Method, device, equipment, medium and product for determining the amount of tobacco in a tobacco filter rod

Abstract

This application relates to a method, apparatus, equipment, medium, and product for determining the packing amount of tobacco filter rods, belonging to the field of tobacco industry filter rod production technology. The method includes: acquiring packing amount data and pressure drop data of sample tobacco filter rods of different rod types; performing correlation analysis, normality test processing, and stability verification processing on the packing amount data and pressure drop data of sample tobacco filter rods of each rod type, respectively, to obtain correlation analysis results, normality test results, and stability verification results; obtaining the target model corresponding to the rod type from multiple candidate models based on the correlation analysis results and stability verification results; and constructing a packing amount data acquisition model corresponding to the rod type based on the target model, normality test results, and the packing amount data and pressure drop data of sample tobacco filter rods of the rod type. This method can improve the model's adaptability to meet the requirements of actual production for controlling the stability of filter rod pressure drop due to the packing amount.

Description

Technical Field

[0001] This application relates to the field of tobacco filter rod production technology, and in particular to a method, apparatus, equipment, medium and product for determining the filling amount of a tobacco filter rod. Background Technology

[0002] In the tobacco industry, filter rods, as the core component of cigarettes, directly affect product quality control and the smoking experience. Pressure drop is a key indicator for measuring filter rod filtration efficiency and comfort, and it needs to be balanced within a specific range. Currently, the ZL29 filter rod forming machine, as the mainstream production equipment in the industry, shows that fluctuations in the filler amount significantly affect the filter rod's pressure drop characteristics. However, in the application of filter rod pressure drop control, existing technologies typically combine data from different rod types into a single model, using a single rod type in production scenarios. The model output is only a deterministic mean, resulting in low model adaptability and difficulty in meeting the requirements of actual tobacco filter rod filler amount for stable pressure drop control. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for determining the filling amount of a tobacco filter rod, which can improve the adaptability of the model and meet the requirements of the filling amount of the tobacco filter rod for stable control of filter rod pressure drop in actual production.

[0004] In a first aspect, this application provides a method for determining the amount of tobacco filling in a tobacco filter rod, including:

[0005] Obtain packing amount data and pressure drop data for sample tobacco filter rods of at least two different rod types;

[0006] For the packing amount data and pressure drop data of any of the sample tobacco filter rods of the aforementioned rod type, correlation analysis, normality test processing and stability verification processing are performed on the packing amount data and pressure drop data respectively to obtain the correlation analysis results, normality test results and stability verification results;

[0007] Based on the correlation analysis results and the stability verification results, the target model corresponding to the bar shape is obtained from multiple types of candidate models;

[0008] Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the specified rod type, a packing amount data acquisition model corresponding to the specified rod type is constructed; the packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the specified rod type.

[0009] In one embodiment, obtaining the target model corresponding to the bar shape from multiple types of candidate models based on the correlation analysis results and the stability verification results includes:

[0010] Each candidate model is subjected to parameter estimation and goodness-of-fit evaluation to obtain multiple parameter estimation results and multiple goodness-of-fit evaluation results.

[0011] Based on the correlation analysis results, stability verification results, parameter estimation results, and fit evaluation results, the target model corresponding to the bar shape is obtained from the multiple candidate models.

[0012] In one embodiment, determining the amount of tobacco filling in the tobacco filter rod further includes:

[0013] The least squares method is used to perform parameter estimation on each candidate model to obtain the parameter estimation results of each candidate model;

[0014] Based on the parameter estimation results, correlation analysis results, and preset parameter thresholds, a target candidate model is determined from multiple candidate models.

[0015] The goodness-of-fit of each of the target candidate models is evaluated to obtain the goodness-of-fit evaluation results of each of the target candidate models;

[0016] Based on the correlation analysis results, the stability verification results, and the fit evaluation results, the target model corresponding to the bar shape is obtained from multiple target candidate models.

[0017] In one embodiment, the step of constructing a data acquisition model for the filler quantity of the filter rod corresponding to the rod type, based on the target model, the normality test results, and the filler quantity data and pressure drop data of the sample tobacco filter rod of the rod type, includes:

[0018] Based on the normality test results, determine the error terms of each target model;

[0019] Based on the target model, the error term, and the packing amount data and pressure drop data of the sample tobacco filter rod of the rod type, a packing amount data acquisition model corresponding to the rod type is constructed.

[0020] In one embodiment, after constructing the filler amount data acquisition model corresponding to the rod type based on the target model, the normality test results, and the filler amount data and pressure drop data of the sample tobacco filter rod of the rod type, the method further includes:

[0021] Obtain the target pressure drop data corresponding to each of the aforementioned rod types;

[0022] Based on the filler wire amount data corresponding to each of the rod types, the model and the target pressure drop data are obtained to determine the target filler wire amount data corresponding to each of the rod types.

[0023] In one embodiment, the packing amount data and the pressure drop data for any sample tobacco filter rod of the aforementioned rod type are subjected to a correlation analysis to obtain the correlation analysis results, including:

[0024] For the packing amount data and the pressure drop data of any sample tobacco filter rod of the aforementioned rod type, a first image model is constructed between the packing amount data and the pressure drop data;

[0025] The correlation analysis results are obtained based on the first image model.

[0026] In one embodiment, the packing amount data and pressure drop data for any of the aforementioned rod types are subjected to stability verification processing to obtain stability verification results, including:

[0027] For the packing amount data and pressure drop data of sample tobacco filter rods of any of the aforementioned rod types, multiple coefficients of variation for each of the aforementioned rod types are obtained;

[0028] Based on the coefficient of variation and the filler amount data, a second image model is constructed between the filler amount and the coefficient of variation;

[0029] The stability verification result is obtained based on the second image model and the preset fluctuation threshold.

[0030] Secondly, this application also provides a device for determining the amount of tobacco filter material in a tobacco filter rod, comprising:

[0031] The data acquisition module is used to acquire data on the amount of tobacco filling and pressure drop of sample tobacco filter rods of at least two different rod types;

[0032] The data processing module is used to perform correlation analysis, normality test processing, and stability verification processing on the packing amount data and pressure drop data of any sample tobacco filter rod of the aforementioned rod type, respectively, to obtain the correlation analysis results, normality test results, and stability verification results.

[0033] The model filtering module is used to obtain the target model corresponding to the bar shape from multiple types of candidate models based on the correlation analysis results and the stability verification results.

[0034] The model building module is used to construct a data acquisition model for the filling amount of the sample tobacco filter rod of the rod type based on the target model, the normality test results, and the filling amount data and pressure drop data of the sample tobacco filter rod of the rod type; the data acquisition model for the filling amount of the sample tobacco filter rod of the rod type is used to characterize the correlation between the filling amount data and the pressure drop data of the sample tobacco filter rod of the rod type.

[0035] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0036] Obtain packing amount data and pressure drop data for sample tobacco filter rods of at least two different rod types;

[0037] For the packing amount data and pressure drop data of any of the sample tobacco filter rods of the aforementioned rod type, correlation analysis, normality test processing and stability verification processing are performed on the packing amount data and pressure drop data respectively to obtain the correlation analysis results, normality test results and stability verification results;

[0038] Based on the correlation analysis results and the stability verification results, the target model corresponding to the bar shape is obtained from multiple types of candidate models;

[0039] Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the specified rod type, a packing amount data acquisition model corresponding to the specified rod type is constructed; the packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the specified rod type.

[0040] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0041] Obtain packing amount data and pressure drop data for sample tobacco filter rods of at least two different rod types;

[0042] For the packing amount data and pressure drop data of any of the sample tobacco filter rods of the aforementioned rod type, correlation analysis, normality test processing and stability verification processing are performed on the packing amount data and pressure drop data respectively to obtain the correlation analysis results, normality test results and stability verification results;

[0043] Based on the correlation analysis results and the stability verification results, the target model corresponding to the bar shape is obtained from multiple types of candidate models;

[0044] Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the specified rod type, a packing amount data acquisition model corresponding to the specified rod type is constructed; the packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the specified rod type.

[0045] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0046] Obtain packing amount data and pressure drop data for sample tobacco filter rods of at least two different rod types;

[0047] For the packing amount data and pressure drop data of any of the sample tobacco filter rods of the aforementioned rod type, correlation analysis, normality test processing and stability verification processing are performed on the packing amount data and pressure drop data respectively to obtain the correlation analysis results, normality test results and stability verification results;

[0048] Based on the correlation analysis results and the stability verification results, the target model corresponding to the bar shape is obtained from multiple types of candidate models;

[0049] Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the specified rod type, a packing amount data acquisition model corresponding to the specified rod type is constructed; the packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the specified rod type.

[0050] The aforementioned method, apparatus, computer equipment, computer-readable storage medium, and computer program product for determining the packing amount of tobacco filter rods involve acquiring packing amount data and pressure drop data of sample tobacco filter rods of at least two different rod types. For the packing amount data and pressure drop data of any given rod type, correlation analysis, normality test processing, and stability verification processing are performed on the packing amount data and pressure drop data to obtain correlation analysis results, normality test results, and stability verification results. Then, based on the correlation analysis results and stability verification results, a target model corresponding to the rod type is obtained from multiple types of candidate models. Finally, based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the given rod type, a packing amount data acquisition model corresponding to the rod type is constructed. This packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the rod type. This method enables separate modeling of the different correlation characteristics of tobacco filter rods with different rod types due to the different filament filling morphology. This avoids the parameter deviation caused by merging modeling, thereby improving the adaptability of the model and providing more comprehensive reference information for production control, so as to meet the requirements of the filling amount of tobacco filter rods in actual production for the stability control of filter rod pressure drop. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating a method for determining the amount of tobacco filling in a tobacco filter rod in one embodiment.

[0053] Figure 2 This is a schematic diagram illustrating the fitting effect of the filler wire amount and pressure drop on the rod in one embodiment;

[0054] Figure 3 This is a schematic diagram illustrating the fitting effect of filler wire amount and pressure drop in one embodiment;

[0055] Figure 4 This is a schematic diagram illustrating the fitting effect of the filler wire amount and pressure drop on the large bar in one embodiment;

[0056] Figure 5 This is a structural block diagram of a device for determining the amount of tobacco filter material in one embodiment;

[0057] Figure 6This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0059] As described in the background section, the pressure drop control methods for tobacco filter rods in related technologies suffer from low model adaptability, making it difficult to meet the requirements of actual production regarding the stability of filter rod pressure drop control based on the amount of filler. The inventors have discovered that this problem arises because, in the tobacco industry, the filter rod, as a core component of cigarettes, directly impacts product quality control and the consumer's smoking experience due to its pressure drop performance. Pressure drop is a key indicator for measuring filter rod filtration efficiency and comfort, requiring a balance within a specific range. Excessive pressure drop leads to excessive draw resistance, affecting the user experience; insufficient pressure drop may reduce the efficiency of harmful substance retention, while excessive smoke volume also affects taste. Filter rod pressure drop also impacts the cigarette production process. Excessive pressure drop may increase equipment energy consumption and reduce production efficiency; unstable pressure drop may cause equipment malfunctions during production, increasing maintenance costs. Therefore, the accuracy of filter rod pressure drop control directly affects the cigarette product qualification rate, production efficiency, and the economic benefits of the tobacco industry. Currently, the ZL29 filter rod forming machine, as the mainstream production equipment in the industry, shows that fluctuations in the filler amount significantly affect the filter rod pressure drop characteristics. However, existing technologies for filter rod pressure drop control still have the following technical shortcomings: First, they ignore differences in rod type and merge data from different rod types into a single model, leading to a decrease in the applicability of the model in a single rod type production scenario. Second, they focus on fitting a single model without systematically comparing the applicability of multiple models, and do not consider the impact of random fluctuations in filler quantity caused by uneven filament bundles during production on pressure drop. The model output is only a deterministic mean, which cannot provide a reference for the fluctuation range and is difficult to meet the requirements for stable control of filter rod pressure drop in actual production. In addition, related technologies have proposed gradually reducing the filament bundle feed rate until the filter rod shows a 1mm shrinkage, obtaining a filter rod sample and testing the suction resistance and filament bundle filler quantity; then gradually increasing the filament bundle feed rate until the roller wraps the filament bundle, repeating the above operation, and finally establishing a regression equation between filament bundle filler quantity and suction resistance under standard circumference based on the obtained data. However, the regression relationship between filler quantity and pressure drop established by this method is relatively simple, and the consideration of the influence of multiple parameters is slightly insufficient. For example, related technologies have also clarified the correlation between tow specifications, filler quantity, and pressure drop by testing the forming ability and filter rod performance of different specifications of cellulose acetate tows. This provides a quantitative basis for the selection of tow specifications for medium and fine cigarette filters. The method clarifies the range of cellulose acetate tow specifications suitable for medium and fine cigarettes and establishes a quantitative relationship model between filler quantity and pressure drop within this specification range. However, it does not include the fluctuations in pressure drop data that exist in actual production in the model, which will have a certain impact on the predictive ability.

[0060] For the reasons mentioned above, this application provides a method for determining the amount of tobacco filling in a tobacco filter rod, which aims to improve adaptability and meet the requirements of the amount of tobacco filling in the filter rod for stable control of filter rod pressure drop in actual production.

[0061] In one embodiment, such as Figure 1As shown, a task scheduling method for multiple Jenkins instances is provided. This embodiment illustrates the method by applying it to a server. It is understood that this method can also be applied to terminals, and to systems including terminals and servers, and is implemented through interaction between the terminals and servers. The terminals can be, but are not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, projection devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. In this embodiment, the method includes the following steps:

[0062] Step S102: Obtain the packing amount data and pressure drop data of at least two different types of sample tobacco filter rods.

[0063] Here, "rod type" can refer to different specifications or models of tobacco filter rods. In the production of ZL29 filter rod forming machines, different rod types may correspond to different lengths, circumferences, filament specifications, or design target parameters. For example, based on the filament filling morphology, they can be divided into three categories: small rods (filament end face is lower than the forming paper, with a constricted head), medium rods (no constricted head / protruding head, normal), and large rods (filament end face is higher than the forming paper, with a protruding head). Different rod types have different physical differences in filament filling morphology (end face height, density distribution), resulting in different correlation characteristics between filament filling amount and pressure drop.

[0064] Among them, the sample tobacco filter rods can refer to representative filter rod samples collected from the actual production process for building and validating the model. For example, 100 samples can be collected independently for each type of filter rod to obtain data on the amount of tobacco filling and the pressure drop.

[0065] Among them, the filling amount data can refer to the mass or volume of the filament bundle (filter material) filled in a unit length or a single tobacco filter rod, which is a key process parameter affecting the pressure drop of the filter rod.

[0066] Pressure drop data refers to the pressure loss value generated when airflow passes through the tobacco filter rod, which is used to measure the filter rod's filtration efficiency and suction resistance.

[0067] Optionally, the server acquires data on the filling amount and pressure drop of sample tobacco filter rods of at least two different rod types. For example, it acquires data on the filling amount and pressure drop of sample tobacco filter rods of three rod types: large, medium, and small. Specifically, during data acquisition, the 3.8Y / 34000 specification acetate fiber tow filter rod produced by the ZL29 filter rod forming machine is used as the object. During data acquisition, the forming machine parameters are uniformly set (such as forming speed 400 rods / minute, filter rod circumference 24.0mm, opening ratio 1:1.2, and roller pressure 0.3MPa), and the ambient temperature and humidity are controlled (such as 22±2℃, 55±5%) to eliminate interference. It is understandable that the correlation characteristics between filling amount and pressure drop are different for the same rod type due to physical differences in the tow filling shape (end face height, density distribution). Collecting data by rod type can retain the specific data of each rod type, avoid parameter deviations caused by merging modeling, and lay a data foundation for subsequent accurate modeling.

[0068] Optionally, after the server acquires the packing amount and pressure drop data of at least two different types of sample tobacco filter rods, the server can preprocess the collected packing amount and pressure drop data to improve data quality. For example, it can use box plots to remove outliers, calculate the first quartile (Q1) and third quartile (Q3) of each data set, and use 1.5×(Q3-Q1) as a threshold to remove data above Q3+1.5×(Q3-Q1) or below Q1-1.5×(Q3-Q1), ultimately forming a data table corresponding to the packing amount and pressure drop data for each filter rod type, which facilitates subsequent processing. It is understood that uneven fiber bundles and instantaneous equipment vibrations during production may generate abnormal data (such as sudden increases / decreases in packing amount). Such data can interfere with the model fitting accuracy. Identifying outliers through statistical methods can ensure the consistency and validity of the data and reduce systematic errors in subsequent modeling.

[0069] Step S104: For the packing amount data and pressure drop data of any type of sample tobacco filter rod, perform correlation analysis, normality test processing and stability verification processing on the packing amount data and pressure drop data respectively, and obtain the correlation analysis results, normality test results and stability verification results.

[0070] Correlation analysis can refer to the process of determining whether there is a linear or nonlinear relationship between filler quantity data and pressure drop data through statistical methods (such as Pearson correlation coefficient, scatter plot observation, etc.); the results of correlation analysis can be the analytical conclusions obtained after correlation analysis, confirming the correlation between filler quantity and pressure drop.

[0071] Among them, the normality test can refer to testing the residual distribution of the filler amount data or pressure drop data to determine whether the data conforms to a normal distribution, thereby determining the error term processing method of the subsequent model; the normality test result can be the conclusion obtained after performing normality test processing on the filler amount data and pressure drop data.

[0072] Among them, stability verification processing can refer to the process of evaluating the fluctuation of filler quantity data and pressure drop data during the production process and determining whether they are under control; stability verification result can be the evaluation result after stability verification of filler quantity data and pressure drop data.

[0073] Optionally, the server performs correlation analysis on the filler quantity and pressure drop data of any type of sample tobacco filter rod, determining the trend of change between filler quantity and pressure drop, whether there is a correlation between the two and the direction of the correlation, and obtaining the correlation analysis results; it also performs normality test processing on the filler quantity and pressure drop data to test whether the pressure drop data follows a normal distribution, obtaining the normality test results; and it performs stability verification processing on the filler quantity and pressure drop data, obtaining the stability verification results. The above analyses yield correlation analysis results, normality test results, and stability verification results, respectively, providing a basis for subsequent model selection.

[0074] Step S106: Based on the correlation analysis results and stability verification results, obtain the target model corresponding to the bar shape from multiple types of candidate models.

[0075] Candidate models can refer to various types of mathematical models that are considered as alternatives in the model selection process, such as, but not limited to, linear regression models, cubic curve regression models, logarithmic curve regression models, and composite curve regression models.

[0076] The target model can refer to the optimal model type that is finally determined after screening multiple models to build a model for obtaining the filling amount data.

[0077] Optionally, the server can compare and evaluate various candidate models of multiple types based on the correlation analysis results and stability verification results, and comprehensively select the target model that best suits the current rod type's filler wire data characteristics and pressure drop data characteristics.

[0078] Step S108: Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the rod type, construct a packing amount data acquisition model corresponding to the rod type.

[0079] Among them, the filler wire data acquisition model is used to characterize the correlation between the filler wire data corresponding to the rod type and the pressure drop data corresponding to the rod type.

[0080] Optionally, based on the selected target model, the server introduces an error term by combining the normality test results to construct an initial filler quantity data acquisition model. Then, based on the filler quantity data and pressure drop data of the sample tobacco filter rods of the rod type, the server solves for the unknown parameters in the initial filler quantity data acquisition model to obtain the final filler quantity data acquisition model corresponding to the rod type. This model is then used in subsequent production processes to deduce the required filler quantity for various rod types based on the required pressure drop.

[0081] In the aforementioned method for determining the filler amount of tobacco filter rods, the method acquires filler amount data and pressure drop data of sample tobacco filter rods of at least two different rod types. For the filler amount data and pressure drop data of any rod type, correlation analysis, normality test processing, and stability verification processing are performed on the filler amount data and pressure drop data, respectively, to obtain the correlation analysis results, normality test results, and stability verification results. Then, based on the correlation analysis results and stability verification results, the target model corresponding to the rod type is obtained from multiple types of candidate models. Finally, based on the target model, the normality test results, and the filler amount data and pressure drop data of the sample tobacco filter rods of the rod type, a filler amount data acquisition model corresponding to the rod type is constructed. This filler amount data acquisition model is used to characterize the correlation between the filler amount data and the pressure drop data corresponding to the rod type. This method enables separate modeling of the different correlation characteristics of tobacco filter rods with different rod types due to the different filament filling morphology. This avoids the parameter deviation caused by merging modeling, thereby improving the adaptability of the model and providing more comprehensive reference information for production control, so as to meet the requirements of the filling amount of tobacco filter rods in actual production for the stability control of filter rod pressure drop.

[0082] In an exemplary embodiment, step S106, based on the correlation analysis results and stability verification results, obtains the target model corresponding to the bar shape from multiple types of candidate models, including:

[0083] Each candidate model is subjected to parameter estimation and goodness-of-fit evaluation, resulting in multiple parameter estimation results and multiple goodness-of-fit evaluation results. Based on the correlation analysis results, stability verification results, parameter estimation results, and goodness-of-fit evaluation results, the target model corresponding to the bar shape is obtained from the multiple candidate models.

[0084] Among them, parameter estimation processing can refer to the process of calculating model parameters (such as the slope and intercept of a linear model) through methods such as least squares, with the aim of enabling the model to best fit the observed data.

[0085] Among them, the goodness-of-fit evaluation can refer to the process of evaluating the accuracy and explanatory power of the established model. In this embodiment, the goodness-of-fit evaluation indicators include the coefficient of determination R² (which measures the proportion of variation of the explanatory variables in the model) and the mean absolute percentage error (which measures the prediction accuracy).

[0086] The fit evaluation result can refer to the quantitative index value obtained through the fit evaluation, such as the R² value and MAPE value. These indicators are used to compare the fit effect of different candidate models and help to select the target model.

[0087] Optionally, candidate models may include linear regression models, cubic curve regression models, logarithmic curve regression models, composite curve regression models, etc. The server performs parameter estimation and goodness-of-fit evaluation on each candidate model, obtaining multiple parameter estimation results and multiple goodness-of-fit evaluation results. Then, based on the correlation analysis results (such as data showing a linear trend) and stability verification results (such as uniform residual distribution), the optimal target model is selected through significance testing and goodness-of-fit evaluation. For example, the parameters of the candidate models are estimated, models with excessively high coefficient significance levels are eliminated, and the coefficient of determination R² and mean absolute percentage error (MAPE) of each model are calculated to evaluate the goodness of fit, thereby determining the target model.

[0088] In this embodiment, a significance test is performed through parameter estimation, and a dual screening mechanism of goodness-of-fit evaluation is combined to select the optimal model from multiple candidate models. The significance test eliminates statistically insignificant models, thus avoiding overfitting. The goodness-of-fit evaluation quantifies the model's prediction accuracy. The final selected target model ensures prediction accuracy while having good operability and extrapolation stability, solving the problems of single model and poor adaptability in the prior art.

[0089] In one exemplary embodiment, the method for determining the filler amount of the tobacco filter rod described in the above embodiments further includes:

[0090] The least squares method is used to estimate the parameters of each candidate model, and the parameter estimation results of each candidate model are obtained. Based on the parameter estimation results, the correlation analysis results, and the preset parameter thresholds, the target candidate model is determined from multiple candidate models. The goodness of fit of each target candidate model is evaluated, and the goodness of fit evaluation results of each target candidate model are obtained. Based on the correlation analysis results, the stability verification results, and the goodness of fit evaluation results, the target model corresponding to the bar shape is obtained from multiple target candidate models.

[0091] Optionally, the server uses the least squares method to perform parameter estimation on each candidate model, obtaining parameter estimation results for each candidate model. Specifically, for linear regression, cubic regression, logarithmic regression, and composite regression models among the candidate models, parameter estimates are calculated respectively. Then, based on the parameter estimation results, correlation analysis results, and preset parameter thresholds, the server determines the target candidate model from multiple candidate models. The preset parameter threshold can be a significance level threshold (e.g., 0.05). The server determines the significance level (p-value) of each parameter estimate in each candidate model. If the p-value of a parameter in a candidate model is greater than 0.05, it indicates that the explanatory power of that parameter is not significant, and the candidate model is eliminated, thus obtaining the remaining target candidate models. For example, in a cubic regression model, if the p-value of the coefficient of the X³ term is >0.05, it indicates that the cubic term has no significant contribution, and the candidate model is unsuitable. Further, the server evaluates the fit of each target candidate model, obtaining the fit evaluation results for each target candidate model. Specifically, for candidate models that pass the significance test, i.e. target candidate models, the server calculates their coefficient of determination R² and mean absolute percentage error (MAPE). The closer R² is to 1 and the smaller the MAPE, the better the target candidate model fits. Then, the server comprehensively considers the correlation analysis results, stability verification results, and various fit evaluation results to obtain the target model corresponding to the bar from multiple target candidate models. For example, although the R² of the logarithmic model may be slightly higher, the linear model has the advantages of simple form, intuitive parameter interpretation, and strong extrapolation stability, and is more suitable for the actual needs of industrial production scenarios. Therefore, the linear model can be finally determined as the target model.

[0092] In this embodiment, the reliability and applicability of the selected model are ensured through a systematic model screening process. The significance test eliminates complex models with insignificant parameter interpretation, thus avoiding the risk of overfitting. The goodness-of-fit evaluation quantifies the model's predictive accuracy. By comprehensively considering the model's simplicity and interpretability, the final selected target model is not only highly accurate but also easy to understand and apply in actual production, solving the problems of single models and poor adaptability in the prior art.

[0093] In an exemplary embodiment, step S108, which involves constructing the filler data acquisition model content corresponding to the rod type based on the target model, the normality test results, and the filler amount data and pressure drop data of the sample tobacco filter rod, includes:

[0094] Based on the results of each normality test, the error terms of each target model are determined; based on the target model, the error terms, and the packing amount data and pressure drop data of the sample tobacco filter rods of the rod type, a data acquisition model for the packing amount data corresponding to the rod type is constructed.

[0095] The error term can refer to a variable in the model that represents data fluctuations caused by random factors.

[0096] The packing quantity data acquisition model refers to the final mathematical model constructed to characterize the relationship between packing quantity data and pressure drop data for different types of filter rods. This model can infer the required target packing quantity based on a given target pressure drop and output the mean and confidence interval of the pressure drop prediction.

[0097] Optionally, when the server performs normality verification on the filler material and pressure drop data of the sample tobacco filter rods, it has already concluded that the pressure drop data for each rod type follows a normal distribution. For example, by plotting a histogram of the pressure drop data corresponding to the filler material data for each rod type, it can be observed that the data exhibits a symmetrical distribution with dense data in the middle and sparse data at both ends; by plotting a normal Q–Q plot (Quantile–Quantile plot), it can be observed that the measured quantiles and the theoretical normal quantiles are basically in line with the reference line, and the normality test results verify the statistical law that the pressure drop data follows a normal distribution. Based on this, the server determines that the error term ε of the target model follows a normal distribution with a mean of 0, i.e., ε ~ N(0, σ²). Here, σ is the standard deviation, reflecting the random fluctuation amplitude of pressure drop caused by factors such as uneven fiber bundle, minor equipment vibration, and sensor errors during the production process. Furthermore, the server constructs a data acquisition model for the filling amount of the tobacco filter rod corresponding to the rod type based on the target model, error term, and sample tobacco filter rod filling amount and pressure drop data. For example, taking the linear model Y=β0+β1·X as an example, a Gaussian error term ε is introduced to obtain a Gaussian linear model, i.e., the filling amount data acquisition model is Y=β0+β1·X+ε, ε~N(0,σ²), where Y is the pressure drop data, X is the filling amount data, β1 is the slope, and β0 is the intercept. Using the Gaussian linear model, two aspects of information can be output simultaneously, and the pressure drop prediction mean is... The predicted pressure drop range is: [Ŷ-1.96σ ， Ŷ+1.96σ] This interval corresponds to a 95% confidence level, meaning that filter rods produced with the same filler content have a 95% probability that their measured pressure drop will fall within this interval. The principle for constructing the confidence interval can be found in [reference needed]. Figure 2 , Figure 3 and Figure 4 The diagrams provided illustrate the fitting effects of filler wire amount and pressure drop for the small rod, the medium rod, and the large rod, respectively.

[0098] In this embodiment, by introducing an error term, the deterministic model is extended to a probabilistic model, which quantifies the unavoidable random fluctuations in the production process. This allows the model output to reflect the distribution characteristics of the actual data, and the confidence interval provides a clear fluctuation boundary for quality monitoring. In production, the amount of filler can be adjusted to ensure that the entire interval falls within the acceptable range, thereby effectively controlling the fluctuation amplitude of pressure drop and improving the stability of filter rod quality. This solves the problem of unquantified random fluctuations and lack of stability in the prior art.

[0099] In an exemplary embodiment, after constructing the filler amount data acquisition model corresponding to the rod type based on the target model, the normality test results, and the filler amount data and pressure drop data of the sample tobacco filter rod, step S108 further includes:

[0100] Obtain the target pressure drop data for each rod type; based on the filler wire amount data for each rod type, obtain the model and target pressure drop data, and determine the target filler wire amount data for each rod type.

[0101] The target pressure drop data can be the pressure drop value required by the actual production needs for different types of filter rods, or the ideal pressure drop value set according to product design requirements and quality control standards. For example, for a certain specification of cigarette, the target pressure drop during the production of small rods may be 2050Pa, medium rods 3050Pa, and large rods 3500Pa.

[0102] Optionally, the server obtains the target pressure drop data corresponding to each rod type, and then inputs the target pressure drop data corresponding to each rod type into the filler wire amount data acquisition model for each rod type to determine the target filler wire amount data for each rod type. For example, taking the medium rod model as an example, from Y=6497.358X-1250.691, let Y=3000, we get X_target=(3000+1250.691) / 6497.358≈0.654g / cm³, where the target pressure drop Y is substituted into the filler wire amount data acquisition model for the corresponding rod type to solve for the target filler wire amount X_target.

[0103] Meanwhile, based on the standard deviation σ=31.22Pa of the filler data model, the 95% confidence interval of the pressure drop under this filler amount can be calculated as [3000-1.96×31.22, 3000+1.96×31.22]=[2938.8, 3061.2]Pa. During the production process, the server can check whether the pressure drop of the filter rod produced in real time falls within this interval. If it exceeds the interval, it will issue an early warning and make adjustments in time.

[0104] In this embodiment, by applying the filler quantity data acquisition model to actual production control, the reverse derivation from the target pressure drop to the target filler quantity is realized, providing a quantitative control basis for the accurate realization of the target pressure drop and avoiding the blindness of relying on experience for adjustment; combined with process monitoring of the prediction interval, production anomalies can be detected in a timely manner, preventing the occurrence of quality problems and improving the quality control level of filter rod production.

[0105] In an exemplary embodiment, step S104 involves performing a correlation analysis on the filler material data and pressure drop data for any type of sample tobacco filter rod, yielding the correlation analysis results, including:

[0106] For sample tobacco filter rods of any rod type, a first image model is constructed to connect the filler amount data and pressure drop data; based on the first image model, the correlation analysis results are obtained.

[0107] The first image model can refer to the graphical representation used for correlation analysis, specifically a scatter plot between the filler data and the pressure drop data. The correlation trend between variables can be observed intuitively through the first image model.

[0108] Optionally, for the packing amount data and pressure drop data of any type of sample tobacco filter rod, a first image model is constructed between the packing amount data and the pressure drop data. The correlation analysis results are obtained based on the first image model. The first image model can be a scatter plot. For example, a scatter plot is drawn using 100 pairs of packing amount data and pressure drop data as coordinate points. By observing the distribution trend of the scatter points, it is possible to intuitively determine whether there is a correlation between the packing amount data and the pressure drop data and the direction of the correlation. If the pressure drop increases synchronously with the increase of the packing amount and the scatter points show a band distribution from the lower left to the upper right, it can be determined that the two are significantly positively correlated.

[0109] The inventors discovered that, in practical implementation, scatter plots were drawn for the three types of rods—small, medium, and large—and a comparison revealed that, although all three types of rods showed a positive correlation trend, their slopes (i.e., the degree of influence of changes in the amount of filler on the pressure drop) differed significantly. This further confirms the necessity of modeling by rod type.

[0110] In this embodiment, correlation analysis is achieved through an image model. The scatter plot intuitively displays the correlation trend between variables, providing a visual basis for subsequent model type selection. At the same time, by comparing scatter plots of different rod types, the influence of rod type differences on the relationship between filler wire amount data and pressure drop data can be intuitively understood, strengthening the rationality of rod type-based modeling.

[0111] In an exemplary embodiment, step S104 involves performing stability verification on the packing amount data and pressure drop data for any type of sample tobacco filter rod, resulting in the stability verification result, which includes:

[0112] For sample tobacco filter rods of any rod type, the packing amount data and pressure drop data are used to obtain multiple coefficients of variation for each rod type; based on the coefficients of variation and packing amount data, a second image model is constructed between the packing amount and the coefficients of variation; based on the second image model and a preset fluctuation threshold, the stability verification results are obtained.

[0113] The coefficient of variation can be the ratio of the standard deviation to the mean, which is used to measure the relative dispersion of the data. In this embodiment, the repeatability and volatility of the data are evaluated by calculating the coefficient of variation of the pressure drop data under different filler wire levels.

[0114] The second image model can refer to the graphical representation used for stability verification, specifically the error scatter plot between the filler amount and the coefficient of variation. The distribution characteristics and fluctuation range of the model residuals can be observed intuitively through the second image model.

[0115] The fluctuation threshold can be a preset critical value used to judge the stability of data. For example, in this embodiment, the preset fluctuation threshold is 5%. If the coefficient of variation at each filler amount level is less than 5%, the data stability is judged to meet the modeling requirements.

[0116] Optionally, the server obtains multiple coefficients of variation for each type of tobacco filter rod based on the filler weight and pressure drop data. Specifically, the coefficient of variation is the ratio of the standard deviation to the mean, used to measure the relative dispersion of the data. For each filler weight data level (or filler weight interval), the server calculates the coefficient of variation for the pressure drop data of multiple sample tobacco filter rods at that level. Then, with the filler weight data as the x-axis and the coefficient of variation as the y-axis, an error scatter plot is plotted to obtain the second image model. The preset fluctuation threshold can be 5%. If the server detects that the coefficient of variation at each filler weight data level is less than 5%, and the error bars (representing confidence intervals) in the second image model are evenly distributed without a significant trend of change with the filler weight, it indicates that the data repeatability is good, the model fitting residuals are evenly distributed, and the model has universality across the entire filler weight range.

[0117] In this embodiment, by calculating the coefficient of variation of each rod type and constructing an error scatter plot, i.e., the second image model, stability verification is performed, and the repeatability and volatility of the data are quantitatively evaluated, ensuring that the data foundation on which the subsequent filling amount data acquisition model is based is stable and reliable. The error scatter plot intuitively shows the distribution characteristics of the model residuals, which can be used to verify whether the model assumptions (such as homogeneity of variance) are valid, providing support for the effectiveness of the subsequent filling amount data acquisition model.

[0118] In one exemplary embodiment, another method for determining the packing amount of a tobacco filter rod is provided, specifically including:

[0119] Step 1: Collect data in a bar-like pattern.

[0120] For example, taking 3.8Y / 34000 specification cellulose acetate tow filter rods produced by the ZL29 filter rod forming machine as the object, three types of rods were clearly defined: "small rods (the end face of the tow is lower than the forming paper, with a shrinkage), medium rods (no shrinkage / emergence, normal), and large rods (the end face of the tow is higher than the forming paper, with an emergence)". 100 rods of each type were independently sampled. During data collection, the forming machine parameters were uniformly set (forming speed 400 rods / minute, filter rod circumference 24.0 mm, opening ratio 1:1.2, roller pressure 0.3 MPa), and the ambient temperature and humidity were controlled (22±2℃, 55±5%) to eliminate interference. Principle: Different rod types have physical differences in the tow filling morphology (end face height, density distribution), resulting in different correlation characteristics between the filling amount and pressure drop. Segmented data collection by rod type retains the specific data of each rod type, avoiding parameter deviations caused by merging modeling, and laying a data foundation for subsequent accurate modeling.

[0121] Step 2, data preprocessing.

[0122] For example, outliers are removed using box plots. The first quartile (Q1) and third quartile (Q3) of each data set are calculated. Using 1.5 × (Q3 - Q1) as a threshold, data above Q3 + 1.5 × (Q3 - Q1) or below Q1 - 1.5 × (Q3 - Q1) are removed, ultimately forming a bar-shaped data table of filler quantity and pressure drop. Principle: Uneven filament bundles and instantaneous equipment vibrations during production may generate abnormal data (such as sudden increases / decreases in filler quantity). Such data can interfere with the model's fitting accuracy. Identifying outliers through statistical methods ensures data consistency and validity, reducing systematic errors in subsequent modeling.

[0123] Step 3, Statistical Analysis of Data.

[0124] For example, a scatter plot of filler amount versus pressure drop was plotted, revealing a significant positive correlation between the two (pressure drop increases synchronously with increasing filler amount), clarifying the direction of variable association and providing a basis for model type selection. Normality test: Histograms (data exhibiting a symmetrical distribution of "dense in the middle and sparse at both ends") and normal QQ plots (measured quantiles and theoretical normal quantiles closely approximate the reference line) were plotted for pressure drop data corresponding to typical filler amounts, verifying that the pressure drop data follows a normal distribution under a fixed filler amount, providing a statistical basis for subsequently introducing a Gaussian error term. Stability verification: An error scatter plot containing a 95% confidence interval was plotted using the mean pressure drop as a baseline, revealing that the coefficient of variation for each filler amount across the three types of rods fluctuated by less than 5%, demonstrating good data repeatability and a uniform distribution of model fitting residuals, ensuring the model's universality across the entire filler amount range.

[0125] Step 4: Multi-model screening and determination of the optimal model.

[0126] For example, based on the linear trend characteristics of the data, four candidate models—linear regression, cubic regression, logarithmic regression, and composite regression—were selected. The least squares method (core principle: minimizing the sum of squared residuals between observed and predicted values ​​to achieve optimal model fit) was used for bar-shaped parameter estimation. Models were screened through "significance test + goodness-of-fit evaluation": Significance test: The coefficients of the cubic regression model had a significance level > 0.05, indicating insignificant parameter explanation, and were therefore eliminated. Goodness-of-fit evaluation: The coefficient of determination (R²) and mean absolute percentage error (MAPE) were calculated. Although the logarithmic model had a slightly higher R², the linear model had advantages such as "simple form, intuitive parameter interpretation, and strong extrapolation stability," making it more suitable for industrial production scenarios. Ultimately, the linear model was determined as the basic model.

[0127] Step 5: Constructing the Gaussian linear model.

[0128] For example, a Gaussian error term is introduced into the optimal linear model (based on the normality test results mentioned above, the error follows a normal distribution) to construct a Gaussian linear model. This model can simultaneously output the predicted mean of pressure drop and a 95% confidence interval (e.g., the mean standard deviation of the Gaussian linear model for small rods is 24.96, for medium rods 31.22, and for large rods 25.99). The principle is that uneven fiber bundles and minor equipment vibrations during production can cause random fluctuations in pressure drop data. Existing technologies only output a deterministic mean and cannot provide a reference for the fluctuation range. By quantifying the fluctuations through the Gaussian error term, the 95% confidence interval can cover the vast majority of measured values, providing a dual basis of "mean + fluctuation boundary" for adjusting the filler amount during production, thus avoiding pressure drop exceeding the tolerance due to fluctuations.

[0129] In one embodiment tested and researched by the inventor, the following equipment and materials were prepared: a ZL29 filter rod forming machine, 3.8Y / 34000 cellulose acetate tow, and forming paper. The process parameters were configured as follows: forming speed 400 rods / minute, filter rod circumference 24.0mm, opening ratio 1:1.2, roller pressure 0.3MPa, and ambient temperature and humidity 22±2℃ and 55±5%.

[0130] Implementation steps: Data acquisition (step 1): Produce 100 small rods, 100 medium rods, and 100 large rods respectively, and record the packing amount (g / cm³) and pressure drop (Pa) of each filter rod online.

[0131] Data preprocessing (step 2): Use box plots to remove outliers from each bar dataset.

[0132] Modeling and Optimization (Steps 3-5): Following the process described in the above embodiments, Gaussian linear models are established for the three types of bars respectively, where: Small bar: Y=7743.243X-2310.239, standard deviation σ=24.96Pa; Medium bar: Y=6497.358X-1250.691, standard deviation σ=31.22Pa; Large bar: Y=10456.538X-3955.448, standard deviation σ=25.99Pa. Table 1 provides a comparison table of measured and predicted values ​​of filter rod pressure drop. The three core equations and their corresponding standard deviations are shown. Five samples each of small, medium and large filter rods were extracted from different batches of samples from the same equipment. After being substituted into the above model, the results are shown in Table 1. It can be seen from the table that all the measured values ​​of filter rod pressure drop fall within the predicted range, and the deviation between the measured values ​​and the center of the predicted range is less than 2%. This indicates that the Gaussian linear model has good prediction accuracy and stability and can effectively reflect the quantitative relationship between the filler amount and the pressure drop.

[0133] Table 1

[0134]

[0135] Optionally, during production, the application control sets the target pressure drop to TPa. Substituting T into the Gaussian linear model corresponding to the filter rod type, the target filler quantity X_target is derived, and the filler quantity control device is set near X_target. Simultaneously, the quality monitoring system checks whether the real-time filter rod pressure drop falls within the range of [Y_predicted - 1.96σ, Y_predicted + 1.96σ].

[0136] In this embodiment, the first core defect of the prior art lies in its failure to recognize that the physical differences in the filament filling morphology of small, medium, and large rods lead to completely different correlations between the filling amount and pressure drop. Merging models would forcibly smooth out the specific correlations, causing prediction bias. This embodiment clarifies the core impact and underlying principles of the physical differences in rod types; it collects data by rod type to preserve specific correlations; it selects models and calculates parameters separately for the three rod types, ultimately obtaining differentiated Gaussian linear equations. The determination coefficients R² of the rod type-specific models are all >0.9 (0.907 for small rods, 0.963 for medium rods, and 0.956 for large rods), and the mean absolute percentage error (MAPE) is <0.6%, which is far superior to merged modeling. In independent batch verification, the deviation between the measured pressure drop values ​​and the center of the prediction interval for small, medium, and large rods is all <2%, solving the problem of low accuracy in merged modeling. The second major defect of the prior art is that it only constructs a simple regression model and outputs a deterministic mean. However, factors such as uneven filament bundles, minor equipment vibrations, and sensor errors in production can cause random fluctuations in pressure drop, and the mean alone cannot guide stable production. This embodiment provides a basis for quantification by verifying the statistical regularity of fluctuations, introduces a Gaussian error term to quantify the fluctuation range, and uses multi-model screening to ensure the accuracy of mean prediction. The output 95% confidence interval can accurately cover actual fluctuations. In production, the interval can be made to fall completely within the qualified range by adjusting the filler amount, realizing stable and controllable production guidance. This ensures that the model can accurately adapt to different rod production scenarios, providing a more scientific basis for adjusting the filler amount of the ZL29 filter rod forming machine, thereby improving the stability of filter rod pressure drop control.

[0137] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0138] Based on the same inventive concept, this application also provides a device for determining the amount of tobacco filter material in order to implement the method for determining the amount of tobacco filter material described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the device for determining the amount of tobacco filter material provided below can be found in the limitations of the method for determining the amount of tobacco filter material described above, and will not be repeated here.

[0139] In one exemplary embodiment, such as Figure 5 As shown, a device 500 for determining the amount of tobacco filter material is provided, comprising: a data acquisition module 501, a data processing module 502, a model screening module 503, and a model construction module 504, wherein:

[0140] The data acquisition module 501 is used to acquire the packing amount data and pressure drop data of sample tobacco filter rods of at least two different rod types;

[0141] The data processing module 502 is used to perform correlation analysis, normality test processing and stability verification processing on the filling amount data and pressure drop data of sample tobacco filter rods of any rod type, and obtain the correlation analysis results, normality test results and stability verification results.

[0142] The model screening module 503 is used to obtain the target model corresponding to the bar shape from multiple types of candidate models based on the correlation analysis results and stability verification results.

[0143] The model building module 504 is used to build a data acquisition model for the filling amount of the bar-shaped sample tobacco filter rod based on the target model, the normality test results, and the filling amount data and pressure drop data of the bar-shaped sample tobacco filter rod. The data acquisition model for the filling amount of the bar-shaped sample tobacco filter rod is used to characterize the correlation between the filling amount data and the pressure drop data of the bar-shaped sample tobacco filter rod.

[0144] Furthermore, in one embodiment, the model screening module 503 is also used to perform parameter estimation processing and goodness-of-fit evaluation on each candidate model to obtain multiple parameter estimation results and multiple goodness-of-fit evaluation results; and to obtain the target model corresponding to the bar shape from the multiple candidate models based on the correlation analysis results, stability verification results, parameter estimation results and goodness-of-fit evaluation results.

[0145] Furthermore, in one embodiment, the model screening module 503 is also used to perform parameter estimation processing on each candidate model using the least squares method to obtain the parameter estimation results of each candidate model; determine the target candidate model from multiple candidate models based on the parameter estimation results, the correlation analysis results, and the preset parameter thresholds; evaluate the fit of each target candidate model to obtain the fit evaluation results of each target candidate model; and obtain the target model corresponding to the bar shape from multiple target candidate models based on the correlation analysis results, the stability verification results, and the fit evaluation results.

[0146] Furthermore, in one embodiment, the model building module 504 is also used to determine the error terms of each target model based on the results of each normality test; and to build a data acquisition model for the filling amount of the rod corresponding to the rod type based on the target model, the error terms, and the filling amount data and pressure drop data of the sample tobacco filter rod.

[0147] Furthermore, in one embodiment, the model building module 504 is also used to obtain target pressure drop data corresponding to each rod type; obtain the model and target pressure drop data according to the filler wire amount data corresponding to each rod type, and determine the target filler wire amount data corresponding to each rod type.

[0148] Furthermore, in one embodiment, the data processing module 502 is also used to construct a first image model between the filler amount data and the pressure drop data for sample tobacco filter rods of any rod type; and to obtain the correlation analysis results based on the first image model.

[0149] Furthermore, in one embodiment, the data processing module 502 is also used to obtain multiple coefficients of variation for each type of sample tobacco filter rod based on the filler amount data and pressure drop data; construct a second image model between the filler amount and the coefficient of variation based on the coefficients of variation and the filler amount data; and obtain a stability verification result based on the second image model and a preset fluctuation threshold.

[0150] Each module in the tobacco filter rod filling amount determination device 500 can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0151] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs in the non-volatile storage media to run. The database stores data such as filler weight data, pressure drop data, correlation analysis results, normality test results, and stability verification results. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for determining the filler weight of a tobacco filter rod.

[0152] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0153] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0154] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0155] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0156] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0157] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0158] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for determining the amount of tobacco filling in a tobacco filter rod, characterized in that, The method includes: Obtain packing amount data and pressure drop data for sample tobacco filter rods of at least two different rod types; For the packing amount data and pressure drop data of any of the sample tobacco filter rods of the aforementioned rod type, correlation analysis, normality test processing and stability verification processing are performed on the packing amount data and pressure drop data respectively to obtain the correlation analysis results, normality test results and stability verification results; Based on the correlation analysis results and the stability verification results, the target model corresponding to the bar shape is obtained from multiple types of candidate models; Based on the target model, the normality test results, and the packing amount data and pressure drop data of the sample tobacco filter rods of the specified rod type, a packing amount data acquisition model corresponding to the specified rod type is constructed; the packing amount data acquisition model is used to characterize the correlation between the packing amount data and the pressure drop data corresponding to the specified rod type.

2. The method according to claim 1, characterized in that, The step of obtaining the target model corresponding to the bar shape from multiple types of candidate models based on the correlation analysis results and the stability verification results includes: Each candidate model is subjected to parameter estimation and goodness-of-fit evaluation to obtain multiple parameter estimation results and multiple goodness-of-fit evaluation results. Based on the correlation analysis results, stability verification results, parameter estimation results, and fit evaluation results, the target model corresponding to the bar shape is obtained from the multiple candidate models.

3. The method according to claim 2, characterized in that, The method further includes: The least squares method is used to perform parameter estimation on each candidate model to obtain the parameter estimation results of each candidate model; Based on the parameter estimation results, correlation analysis results, and preset parameter thresholds, a target candidate model is determined from multiple candidate models. The goodness-of-fit of each of the target candidate models is evaluated to obtain the goodness-of-fit evaluation results of each of the target candidate models; Based on the correlation analysis results, the stability verification results, and the fit evaluation results, the target model corresponding to the bar shape is obtained from multiple target candidate models.

4. The method according to claim 1, characterized in that, The step of constructing a data acquisition model for the filling amount of the specified tobacco filter rod, based on the target model, the normality test results, and the filling amount and pressure drop data of the sample tobacco filter rods of the specified rod type, includes: Based on the normality test results, determine the error terms of each target model; Based on the target model, the error term, and the packing amount data and pressure drop data of the sample tobacco filter rod of the rod type, a packing amount data acquisition model corresponding to the rod type is constructed.

5. The method according to claim 1, characterized in that, After constructing the data acquisition model for the filling amount of the specified tobacco filter rod based on the target model, the normality test results, and the filling amount and pressure drop data of the sample tobacco filter rod, the method further includes: Obtain the target pressure drop data corresponding to each of the aforementioned rod types; Based on the filler wire amount data corresponding to each of the rod types, the model and the target pressure drop data are obtained to determine the target filler wire amount data corresponding to each of the rod types.

6. The method according to any one of claims 1 to 5, characterized in that, The data on the packing material amount and the pressure drop for any sample tobacco filter rod of the aforementioned rod type are used to perform a correlation analysis on the packing material amount data and the pressure drop data, and the correlation analysis results are obtained, including: For the packing amount data and the pressure drop data of any sample tobacco filter rod of the aforementioned rod type, a first image model is constructed between the packing amount data and the pressure drop data; The correlation analysis results are obtained based on the first image model.

7. The method according to any one of claims 1 to 5, characterized in that, The packing amount data and pressure drop data for any of the aforementioned rod types of sample tobacco filter rods are subjected to stability verification processing to obtain stability verification results, including: For the packing amount data and pressure drop data of sample tobacco filter rods of any of the aforementioned rod types, multiple coefficients of variation for each of the aforementioned rod types are obtained; Based on the coefficient of variation and the filler amount data, a second image model is constructed between the filler amount and the coefficient of variation; The stability verification result is obtained based on the second image model and the preset fluctuation threshold.

8. A device for determining the amount of tobacco filling in a tobacco filter rod, characterized in that, The device includes: The data acquisition module is used to acquire data on the amount of tobacco filling and pressure drop of sample tobacco filter rods of at least two different rod types; The data processing module is used to perform correlation analysis, normality test processing, and stability verification processing on the packing amount data and pressure drop data of any sample tobacco filter rod of the aforementioned rod type, respectively, to obtain the correlation analysis results, normality test results, and stability verification results. The model filtering module is used to obtain the target model corresponding to the bar shape from multiple types of candidate models based on the correlation analysis results and the stability verification results. The model building module is used to construct a data acquisition model for the filling amount of the sample tobacco filter rod of the rod type based on the target model, the normality test results, and the filling amount data and pressure drop data of the sample tobacco filter rod of the rod type; the data acquisition model for the filling amount of the sample tobacco filter rod of the rod type is used to characterize the correlation between the filling amount data and the pressure drop data of the sample tobacco filter rod of the rod type.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images