pyrolysis temperatures and the mass change rate corresponding to the characteristic temperature; a prediction module configured to predict the origin of the tobacco product raw materials by using at least one parameter among the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data, and by using the trained prediction model.

[0015] According to some other embodiments of the present disclosure, there is provided an electronic device, including: a memory; and a processor coupled to the memory. The processor is configured to execute the method for predicting the origin of tobacco product raw materials or the method for training the prediction model in any one of the above embodiments based on instructions stored in the memory device.

[0016] According to some further embodiments of the present disclosure, there is provided a computer-readable storage medium having stored thereon computer instructions which, when executed by a processor, implement the method for predicting the origin of tobacco product raw materials or the training method of the prediction model in any one of the above embodiments.

[0017] According to some further embodiments of the present disclosure, there is also provided a computer program product including instructions which, when executed by a processor, cause the processor to execute the method for predicting the origin of tobacco product raw materials or the training method of the prediction model in any one of the above embodiments.

[0018] In the above embodiments, the characteristic data determined from the thermogravimetric analysis data of the tobacco product raw materials includes two parameters that can characterize the pyrolysis characteristics of the tobacco product raw materials, namely, the characteristic temperature and the mass change rate corresponding to the characteristic temperature. The trained prediction model is used to perform a fusion process on at least one of these two parameters that can characterize the pyrolysis characteristics of the tobacco product raw materials and the chemical composition data of the tobacco product raw materials to predict the origin of the tobacco product raw materials. Since the origin of the tobacco product raw materials affects the pyrolysis characteristics and the content of various chemical components of the raw materials, that is, there is a correlation between the origin of the tobacco product raw materials and their pyrolysis characteristics and the content of chemical components. Moreover, considering the pyrolysis characteristics and the content of chemical components of the tobacco product raw materials simultaneously can better describe the raw material characteristics. Therefore, combining the pyrolysis characteristics of the tobacco product raw materials and the content of various chemical components in the tobacco product raw materials for predicting the origin can improve the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings forming a part of the specification depict embodiments of the present disclosure and, together with the description, are used to explain the principles of the present disclosure.

[0020] Referring to the drawings, the present disclosure can be more clearly understood from the following detailed description, wherein:

[0021] Figure 1 A flowchart showing some embodiments of the method for predicting the origin of tobacco product raw materials of the present disclosure;

[0022] Figure 2 A schematic diagram showing some embodiments of the DTG curve of the present disclosure;

[0023] Figure 3 A schematic diagram showing some other embodiments of the DTG curve of the present disclosure;

[0024] Figure 4 A flowchart showing some embodiments of the training method of the prediction model of the present disclosure;

[0025] Figure 5Flowcharts showing other embodiments of the training method of the prediction model of the present disclosure;

[0026] Figure 6 Block diagrams showing some embodiments of the device for predicting the origin of tobacco product raw materials of the present disclosure;

[0027] Figure 7 Block diagrams showing some embodiments of the training device of the prediction model of the present disclosure;

[0028] Figure 8 Block diagrams showing some embodiments of the electronic device of the present disclosure;

[0029] Figure 9 Block diagrams showing other embodiments of the electronic device of the present disclosure. Detailed Description of the Invention

[0030] Various exemplary embodiments of the present disclosure will now be described in detail with reference to the accompanying drawings. It should be noted that: Unless otherwise specifically stated, the relative arrangements of components and steps, numerical expressions, and numerical values set forth in these embodiments do not limit the scope of the present disclosure.

[0031] The following description of at least one exemplary embodiment is merely illustrative in nature and in no way limits the present disclosure, its application, or its use.

[0032] Techniques, methods, and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be regarded as part of the specification.

[0033] In all examples shown and discussed herein, any specific values should be construed as merely exemplary and not as limitations. Thus, other examples of the exemplary embodiments may have different values.

[0034] It should be noted that: Like reference numerals and letters denote like items in the following drawings, and thus, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0035] In the related art, the near-infrared spectrum information of tobacco product raw materials is usually used to predict the origin of tobacco product raw materials. However, in this way of predicting the origin only by using a single-dimensional index of the same type, the obtained prediction result of the origin is often not accurate enough.

[0036] The inventors of the present disclosure have found through research that the characteristics of the smoke formed by tobacco products are affected by the chemical composition content and pyrolysis characteristics of the tobacco product raw materials. Due to differences in factors such as soil composition, climate conditions, and planting techniques, the tobacco product raw materials from different origins will have differences in their chemical composition content and pyrolysis characteristics. For example, the tobacco leaves produced in Zimbabwe have a relatively high nicotine content, and the smoke generated during pyrolysis is relatively rich and has a complex aroma, while the tobacco leaves produced in Yunnan have a moderate nicotine content, and the smoke generated during pyrolysis is relatively mild and has a long-lasting aroma.

[0037] If only the chemical composition content of the tobacco product raw materials is analyzed, it can provide a certain basis for origin prediction. However, due to the large variety of chemical components in the tobacco product raw materials, it is difficult to accurately measure the content of some chemical components, and the tobacco product raw materials from different origins may also have similar chemical composition contents, resulting in inaccurate origin prediction results obtained solely based on the chemical composition content. Pyrolysis reaction is one of the main chemical reactions that occur when the tobacco product raw materials are continuously heated in the temperature range above the pyrolysis threshold temperature. The pyrolysis reaction can occur in the heat-not-burn stage or in the combustion stage. The temperature in the former stage is generally lower than that in the latter stage. The pyrolysis characteristics can reflect the behavioral characteristics of the tobacco product raw materials during the pyrolysis reaction. Although analyzing only the pyrolysis characteristics of the tobacco product raw materials can also provide a certain basis for origin prediction, however, the pyrolysis reaction is affected by various factors such as heating rate and heating temperature. These factors may cause the tobacco product raw materials from the same origin to exhibit different pyrolysis characteristics. In addition, the pyrolysis characteristics of the tobacco product raw materials are often characterized by collecting data such as the temperature and mass change rate of the raw materials during the reaction process. The selection of characteristic points during the collection process will affect the accuracy of the characterization. These make the origin prediction based solely on the pyrolysis characteristics equally inaccurate.

[0038] Therefore, neither the pyrolysis characteristics nor the chemical composition content alone is sufficient to accurately predict the origin of the tobacco product raw materials. To improve the accuracy of the origin prediction of the tobacco product raw materials, in the differentiation of the origin of the tobacco product raw materials, multiple dimensions of indicators including chemical composition content and pyrolysis characteristics can be used in combination. By combining data from multiple different dimensions, the complementary advantages between them can be utilized to reduce the deviation caused by a single data source, so as to more comprehensively capture the origin characteristics of the tobacco and make the prediction results more reliable.

[0039] In view of this, an embodiment of the present disclosure provides a technical solution for predicting the origin of tobacco product raw materials, which can fuse two types of indicators, namely, thermogravimetric analysis data and chemical composition data of tobacco product raw materials, by using a trained prediction model to achieve the origin prediction of tobacco product raw materials. Compared with the method of predicting the origin of tobacco product raw materials by only using a single dimension of indicators of the same type, the accuracy of the origin prediction of tobacco product raw materials is effectively improved.

[0040] Figure 1 The flowchart showing some embodiments of the method for predicting the origin of tobacco product raw materials according to the present disclosure is as follows. Figure 1 As shown, for example, the method for predicting the origin of tobacco product raw materials may include step 110 to step 130.

[0041] In step 110, obtain the thermogravimetric analysis data and chemical composition data of the tobacco product raw materials.

[0042] In the present disclosure, a tobacco product is a product made entirely or partially from tobacco leaves as raw materials and is used for smoking, chewing, nasal inhalation, or other uses. Tobacco products may include, but are not limited to, cigarettes, cigars, pipe tobacco, or hookah. From the perspective of tobacco leaf formulation and processing technology, cigarettes may include, but are not limited to, flue-cured tobacco, burley tobacco, oriental tobacco, or sun-cured tobacco. Tobacco products release chemical substances such as nicotine (commonly known as "nicotine") through heating (e.g., heated tobacco products) and / or combustion (e.g., cigarettes, cigars, etc.). Tobacco products may or may not have a cigarette paper wrapper and may or may not have a filter tip. In the present disclosure, for the convenience of description, in some cases, cigarettes are used as examples of tobacco products for illustration. However, it should be recognized that various features or limitations described herein regarding cigarettes also apply to other types of tobacco products.

[0043] Tobacco product raw materials may include various forms of tobacco leaves, for example, they may be tobacco leaves or cut tobacco, tobacco flakes, tobacco powder, or tobacco blocks formed by processing tobacco leaves.

[0044] The chemical composition data of the tobacco product raw materials may include the content of each chemical component in various chemical components of the tobacco product raw materials. For example, the content of each chemical component in various chemical components of the tobacco product raw materials can be obtained by using a continuous flow method.

[0045] In some embodiments, the various chemical components may include at least one of total nitrogen, nicotine, total sugar, reducing sugar, potassium, and chlorine.

[0046] The thermogravimetric analysis data of the tobacco product raw materials includes multiple pyrolysis temperatures of the pyrolysis process of the tobacco product raw materials and the mass change rate of the tobacco product raw materials corresponding to each pyrolysis temperature among the multiple pyrolysis temperatures.

[0047] For example, the thermogravimetric analysis data of tobacco product raw materials can be obtained by conducting a pyrolysis experiment on the tobacco product raw materials and recording multiple pyrolysis temperatures during the pyrolysis process and the mass change rate of the tobacco product raw materials corresponding to each pyrolysis temperature. The mass change rate can be determined by taking the derivative of the mass of the tobacco product raw materials with respect to temperature during the pyrolysis process.

[0048] Taking a certain tobacco leaf as an example of tobacco product raw materials, the process of conducting a pyrolysis experiment on the tobacco product raw materials will be exemplarily introduced below.

[0049] First, place a certain tobacco leaf in a constant temperature and humidity chamber (for example, at a temperature of (22 ± 1) °C and a relative humidity of (60 ± 2)%) for a period of time to balance (for example, 48 h). Then, grind the tobacco leaf, filter and fully mix the ground tobacco powder through a sieve (for example, a sieve with a mesh number in the range of 80 to 100 or a sieve with a sieve hole diameter in the range of 150 micrometers (μm) to 180 micrometers), and weigh, for example, 20 mg of the tobacco powder as a sample. In a thermal analyzer, heat the sample in a nitrogen atmosphere (for example, the flow rate of nitrogen is 100 milliliters per minute (i.e., the carrier gas flow rate is 100 mL / min)) at a certain rate (for example, 10 Kelvin per minute (10 K / min)) to a certain temperature (for example, 373 Kelvin (K)), and keep it at a constant temperature for a period of time to remove the influence brought by different moisture contents of the tobacco leaf. Then, continue to heat the sample to a higher temperature (for example, 873 Kelvin (K)) at the same rate. Record the relationship between the mass and temperature of the sample during the heating process to obtain a weight loss (TG, Thermo - Gravimetry) curve. To make the tobacco pyrolysis behavior more obvious, take the derivative of the TG curve to obtain a derivative thermogravimetry (DTG) curve, which reflects the relationship between the mass change rate and temperature, that is, the "thermogravimetric analysis data" described in this article. If the thermogravimetric analysis data is presented in the form of a curve graph (i.e., the DTG curve), the abscissa represents the temperature (i.e., the "pyrolysis temperature" described in this article), with the unit of Kelvin (K), and the ordinate represents the mass change rate of the tobacco product raw materials during the pyrolysis process, with the unit of percentage / kelvin (% / K). This curve is also called a thermal analysis spectrum.

[0050] Figure 2 Schematic diagrams showing some embodiments of the DTG curve of the present disclosure.

[0051] For example, in the above example, data between 401.138 K and 801.138 K with a step size of 0.5 K can be taken as the thermogravimetric analysis data of the tobacco leaf, and presented in the form of a curve to obtain a DTG curve as shown by the black solid line in Figure 2 the following figure.

[0052] It can be understood here that the pyrolysis temperature in the thermogravimetric analysis data (i.e., the abscissa of the DTG curve) can reflect the temperature change range of the tobacco product raw material during the pyrolysis process, and different pyrolysis temperatures may correspond to different pyrolysis stages in the pyrolysis reaction. The mass change rate in the thermogravimetric analysis data (i.e., the ordinate of the DTG curve) can reflect the intensity of the reaction during the pyrolysis process of the tobacco product raw material, thereby revealing the kinetic characteristics of the pyrolysis reaction of the tobacco product raw material. That is to say, the pyrolysis temperature and the mass change rate in the thermogravimetric analysis data of the tobacco product raw material can reflect the pyrolysis characteristics of the tobacco product raw material from two different dimensions.

[0053] The multiple pyrolysis temperatures in the thermogravimetric analysis data can be evenly distributed or unevenly distributed. In order to more accurately characterize the pyrolysis characteristics of the tobacco product raw material, the temperature difference between adjacent pyrolysis temperatures among the multiple pyrolysis temperatures can be as small as possible, for example, it can be 10K, 5K, 1K, 0.5K or other values.

[0054] In step 120, characteristic data is determined from the thermogravimetric analysis data. For example, the characteristic data includes the characteristic temperature among the multiple pyrolysis temperatures and the mass change rate corresponding to the characteristic temperature.

[0055] In some embodiments, a part of the characteristic points can be selected from the DTG curve corresponding to the thermogravimetric analysis data. The value of the abscissa of this part of the characteristic points in the DTG curve is determined as the characteristic temperature in the characteristic data, and the value of the ordinate of this part of the characteristic points in the DTG curve is determined as the mass change rate corresponding to the characteristic temperature in the characteristic data.

[0056] In step 130, based on at least one parameter among the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data, the origin of the tobacco product raw material is predicted using the trained prediction model.

[0057] In some embodiments, at least one parameter among the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data can be jointly used as the input of the trained prediction model to obtain the prediction result of the origin of the tobacco product raw material output by the prediction model.

[0058] For example, at least one parameter among the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data can be jointly used as the input of the trained prediction model to obtain the prediction label output by the prediction model, and the origin of the tobacco product raw material is determined based on the prediction label. For example, different origins can correspond to different prediction labels. For example, the origin of Yunnan is represented by the number 1, and the origin of Fujian is represented by the number 2. In response to the prediction label output by the prediction model being 1, it can be determined that the origin of the tobacco product raw material is Yunnan.

[0059] It can be understood here that the number of characteristic temperatures can be one or more. Correspondingly, the number of mass change rates corresponding to the characteristic temperatures can also be one or more. The characteristic temperature and the mass change rate corresponding to the characteristic temperature are two types of parameters in different dimensions. That is to say, in step 130, based on the chemical composition data, any one type of parameter among the characteristic temperature and the mass change rate corresponding to the characteristic temperature or both types of parameters, namely the characteristic temperature and the mass change rate corresponding to the characteristic temperature, can be combined, and the trained prediction model can be used to predict the origin of the tobacco product raw material.

[0060] For example, the characteristic data may include a plurality of characteristic temperatures and the mass change rate corresponding to each characteristic temperature among the plurality of characteristic temperatures.

[0061] For example, a plurality of characteristic temperatures and the chemical composition data can be jointly input into the trained prediction model to predict the origin of the tobacco product raw material; or the mass change rate corresponding to each characteristic temperature among the plurality of characteristic temperatures and the chemical composition data can be jointly input into the trained prediction model to predict the origin of the tobacco product raw material; or a plurality of characteristic temperatures, the mass change rate corresponding to each characteristic temperature among the plurality of characteristic temperatures, and the chemical composition data can be jointly input into the trained prediction model to predict the origin of the tobacco product raw material.

[0062] In the above embodiments, the characteristic data determined from the thermogravimetric analysis data of the tobacco product raw material includes two parameters, namely the characteristic temperature and the mass change rate corresponding to the characteristic temperature, which can characterize the pyrolysis characteristics of the tobacco product raw material. The trained prediction model is used to perform fusion processing on at least one of these two parameters that can characterize the pyrolysis characteristics of the tobacco product raw material and the chemical composition data of the tobacco product raw material to predict the origin of the tobacco product raw material. In this way, in the process of predicting the origin of the tobacco product raw material, the pyrolysis characteristics of the tobacco product raw material and the contents of various chemical components in the tobacco product raw material are considered together, thereby realizing accurate prediction of the origin of the tobacco product raw material.

[0063] In some embodiments, the origin of the tobacco product raw material can be predicted based on the characteristic temperature, the mass change rate corresponding to the characteristic temperature, and the chemical composition data. For example, the characteristic temperature, the mass change rate corresponding to the characteristic temperature, and the chemical composition data can be jointly used as the input of the trained prediction model to obtain the prediction label output by the prediction model, and the origin of the tobacco product raw material can be determined based on this prediction label.

[0064] In this way, since the combination of the pyrolysis temperature and the mass change rate can more accurately reveal the pyrolysis characteristics of the tobacco product raw material, on the basis of the chemical composition data, further inputting the characteristic temperature included in the characteristic data and the mass change rate corresponding to the characteristic temperature into the trained prediction model can more accurately predict the origin of the tobacco product raw material.

[0065] The following exemplarily illustrates the characteristic data in step 120 in conjunction with some embodiments.

[0066] In some embodiments, the characteristic temperature may include the starting temperature and / or the ending temperature among multiple pyrolysis temperatures.

[0067] That is to say, the characteristic data determined from the thermogravimetric analysis data may include the starting temperature among multiple pyrolysis temperatures and the mass change rate corresponding to the starting temperature, and / or the characteristic data determined from the thermogravimetric analysis data may include the ending temperature among multiple pyrolysis temperatures and the mass change rate corresponding to the ending temperature.

[0068] The starting temperature among multiple pyrolysis temperatures can be used to represent the start of the pyrolysis process, that is, the tobacco product raw material begins to undergo thermal decomposition. The ending temperature among multiple pyrolysis temperatures can be used to represent the termination of the pyrolysis process, that is, the tobacco product raw material has completed thermal decomposition.

[0069] For example, please continue to refer to Figure 2 , as mentioned above, taking 801 data between 401.138 K and 801.138 K at a step of 0.5 K as the thermogravimetric analysis data of the tobacco leaf to obtain the Figure 2 DTG curve shown by the black solid line. That is, in the Figure 2 example shown, the starting temperature of the pyrolysis process is 401.138 K and the ending temperature is 801.138 K.

[0070] Select the starting point A1 and the ending point A2 from the DTG curve as characteristic points. The value of the abscissa of the starting point A1 is the starting temperature TA1, and the value of the ordinate of the starting point A1 is the mass change rate WA1 corresponding to the starting temperature. The value of the abscissa of the ending point A2 is the ending temperature TA2, and the value of the ordinate of the ending point A2 is the mass change rate WA2 corresponding to the ending temperature.

[0071] For example, any set of data among (TA1, TA2), (WA1, WA2), (TA1, TA2, WA1, WA2) and the chemical composition data of the tobacco leaf can be input into the trained prediction model to predict the origin of the tobacco leaf.

[0072] In the above embodiments, since the starting temperature of the pyrolysis process marks the start of thermal decomposition of the tobacco product raw material, and the ending temperature of the pyrolysis process marks the completion of thermal decomposition of the tobacco product raw material, therefore, making the characteristic data include the starting temperature, the mass change rate corresponding to the starting temperature, the ending temperature, and the mass change rate corresponding to the ending temperature can effectively utilize the thermal stability characteristics of the tobacco product raw material for origin prediction. Thus, the accuracy of origin prediction of the tobacco product raw material can be further improved.

[0073] In some embodiments, multiple pyrolysis temperatures form a temperature range. The mass change rate corresponding to the characteristic temperature includes one or more extreme values of the mass change rate of the tobacco product raw material within the temperature range. That is to say, the characteristic data determined from the thermogravimetric analysis data may include at least one extreme value of the mass change rate of the tobacco product raw material and the temperature corresponding to this at least one extreme value. The at least one extreme value may include a maximum value and / or a minimum value. Here, if the mass change rate is differentiated with respect to temperature, the extreme value is the mass change rate corresponding to when the derivative is 0. Or, if the pyrolysis temperature and the mass change rate are plotted as a curve, the extreme value corresponds to the peak or valley value on the curve.

[0074] For example, as Figure 2 shown, the lower limit of the temperature range formed by multiple pyrolysis temperatures is 401.138 K and the upper limit is 801.138 K. Multiple points (B1 - B5) with a slope of 0 on the DTG curve are selected as characteristic points. Among them, B1 and B4 are the trough points on the DTG curve, and B2, B3, and B5 are the peak points on the DTG curve.

[0075] The ordinate values (WB1 - WB5) of these characteristic points are the extreme values of the mass change rate of the tobacco leaf within the temperature range of 401.138 K to 801.138 K, and the abscissa values (TB1 - TB5) of these characteristic points are the temperatures corresponding to the extreme values (i.e., the characteristic temperatures).

[0076] For example, any one set of data among (TB1 - TB5), (WB1 - WB5), ((TB1, WB1) - (TB5, WB5)) and the chemical composition data of the tobacco leaf can be input into the trained prediction model to predict the origin of the tobacco leaf.

[0077] In the above embodiments, since the peak points and trough points in the curve corresponding to the thermogravimetric analysis data can reflect the main thermal decomposition stages of the tobacco product raw material, therefore, making the characteristic data include one or more extreme values of the mass change rate of the tobacco product raw material within the temperature range of the pyrolysis process and the corresponding characteristic temperatures can effectively utilize the characteristics of the tobacco product raw material in each thermal decomposition stage for origin prediction. Thus, the accuracy of origin prediction of the tobacco product raw material can be further improved.

[0078] In some embodiments, the characteristic data determined from the thermogravimetric analysis data may include: the starting temperature and the mass change rate corresponding to the starting temperature, the termination temperature and the mass change rate corresponding to the termination temperature, at least one extreme value of the mass change rate of the tobacco product raw material, and the temperature corresponding to this at least one extreme value.

[0079] For example, the starting point A1, the termination point A2, and the characteristic points B1 - B5 can be selected together from the DTG curve to obtain the characteristic data for predicting the origin of the tobacco product raw material.

[0080] Please continue to refer to Figure 2 , the curve obtained after curve fitting based on these 7 characteristic points A1, A2, and B1 to B5 is as Figure 2 shown by the dashed line in. It can be seen that the curve fitted based on these 7 characteristic points can more accurately reflect the trend of the actual DTG curve. This indicates that the characteristic data obtained using these 7 characteristic points can more accurately reflect the pyrolysis characteristics of the tobacco product raw material during the pyrolysis process. Therefore, using these characteristic data can help accurately predict the origin of the tobacco product raw material.

[0081] In some embodiments, multiple pyrolysis temperatures form adjacent first and second temperature intervals.

[0082] The mass change rate corresponding to the characteristic temperature includes the intermediate value of the first extreme value of the mass change rate of the tobacco product raw material within the first temperature interval and the second extreme value of the mass change rate of the tobacco product raw material within the second temperature interval. In other words, the mass change rate corresponding to the characteristic temperature can include the intermediate value of the two extreme values corresponding to the mass change rate of the tobacco product raw material within two adjacent temperature intervals.

[0083] In some embodiments, the characteristic temperature corresponding to the intermediate value includes the first temperature corresponding to the intermediate value in the thermogravimetric analysis data. The temperature corresponding to the first extreme value is less than the temperature corresponding to the second extreme value. The first temperature is greater than the temperature corresponding to the first extreme value and less than the temperature corresponding to the second extreme value. In other words, the characteristic temperature corresponding to the intermediate value can include the first temperature between the temperature corresponding to the first extreme value and the temperature corresponding to the second extreme value.

[0084] In this way, further enabling the characteristic data to include the intermediate value of the two extreme values corresponding to the mass change rate of the tobacco product raw material within two adjacent temperature intervals and the first temperature corresponding to the intermediate value can further effectively utilize the characteristics of the tobacco product raw material in each pyrolysis stage for origin prediction. Thus, the accuracy of predicting the origin of the tobacco product raw material can be further improved.

[0085] In some embodiments, the characteristic temperature corresponding to the intermediate value includes a second temperature corresponding to the intermediate value in the thermogravimetric analysis data. For example, the second temperature is less than the temperature corresponding to the first extreme value or greater than the temperature corresponding to the second extreme value. In other words, the characteristic temperature corresponding to the intermediate value may include a second temperature less than the temperature corresponding to the first extreme value or a second temperature greater than the temperature corresponding to the second extreme value.

[0086] Thus, considering that the intermediate value may correspond to multiple different temperatures in the thermogravimetric analysis data, further including the intermediate value and the second temperature corresponding to the intermediate value that is different from the first temperature in the characteristic data can more effectively utilize the characteristics of the tobacco product raw material in each thermal decomposition stage for origin prediction. Therefore, the accuracy of the origin prediction of the tobacco product raw material can be further improved.

[0087] The following combines Figure 3 to further illustrate the case where the characteristic data includes the intermediate value of the two extreme values of the mass change rate of the tobacco product raw material in two adjacent temperature intervals and the temperature corresponding to the intermediate value.

[0088] Figure 3 Schematic diagrams showing other embodiments of the DTG curve of the present disclosure.

[0089] Figure 3 Continuing Figure 2 the example shown, the DTG curve of the example shown above is also shown as a solid black line. As Figure 3 shown, A1 and A2 are the starting point and the ending point on the DTG curve respectively. B1 and B4 are the trough points on the DTG curve, and B2, B3, and B5 are the peak points on the DTG curve.

[0090] For example, B1 and B2 are the extreme values in two adjacent temperature intervals. The value of the ordinate of the characteristic point B1 can be taken as the first extreme value, the value of the ordinate of the characteristic point B2 can be taken as the second extreme value, and the points C1 and D1 corresponding to the intermediate value of the values of the ordinates of the characteristic points B1 and B2 on the DTG curve are selected as characteristic points. Among them, the value of the abscissa of the characteristic point C1 is between the values of the abscissas corresponding to the characteristic points B1 and B2, that is, the value of the abscissa of the characteristic point C1 is the first temperature TC1 corresponding to the intermediate value WC1; the value of the abscissa of the characteristic point D1 is less than the value of the abscissa corresponding to the characteristic point B1, that is, the value of the abscissa of the characteristic point D1 is the second temperature TD1 corresponding to the intermediate value WD1. And so on.

[0091] It can be understood here that the mid-value of the ordinate value of feature point B1 and the ordinate value of feature point B2 can also be selected, and the corresponding point D1' on the DTG curve can be used as a feature point. The abscissa value of feature point D1' is greater than the abscissa value corresponding to feature point B2, that is, the abscissa value of feature point D1' is the second temperature TD1' corresponding to the mid-value WD1'.

[0092] Figure 3 Eight feature points (also known as "half-peak points") C1 to C4 and D1 to D4 selected are schematically shown. Any combination of one or two types of data among the temperatures and mass change rates of these eight feature points can be used as the characteristic temperature and / or the corresponding mass change rate and input into the trained prediction model to predict the growth location of the tobacco leaf.

[0093] For example, any set of data among (TC1~TC4), (WC1~WC4), ((TC1, WC1)~(TC4, WC4)) and the chemical composition data of the tobacco leaf can be input into the trained prediction model to predict the origin of the tobacco leaf.

[0094] For example, any set of data among (TD1~TD4), (WD1~WD4), ((TD1, WD1)~(TD4, WD4)) and the chemical composition data of the tobacco leaf can be input into the trained prediction model to predict the origin of the tobacco leaf.

[0095] For example, any set of data among (TC1~TC4 and TD1~TD4), (WC1~WC4 and WD1~WD4), ((TC1, WC1)~(TC4, WC4), (TD1, WD1)~(TD4, WD4)) and the chemical composition data of the tobacco leaf can be input into the trained prediction model to predict the origin of the tobacco leaf.

[0096] In some embodiments, the characteristic data determined from the thermogravimetric analysis data may include: the starting temperature and the mass change rate corresponding to the starting temperature, the termination temperature and the mass change rate corresponding to the termination temperature, at least one extreme value of the mass change rate of the tobacco product raw material and the temperature corresponding to this at least one extreme value, the mid-value of the two extreme values corresponding to the mass change rate of the tobacco product raw material in two adjacent temperature intervals and the first temperature corresponding to the mid-value, the mid-value of the two extreme values corresponding to the mass change rate of the tobacco product raw material in two adjacent temperature intervals and the second temperature corresponding to the mid-value.

[0097] For example, starting point A1, ending point A2, characteristic points B1 - B5 (also known as "inflection points"), "half - peak points" C1 - C4, and D1 - D4 can be selected from the DTG curve together to obtain characteristic data for predicting the origin of tobacco product raw materials.

[0098] The following schematically shows, through Table 1, the characteristic data (also known as "characteristic parameters") obtained based on these characteristic points.

[0099] Table 1

[0100] Please continue to refer to Figure 3 , the curve obtained after curve fitting based on the 15 characteristic points A1, A2, B1 - B5, C1 - C4, and D1 - D4 is as shown by the dashed line in Figure 3 . It can be seen that the trend of the curve fitted based on these 15 characteristic points is basically the same as that of the actual DTG curve. This indicates that the 30 characteristic data obtained using these 15 characteristic points can quite accurately reflect the pyrolysis characteristics of tobacco product raw materials during the pyrolysis process. Therefore, using these characteristic data can further help accurately predict the origin of tobacco product raw materials.

[0101] In some embodiments, the chemical composition data of the tobacco product raw materials may also include the ratios of the contents of different chemical components among various chemical components in the tobacco product raw materials. For example, the various chemical components may include total nitrogen, nicotine, total sugar, reducing sugar, potassium, and chlorine. The ratios of the contents of different chemical components among the various chemical components may include at least one of the potassium - to - chlorine ratio (i.e., the ratio of the contents of potassium and chlorine), the nitrogen - to - nicotine ratio (i.e., the ratio of the contents of total nitrogen and nicotine), the sugar - to - nicotine ratio (i.e., the ratio of the contents of total sugar and nicotine), and the two - sugar ratio (i.e., the ratio of the contents of reducing sugar and total sugar).

[0102] In this way, through the analysis of the ratios of the contents of different chemical components in the tobacco product raw materials, the origin characteristics of the tobacco product raw materials can be comprehensively judged. In the process of predicting the origin of tobacco product raw materials, further considering the contents of various chemical components and the ratios of the contents in the tobacco product raw materials can more accurately predict the origin of the tobacco product raw materials.

[0103] The foregoing introduced the related implementation of predicting the origin of tobacco product raw materials using a trained prediction model. The following further introduces the training method of the prediction model for predicting the origin of tobacco product raw materials.

[0104] Figure 4 The flowchart showing some embodiments of the training method of the prediction model of the present disclosure is shown. As Figure 4 shown, for example, the training method of the prediction model may include step 410 to step 430. Among them, this prediction model is used to predict the origin of tobacco product raw materials.

[0105] In step 410, thermogravimetric analysis data and chemical composition data of each tobacco product raw material sample among multiple tobacco product raw material samples from multiple different origins are obtained.

[0106] The thermogravimetric analysis data of each tobacco product raw material sample includes multiple sample pyrolysis temperatures in the pyrolysis process of each tobacco product raw material sample and the mass change rate corresponding to each sample pyrolysis temperature among the multiple sample pyrolysis temperatures. The chemical composition data of each tobacco product raw material sample includes the content of each chemical component among multiple chemical components in each tobacco product raw material sample.

[0107] To improve the generalization ability of model training, when selecting training data, it is possible to consider selecting tobacco product raw material samples from multiple origins covering multiple different flavor styles. In some embodiments, the multiple origins can cover three flavor styles: light flavor type, intermediate flavor type, and strong flavor type. For a similar purpose, when selecting training data, it is further possible to consider selecting tobacco product raw material samples covering multiple growth parts for each origin. Here, the growth part to which the tobacco product raw material belongs is, for example, the growth part of the tobacco leaf on the tobacco plant, such as the upper part, the middle part, or the lower part. In some embodiments, the tobacco product raw material samples from one or more origins may include upper tobacco leaves, middle tobacco leaves, and lower tobacco leaves. For example, at least one corresponding origin's tobacco product raw material sample can be selected for the light flavor type, intermediate flavor type, and strong flavor type respectively, and the selected tobacco product raw material samples for each origin cover upper tobacco leaves, middle tobacco leaves, and lower tobacco leaves, thereby constructing a training data set.

[0108] In this way, the constructed training data set can more comprehensively cover the characteristics of various types of tobacco product raw materials. Using such a training data set to train the prediction model can help the prediction model learn more extensive characteristics of tobacco product raw materials from different origins, improve the generalization ability of the model, and thus contribute to improving the prediction performance of the model.

[0109] In step 420, characteristic data of each tobacco product raw material sample is determined from the thermogravimetric analysis data of each tobacco product raw material sample.

[0110] The characteristic data of each tobacco product raw material sample includes the sample characteristic temperature among the multiple sample pyrolysis temperatures and the mass change rate corresponding to the sample characteristic temperature.

[0111] The selection method of the characteristic data of the tobacco product raw material sample during the training process of the prediction model is similar to the selection method of the characteristic data of the tobacco product raw material during the prediction process using the prediction model. For specific descriptions, reference can be made to the relevant embodiments above. Details are not repeated here.

[0112] In step 430, the prediction model is trained with at least one of the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature, as well as the chemical composition data of each tobacco product raw material sample as the input, and the predicted origin of each tobacco product raw material sample as the output, until the training end condition is met. The training end condition can be, for example, that the number of training times reaches a specified number and / or the difference between the predicted origin of each tobacco product raw material sample and the actual origin of each tobacco product raw material sample is less than a threshold. The difference between the predicted origin and the actual origin can be characterized by the loss function value of the prediction model.

[0113] It can be understood here that the types of data input into the prediction model during the training of the prediction model are the same as the types of data input into the prediction model during the prediction using the prediction model.

[0114] For example, if the prediction model is trained with the sample characteristic temperature and the chemical composition data of each tobacco product raw material sample as the input and the predicted origin of each tobacco product raw material sample as the output, then during the prediction using the trained prediction model, the characteristic temperature in the characteristic data and the chemical composition data of the tobacco product raw material are jointly input into the prediction model to obtain the prediction result of the origin of the tobacco product raw material.

[0115] For example, if the prediction model is trained with the sample characteristic temperature, the mass change rate corresponding to the sample characteristic temperature, and the chemical composition data of each tobacco product raw material sample as the input and the predicted origin of each tobacco product raw material sample as the output, then during the prediction using the trained prediction model, the characteristic temperature in the characteristic data, the mass change rate corresponding to the characteristic temperature, and the chemical composition data of the tobacco product raw material are jointly input into the prediction model to obtain the prediction result of the origin of the tobacco product raw material.

[0116] In some embodiments, the prediction model may include a random forest model. Since the random forest model has good interpretability and is more suitable for large-scale data sets and structured data, using the random forest model for predicting the origin of tobacco product raw materials can obtain better prediction results.

[0117] Figure 5 A flowchart showing other embodiments of the training method of the prediction model of the present disclosure.

[0118] As Figure 5 shown, for example, the training method of the prediction model may include steps 510 to 580. This method can be executed as Figure 4 a specific implementation of the method in

[0119] In step 510, tobacco product raw material samples are selected. For example, tobacco product raw material samples from 8 ecological zones are selected according to light aroma type, medium aroma type, and strong aroma type, each producing zone covers at least 30 different tobacco product raw material samples, and the selected tobacco product raw material samples cover upper tobacco leaves, middle tobacco leaves, and lower tobacco leaves.

[0120] In step 520, chemical composition data for each tobacco product raw material sample is obtained.

[0121] For example, the chemical composition data of each tobacco product raw material sample is obtained by a continuous flow method. The chemical composition data of each tobacco product raw material sample may include total nitrogen, nicotine, total sugar, reducing sugar, potassium, chlorine, potassium-chlorine ratio, nitrogen-alkali ratio, sugar-alkali ratio and two sugar ratio.

[0122] In step 530, a pyrolysis experiment is performed on each tobacco product raw material sample.

[0123] For example, each tobacco product raw material sample is placed in a constant temperature and humidity chamber for a period of time (e.g., 48 hours). After being taken out and ground, it is sieved and fully mixed. For example, 20 mg of tobacco powder is weighed as a sample and placed in an alumina flat crucible. The sample is heated from room temperature to 873K in a nitrogen atmosphere in a thermal analyzer.

[0124] In step 540, thermogravimetric analysis data is obtained for each tobacco product raw material sample.

[0125] For example, for each tobacco product raw material sample, with a step length of 0.5 K, a total of 801 data between 401.138 K and 801.138 K are taken as the thermogravimetric analysis data of the tobacco product raw material sample.

[0126] In step 550, characteristic data of each tobacco product raw material sample is determined from the thermogravimetric analysis data of each tobacco product raw material sample. For example, 30 characteristic data are selected from the DTG curve to characterize the thermogravimetric analysis data of the tobacco product raw material sample.

[0127] For example, refer to the previous article Figures 2 to 3 In the relevant example, the 30 characteristic data of each tobacco product raw material sample can be determined by selecting the starting point A1, the ending point A2, the characteristic points B1~B5, the "half-peak points" C1~C4 and D1~D4 from the DTG curve of each tobacco product raw material sample, and then determining the values ​​of the horizontal and vertical coordinates of these 15 characteristic points in the DTG curve.

[0128] In step 560, a training data set is constructed. For example, the training data set is constructed in the form of "feature data - chemical composition data - origin".

[0129] In step 570, a prediction model is trained based on a training data set to obtain a trained prediction model (also referred to as a "style prediction model").

[0130] For example, the prediction model is a random forest model. Using the sample characteristic temperature in the characteristic data of each tobacco product raw material sample, the mass change rate corresponding to the sample characteristic temperature, and the chemical composition data of each tobacco product raw material sample as inputs, and the predicted origin of each tobacco product raw material sample as the output, the random forest model is trained until the training end condition is met. The training end condition can be, for example, that the number of training times reaches a specified number and / or the difference between the predicted origin of each tobacco product raw material sample and the actual origin of each tobacco product raw material sample is less than a threshold. The difference between the predicted origin and the actual origin can be characterized by the loss function value of the prediction model.

[0131] In step 580, the characteristic data and chemical composition data of the tobacco product raw material sample used as a test sample are input into the trained prediction model to obtain a prediction result output by the prediction model.

[0132] It should be recognized that in Figure 2 、 Figure 3 and Figure 5 's examples, the pyrolysis temperature and the number of characteristic points (characteristic temperature and the corresponding mass change rate) are only exemplary.

[0133] For more embodiments of the training method of the prediction model provided by the present disclosure, reference can be made to the description of the method and its embodiments for predicting the origin of tobacco product raw materials in the foregoing text.

[0134] Next, in conjunction with some embodiments, the performance of the prediction model trained by the training method of the present disclosure and the prediction effect of predicting the origin of tobacco product raw materials using the prediction model are exemplarily described.

[0135] 434 tobacco product raw material samples from 8 production areas, namely production area A, production area B, production area C, production area D, production area E, production area F, production area G, and production area H, are selected.

[0136] The 10 chemical composition data of each of the 434 tobacco product raw material samples are measured using a continuous flow method. For example, these 10 chemical composition data include total nitrogen, nicotine, total sugar, reducing sugar, potassium, chlorine, potassium-chlorine ratio, nitrogen-base ratio, sugar-base ratio, and disaccharide ratio.

[0137] For each of the 434 tobacco product raw material samples, a corresponding tobacco powder sample was selected for thermogravimetric testing. In a thermal analyzer, the tobacco powder sample was heated from room temperature to 873 K at a certain heating rate (10 K / min) under a nitrogen atmosphere (carrier gas flow rate of 100 mL / min), and the mass change of the tobacco powder sample during heating was measured to obtain thermogravimetric data (i.e., the data corresponding to the TG curve).

[0138] 801 thermogravimetric analysis data (i.e., the data corresponding to the DTG curve) were obtained by differentiating the thermogravimetric data.

[0139] 30 thermogravimetric analysis data were selected from the 801 thermogravimetric analysis data as characteristic data (also known as "thermogravimetric characteristics") to represent these 801 thermogravimetric analysis data. Each characteristic data includes a characteristic temperature and the mass change rate corresponding to the characteristic temperature.

[0140] A dataset was constructed in the form of "characteristic data - chemical composition data - production area" sample pairs, and the dataset was divided into a training set and a test set in a ratio of 8:2.

[0141] To construct a prediction model, the origin of the sample was used as the target value of the model output. On this basis, the following five types of input data were selected for model training: (1) 10 chemical composition data of each tobacco product raw material sample; (2) 801 thermogravimetric analysis data of each tobacco product raw material sample; (3) 30 thermogravimetric characteristics of each tobacco product raw material sample; (4) 801 thermogravimetric analysis data and 10 chemical composition data of each tobacco product raw material sample; (5) 30 thermogravimetric characteristics and 10 chemical composition data of each tobacco product raw material sample. Among them, the data in categories (1) to (3) are also known as non-fused data, and the data in categories (4) to (5) are also known as fused data.

[0142] Subsequently, three classification algorithms, namely random forest, K-nearest neighbor (KNN), and linear discriminant analysis (LDA), were used to train the model for the above five types of input data, and the classification accuracy was used as an evaluation index for the prediction performance of the trained model.

[0143] The five-fold cross-validation method is used to divide the above-constructed dataset, and the dataset is evenly split into five subsets with equal data volumes. In each validation process, four of these subsets are selected for model training, while the remaining one subset is used for model testing. This division and validation process are repeated five times, labeled as k1, k2, k3, k4, and k5 respectively. In each validation, a different subset is selected as the test set, and the remaining subsets are used as the training set. The prediction accuracy of the trained model on each test set will be recorded, and finally, the average of the prediction accuracies of the five validations will be calculated as a comprehensive evaluation metric for the model's prediction performance.

[0144] Tables 2, 3, and 4 show the prediction results of three algorithms respectively. In the tables, the results with the best comprehensive evaluation metric are underlined, and the results with the second-best comprehensive evaluation metric are italicized and underlined.

[0145] Table 2 Prediction Results of the Random Forest Algorithm for Different Types of Input Data

[0146] Table 3 Prediction Results of the KNN Algorithm for Different Types of Input Data

[0147] Table 4 Prediction Results of the LDA Algorithm for Different Types of Input Data

[0148] Taking the prediction results of the random forest algorithm shown in Table 2 as an example, its prediction effect shows that in non-fused data, the prediction effect of using chemical composition data (i.e., type (1) data) for prediction is relatively poor, with an average prediction accuracy of 84.55%; while the prediction effect of using thermogravimetric analysis data (i.e., type (2) data) for prediction is relatively good, with an average prediction accuracy reaching 87.54%. After fusing the chemical composition data with the thermogravimetric characteristics, the prediction effect of the model is significantly improved, and its average prediction accuracy is increased to 95.40%. Among them, the maximum prediction accuracy of the random forest model reaches 96.55%. This indicates that using the random forest model for origin prediction and using the fused data of chemical composition data and thermogravimetric characteristics as the input of the model can effectively enhance the prediction performance of the model, thereby improving the accuracy of origin prediction.

[0149] Next, it can be seen from the prediction results of the KNN algorithm shown in Table 3 that the prediction methods using the thermogravimetric analysis data and the fusion data of the thermogravimetric characteristics and chemical composition data respectively achieved the optimal and sub-optimal prediction results. In contrast, in the prediction method using only chemical composition data, the average prediction accuracy of the model was relatively low, only 67.51%. Further, the results of the LDA algorithm in Table 4 also show that the prediction method using fusion data has a significant improvement in prediction performance compared with the method using non-fusion data.

[0150] Generally speaking, the prediction performance of the random forest algorithm is better than that of other algorithms, and the minimum value of the average prediction accuracy is still greater than 80%. Moreover, the optimal result obtained by the random forest algorithm (the average accuracy is 95.40% and the highest prediction accuracy is 96.55%) was achieved by using the data fused with the thermogravimetric characteristics and chemical composition data as the input of the model.

[0151] It can be seen that the prediction method using fusion data has better prediction performance than the method using non-fusion data in terms of prediction effect, that is, it can more accurately predict the origin of tobacco product raw materials. This also verifies the technical solution of using the characteristic data that can characterize the pyrolysis characteristics of tobacco product raw materials and the chemical composition data of tobacco product raw materials provided by the embodiments of the present disclosure to predict the origin of tobacco product raw materials, which can effectively improve the prediction accuracy.

[0152] Figure 6 The block diagram showing some embodiments of the device for predicting the origin of tobacco product raw materials of the present disclosure.

[0153] As Figure 6 shown, the device 600 for predicting the origin of tobacco product raw materials includes an acquisition module 601, a determination module 602, and a prediction module 603.

[0154] The acquisition module 601 can be configured to acquire the thermogravimetric analysis data and chemical composition data of tobacco product raw materials. The thermogravimetric analysis data of tobacco product raw materials includes multiple pyrolysis temperatures in the pyrolysis process of tobacco product raw materials and the mass change rate of tobacco product raw materials corresponding to each pyrolysis temperature among the multiple pyrolysis temperatures. The chemical composition data of tobacco product raw materials includes the content of each chemical component among multiple chemical components in tobacco product raw materials.

[0155] The determination module 602 can be configured to determine characteristic data from the thermogravimetric analysis data. The characteristic data includes the characteristic temperature among multiple pyrolysis temperatures and the mass change rate corresponding to the characteristic temperature.

[0156] The prediction module 603 may be configured to predict the origin of the tobacco product raw material by using the trained prediction model based on at least one parameter of the characteristic temperature and the mass change rate corresponding to the characteristic temperature, as well as the chemical composition data.

[0157] In some embodiments, the apparatus 600 for predicting the origin of the tobacco product raw material may further include other modules that perform other operations in the relevant embodiments of the method for predicting the origin of the tobacco product raw material described above.

[0158] Figure 7 The block diagram showing some embodiments of the training apparatus of the prediction model of the present disclosure.

[0159] As Figure 7 shown, the training apparatus 700 of the prediction model includes an acquisition module 701, a determination module 702, and a training module 703. The training apparatus 700 of the prediction model is used to predict the origin of the tobacco product raw material.

[0160] The acquisition module 701 is configured to acquire the thermogravimetric analysis data and the chemical composition data of each tobacco product raw material sample from multiple tobacco product raw material samples from multiple different origins. The thermogravimetric analysis data of each tobacco product raw material sample includes multiple sample pyrolysis temperatures of the pyrolysis process of each tobacco product raw material sample and the mass change rate corresponding to each sample pyrolysis temperature among the multiple sample pyrolysis temperatures. The chemical composition data of each tobacco product raw material sample includes the content of each chemical component among multiple chemical components in each tobacco product raw material sample.

[0161] The determination module 702 is configured to determine the characteristic data of each tobacco product raw material sample from the thermogravimetric analysis data of each tobacco product raw material sample. The characteristic data of each tobacco product raw material sample includes the sample characteristic temperature among the multiple sample pyrolysis temperatures and the mass change rate corresponding to the sample characteristic temperature.

[0162] The training module 703 is configured to train the prediction model with at least one parameter of the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature, as well as the chemical composition data of each tobacco product raw material sample as the input and the predicted origin of each tobacco product raw material sample as the output until the training end condition is met. The training end condition may be, for example, that the number of training times reaches a specified number and / or the difference between the predicted origin of each tobacco product raw material sample and the actual origin of each tobacco product raw material sample is less than a threshold. The difference between the predicted origin and the actual origin can be characterized by the loss function value of the prediction model.

[0163] In some embodiments, the training apparatus 700 of the prediction model may further include other modules that perform other operations in the relevant embodiments of the training method of the prediction model described above.

[0164] Figure 8 Block diagram showing some embodiments of the electronic device of the present disclosure.

[0165] As Figure 8 shown, the electronic device 800 of this embodiment includes: a memory 801 and a processor 802 coupled to the memory 801. The processor 802 is configured to execute the method in any one of the embodiments of the present disclosure based on the instructions stored in the memory 801.

[0166] Among them, the memory 801 may include, for example, a system memory, a fixed non-volatile storage medium, etc. The system memory stores, for example, an operating system, application programs, a boot loader, a database, and other programs.

[0167] In some embodiments, the electronic device 800 may be used as a device for predicting the origin of tobacco product raw materials to perform the operations of any one of the above embodiments. In other embodiments, the electronic device 800 may be used as a training device for a prediction model to perform the operations of any one of the above embodiments.

[0168] Figure 9 Block diagram showing some other embodiments of the electronic device of the present disclosure.

[0169] As Figure 9 shown, the electronic device 900 of this embodiment includes: a memory 901 and a processor 902 coupled to the memory 901. The processor 902 is configured to execute the method in any one of the foregoing embodiments based on the instructions stored in the memory 901.

[0170] The memory 901 may include, for example, a system memory, a fixed non-volatile storage medium, etc. The system memory stores, for example, an operating system, application programs, a boot loader, and other programs.

[0171] The electronic device 900 may further include an input / output interface 903, a network interface 904, a storage interface 905, etc. These interfaces 903, 904, 905 and the memory 901 and the processor 902 may be connected through a bus 906, for example. Among them, the input / output interface 903 provides a connection interface for input / output devices such as a display, a mouse, a keyboard, a touch screen, a microphone, and a speaker. The network interface 904 provides a connection interface for various networking devices. The storage interface 905 provides a connection interface for external storage devices such as an SD card and a USB flash drive.

[0172] In some embodiments, the electronic device 900 may serve as a device for predicting the origin of tobacco product raw materials that performs the operations of any of the above embodiments. In other embodiments, the electronic device 900 may serve as a training device for a prediction model that performs the operations of any of the above embodiments.

[0173] Embodiments of the present disclosure also provide a computer-readable storage medium, including computer program instructions that, when executed by a processor, implement the method of any of the above embodiments.

[0174] Embodiments of the present disclosure also provide a computer program product, including a computer program that, when executed by a processor, implements the method of any of the above embodiments.

[0175] Those skilled in the art should understand that the embodiments of the present disclosure may be provided as a method, a system, or a computer program product. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable non-transitory storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) that contain computer-usable program code.

[0176] So far, the technical solutions for predicting the origin of tobacco product raw materials and the technical solutions for training a prediction model according to the present disclosure have been described in detail. To avoid obscuring the concept of the present disclosure, some details well known in the art have not been described. Those skilled in the art can fully understand how to implement the technical solutions disclosed herein based on the above description.

[0177] The methods and systems of the present disclosure may be implemented in many ways. For example, the methods and systems of the present disclosure may be implemented through software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is only for illustration, and the steps of the method of the present disclosure are not limited to the specific order described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, and these programs include machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0178] Although some specific embodiments of the present disclosure have been described in detail by way of examples, those skilled in the art should understand that the above examples are only for illustration and not for limiting the scope of the present disclosure. Those skilled in the art should understand that the above embodiments may be modified without departing from the scope and spirit of the present disclosure. The scope of the present disclosure is defined by the appended claims.

Claims

1. A method for predicting the origin of raw materials for tobacco products, comprising: Acquire thermogravimetric analysis data and chemical composition data of a tobacco product raw material, wherein the thermogravimetric analysis data includes a plurality of pyrolysis temperatures during the pyrolysis process of the tobacco product raw material and a mass change rate of the tobacco product raw material corresponding to each of the plurality of pyrolysis temperatures, and the chemical composition data includes the content of each of the plurality of chemical components in the tobacco product raw material; Determining characteristic data from the thermogravimetric analysis data, the characteristic data comprising characteristic temperatures among the plurality of pyrolysis temperatures and mass change rates corresponding to the characteristic temperatures; Based on at least one parameter of the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data, the origin of the tobacco product raw material is predicted using a trained prediction model.

2. The method according to claim 1, wherein: The multiple pyrolysis temperatures form a temperature interval, and the mass change rate corresponding to the characteristic temperature includes one or more extreme values ​​of the mass change rate of the tobacco product raw material within the temperature interval.

3. The method according to claim 1, wherein: The multiple pyrolysis temperatures form a first temperature interval and a second temperature interval adjacent to each other. The mass change rate corresponding to the characteristic temperature includes an intermediate value between a first extreme value of the mass change rate of the tobacco product raw material in the first temperature interval and a second extreme value of the mass change rate of the tobacco product raw material in the second temperature interval. The characteristic temperature includes a first temperature corresponding to the intermediate value in the thermogravimetric analysis data, a temperature corresponding to the first extreme value is lower than a temperature corresponding to the second extreme value, and the first temperature is higher than a temperature corresponding to the first extreme value and lower than a temperature corresponding to the second extreme value.

4. The method according to claim 3, wherein: The characteristic temperature includes a second temperature corresponding to the intermediate value in the thermogravimetric analysis data, where the second temperature is less than a temperature corresponding to the first extreme value or greater than a temperature corresponding to the second extreme value.

5. The method according to any one of claims 1 to 4, wherein: The characteristic temperature includes a starting temperature and / or an ending temperature among the plurality of pyrolysis temperatures.

6. The method according to any one of claims 1 to 4, wherein: The chemical composition data also includes the ratio of the content of different chemical components among the multiple chemical components in the tobacco product raw material.

7. The method according to any one of claims 1 to 4, wherein: The trained prediction model is trained in the following manner: Obtaining thermogravimetric analysis data and chemical composition data of each tobacco product raw material sample from a plurality of tobacco product raw material samples from a plurality of different origins, wherein the thermogravimetric analysis data of each tobacco product raw material sample includes a plurality of sample pyrolysis temperatures during the pyrolysis process of each tobacco product raw material sample and a mass change rate corresponding to each sample pyrolysis temperature among the plurality of sample pyrolysis temperatures, and the chemical composition data of each tobacco product raw material sample includes the content of each chemical component among the plurality of chemical components in each tobacco product raw material sample; Determining characteristic data of each tobacco product raw material sample from the thermogravimetric analysis data of each tobacco product raw material sample, wherein the characteristic data of each tobacco product raw material sample includes a sample characteristic temperature among the plurality of sample pyrolysis temperatures and a mass change rate corresponding to the sample characteristic temperature; and The prediction model is trained with at least one parameter of the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature and the chemical composition data of each tobacco product raw material sample as input and the predicted origin of each tobacco product raw material sample as output until the training end condition is met.

8. The method according to any one of claims 1 to 4, wherein: The trained prediction model includes a random forest model.

9. The method according to any one of claims 1 to 4, wherein: The plurality of chemical components include at least one of total nitrogen, nicotine, total sugar, reducing sugar, potassium and chlorine.

10. A method for training a prediction model, the prediction model being used to predict the origin of tobacco product raw materials, the training method comprising: Obtaining thermogravimetric analysis data and chemical composition data of each tobacco product raw material sample from a plurality of tobacco product raw material samples from a plurality of different origins, wherein the thermogravimetric analysis data of each tobacco product raw material sample includes a plurality of sample pyrolysis temperatures during the pyrolysis process of each tobacco product raw material sample and a mass change rate corresponding to each sample pyrolysis temperature among the plurality of sample pyrolysis temperatures, and the chemical composition data of each tobacco product raw material sample includes the content of each chemical component among the plurality of chemical components in each tobacco product raw material sample; Determining characteristic data of each tobacco product raw material sample from the thermogravimetric analysis data of each tobacco product raw material sample, wherein the characteristic data of each tobacco product raw material sample includes a sample characteristic temperature among the plurality of sample pyrolysis temperatures and a mass change rate corresponding to the sample characteristic temperature; and The prediction model is trained with at least one parameter of the sample characteristic temperature and the mass change rate corresponding to the sample characteristic temperature and the chemical composition data of each tobacco product raw material sample as input and the predicted origin of each tobacco product raw material sample as output until the training end conditions are met.

11. A device for predicting the origin of raw materials for tobacco products, comprising: an acquisition module, configured to acquire thermogravimetric analysis data and chemical composition data of a tobacco product raw material, wherein the thermogravimetric analysis data includes a plurality of pyrolysis temperatures during the pyrolysis process of the tobacco product raw material and a mass change rate of the tobacco product raw material corresponding to each of the plurality of pyrolysis temperatures, and the chemical composition data includes a content of each of the plurality of chemical components in the tobacco product raw material; A determination module configured to determine characteristic data from the thermogravimetric analysis data, wherein the characteristic data includes a characteristic temperature among the plurality of pyrolysis temperatures and a mass change rate corresponding to the characteristic temperature; The prediction module is configured to predict the origin of the tobacco product raw material using a trained prediction model based on at least one parameter of the characteristic temperature and the mass change rate corresponding to the characteristic temperature and the chemical composition data.

12. An electronic device, comprising: Memory; and A processor coupled to the memory, the processor being configured to execute the method for predicting the origin of tobacco product raw materials as described in any one of claims 1-9 or the method for training a prediction model as described in claim 10 based on instructions stored in the memory.

13. A computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the method for predicting the origin of tobacco product raw materials as described in any one of claims 1 to 9 or the method for training a prediction model as described in claim 10.

14. A computer program product, comprising instructions, which, when executed by a processor, cause the processor to execute the method for predicting the origin of tobacco product raw materials according to any one of claims 1 to 9 or the method for training a prediction model according to claim 10.

Images