阅读说明：本技术 一种预测烟支燃烧锥落头的方法及其应用 (Method for predicting falling head of cigarette combustion cone and application thereof ) 是由 谭超 金勇� 王诗太 谭海风 李克 刘琦 范红梅 喻赛波 陈潜 于 2018-08-24 设计创作，主要内容包括：本发明公开了一种预测烟支烟燃烧锥落头的方法,通过训练集,创建了分割法模型。所述模型的作用主要是建立燃烧锥落头与烟丝填充状态的关系。本发明还公开了一种预测烟支烟燃烧锥落头的方法的应用。采用该方法得到的预测值与实测值之间的决定系数(R<Sup>2</Sup>)和一致性指标(D指标)都达到0.9以上。与现有的直接测定燃烧锥落头率的方法相比,本发明实现了破坏性检测到无损检测的转换,且能够得出整支烟可能落头的全部位置,位置精度可达1mm。(The invention discloses a method for predicting the falling head of a cigarette combustion cone. The model is mainly used for establishing the relation between the combustion cone falling head and the tobacco shred filling state. The invention also discloses application of the method for predicting the falling end of the cigarette combustion cone. The determination coefficient (R) between the predicted value and the measured value obtained by the method 2 ) And the consistency index (D index) both reach over 0.9. Compared with the existing method for directly measuring the head falling rate of the combustion cone, the method realizes the conversion of destructive detection and nondestructive detection, can obtain all possible head falling positions of the whole cigarette, and has the position precision of 1 mm.)

1. A method for predicting the falling head of a cigarette combustion cone is characterized by comprising the following steps:

(1) collecting the actual density of each site of the cigarette in the training set, and then solving the actual density change rate of each site;

(2) igniting the cigarettes in the training set, and recording the falling point and the non-falling point of the cigarette combustion cone;

(3) fitting the actual density and actual density change rate of the cigarette end falling sites recorded in the step (2) and the actual density and actual density change rate of the cigarette end falling sites to obtain a cigarette combustion cone end falling prediction model: x is the number of₁<A×y₁+B；x₂>A×y₂+B；

In the formula: A. b is a specific fitting parameter; x is the number of₁Is the actual density of the cigarette end point, y₁The actual density change rate of the cigarette end falling point is shown; x is the number of₂Is the actual density of the cigarette end drop site, y₂The actual density change rate of the cigarette non-falling end point is shown;

(4) obtaining the value ranges of A and B through linear planning;

(5) measuring the actual density p of each position of the cigarettes in the test set;

(6) calculating the change rate rho' of the actual density of each position of the cigarettes in the test set;

(7) calculating theoretical density rho of each site of the cigarette: ρ ═ a × ρ' + B;

(8) and predicting whether the cigarettes can generate combustion cone falling heads or not through the relation between the p and the rho.

2. The method of predicting the cone drop of cigarette combustion as set forth in claim 1, wherein: the actual density is determined by microwave method.

3. The method of predicting the cone drop of cigarette combustion as set forth in claim 1, wherein: when the P is smaller than rho, predicting the cigarette falling head; and when the P is larger than or equal to the rho, predicting that the cigarettes do not fall.

4. The method of predicting the cone drop of cigarette combustion as set forth in claim 1, wherein: the value taking range of A and B takes the value near the middle point in the value taking range of A and B.

5. The use of the method for predicting the cone falling head of the combustion of cigarettes according to any one of claims 1 to 4, wherein: the method is applied to predicting the falling rate of the cigarette combustion cone.

6. Use of a method for predicting the cone drop of a cigarette combustion according to claim 4, characterized in that: dividing the cigarettes into N sections, wherein each section comprises 5-12 sites, and the number of the falling-end sites is predicted by the method for predicting the falling-end of the cigarette combustion cone and divided by the total number of the sites to obtain a segmented falling-end rate, wherein the maximum value of the segmented falling-end rate is the falling-end rate of the cigarette combustion cone; wherein N is an integer part of the actual mouth number of the cigarette.

Technical Field

The invention relates to the technical field of cigarette product quality detection, in particular to a method for predicting whether a cigarette combustion cone falls off by using cigarette density and application thereof.

Background

The falling end of the combustion cone can interrupt the smoking of the cigarette, and the recognition degree of the cigarette brand by consumers is reduced; a falling fire head may also cause a fire, impairing the benefits of the consumer.

In view of the common phenomenon of the falling head of the thin cigarette, the tobacco industry will quickly release the industry standard of the falling head inclination test of the cigarette combustion cone. In the future, tobacco industry enterprises will bring the combustion cone falling head into the cigarette quality control category.

The existing research relates to the influence of factors such as tobacco shred structure, cigarette paper burning speed, shredding width, moisture and the like on the burning cone falling head. The nature of the influencing factors is that the stress of the cigarette combustion cone is changed, so that the cigarette end falling performance is changed.

The filling state of tobacco shreds in cigarettes is also an important influence factor for changing the stress of a cigarette combustion cone. However, the current research only relates to the influence of tobacco shred filling on the density of cigarettes by different levels, the influence of the levels on the cigarette blank rate and the influence of the standard deviation of the cigarette density on the end shred dropping amount, and the report of researching the burning cone dropping from the tobacco shred filling angle is rare.

The existing cigarette combustion cone falling head detection method mainly utilizes a machine to knock or flick a burning cigarette. The drawbacks of the method are mainly two: one is that the cigarette must be lit to determine its percentage drop. The sample must be destroyed during the test. The second is that only the first position where the head fall occurs can be measured, and other positions where the head fall is likely cannot be predicted.

Chinese patent CN108303344A discloses a method for judging the falling of a cigarette combustion cone. The method utilizes the reciprocal of the cut tobacco filling value to obtain the optimal theoretical density, and then compares the optimal theoretical density with two judgment intervals of the actually measured cigarette density of the sample to obtain the proportion of the cigarette combustion cone of the sample which possibly falls off. The patent only considers the physical index of the tobacco shred filling value, and the falling tendency of the combustion cone is also influenced by other factors such as the burning speed of the cigarette paper, the moisture, the slices added in the tobacco shreds and the like. The combustion cone drop ratios measured with different cigarette papers are different, but the combustion cone drop ratios predicted by this method do not change. Therefore, the method has a narrow application range.

Disclosure of Invention

Compared with the prior art, the invention has the first aim of providing a method for predicting the falling head of the combustion cone only from the tobacco shred filling state by using the falling head prediction model established by the cigarette density without damaging a sample.

A second object of the invention is to provide an application of the method of predicting the combustion cone falling head.

The invention discloses a method for predicting the falling head of a cigarette combustion cone, which comprises the following steps:

(1) collecting the actual density of each site of the cigarette in the training set, and then solving the actual density change rate of each site;

(2) igniting the cigarettes in the training set, and recording the falling point and the non-falling point of the cigarette combustion cone;

(3) fitting the actual density and actual density change rate of the cigarette end falling sites recorded in the step (2) and the actual density and actual density change rate of the cigarette end falling sites to obtain a cigarette combustion cone end falling prediction model: x is the number of₁<A×y₁+B；x₂>A×y₂+B；

In the formula: A. b is a specific fitting parameter; x is the number of₁Is the actual density of the cigarette end point, y₁The actual density change rate of the cigarette end falling point is shown; x is the number of₂Is the actual density of the cigarette end drop site, y₂The actual density change rate of the cigarette non-falling end point is shown;

(4) obtaining the value ranges of A and B through linear planning;

(5) measuring the actual density P of each position of the cigarettes in the test set;

(6) calculating the change rate rho' of the actual density of each position of the cigarettes in the test set;

(7) calculating theoretical density rho of each site of the cigarette: ρ ═ a × ρ' + B;

(8) and predicting whether the cigarette is subjected to combustion cone falling or not according to the relation between P and rho.

The difficulty of falling the cigarette burning cone head depends on the intensity of the clamping force applied to the tobacco shreds at the root of the burning cone. When the end-falling device knocks the cigarette, the combustion cone in a static state is subjected to instant impact force, so that inertia force separated from the cigarette is generated. If the clamping force is less than the inertial force, the combustion cone will fall, otherwise it will not.

The higher the cigarette density at a certain point is, the more dense the tobacco shred is filled, and the clamping force is large. Meanwhile, the self weight of the combustion cone is larger, and the inertia force is also larger.

Along the moving direction of the cigarette combustion line, if the cigarette density is reduced from large to small, the change rate of the density at the point is negative, and the density of the area adjacent to the root of the combustion cone is suddenly reduced from the side close to the cone to the side close to the unburned tobacco shreds. The high-density cone side has high inertia force and small tobacco shred side density and small clamping force, and the combustion cone has a tendency of falling. If the cigarette density is changed from small to large, the density change rate of the point is positive. This indicates that the density of the area near the base of the combustion cone increases abruptly from the side near the cone to the side near the unburned tobacco. This results in a reduction in the high inertial force of the cone side density, an increase in the grip force with a small tobacco side density, and no tendency for the combustion cone to fall. And if the density of the cigarette does not change, the density change rate at the point is zero.

The inventors have discovered through research that by determining the model parameters from the training set, the factors affecting the combustion cone falling head from the cigarette itself can all be taken into account and integrated into the model parameters A, B to arrive at the correct relationship between cigarette density and combustion cone falling head. When the raw and auxiliary materials of the cigarette are changed, the training set is reset. Therefore, the model has strong self-adaptive capacity, and the original auxiliary materials are changed, so that higher prediction accuracy can be maintained.

The actual density is determined by microwave method.

In the implementation process, the cigarette can be divided into one point every 1-2 mm. Preferably, one site is divided every 1 mm.

Preferably, the cigarette density change rate can be automatically calculated by origin software according to the actual density of each position of the cigarette. Other software with slope calculation functionality is also available.

Preferably, Lingo is selected for linear programming. Other linear programming software may also be implemented.

As shown in fig. 1, when P is smaller than ρ, the cigarette end drop is predicted; and when P is greater than or equal to rho, predicting that the cigarette does not fall.

The preferred value taking range of A and B is the value near the middle point in the value taking range of A and B.

Further preferably, the values of a and B are midpoint values in the value ranges of a and B.

The invention discloses application of a method for predicting the falling head of a cigarette combustion cone, which is applied to predicting the falling head rate of the cigarette combustion cone.

Specifically, the cigarette is divided into N sections, each section comprises 5-12 sites, the number of the falling-head sites is predicted by the method for predicting the falling head of the cigarette combustion cone, the total number of the falling-head sites is divided by the total number of the sites to obtain the falling-head rate of each section, and the maximum value is the falling-head rate of the cigarette combustion cone; wherein N is an integer part of the actual mouth number of the cigarette.

The cigarette segmentation is divided according to the position of each combustion line of the training set sample, and the maximum requirement that each position falls into the divided segment is met, and the segments are not overlapped. The first paragraph is not counted, taking into account the specific situation in which a person smokes.

The predicted head drop rate for the test set samples is the average of the head drop rates for all samples in the test set.

Compared with the prior art, the invention has the following beneficial effects:

1. the invention has the advantages of small root mean square error and normalized root mean square error between the head falling rate value and the measured value, and good prediction effect. The decision coefficient and the consistency index of the method are close to 1, and the consistency between the predicted value and the measured value is high.

2. For cigarettes only changing the filling amount or distribution of tobacco shreds, the falling rate of the combustion cone can be predicted only by measuring the density data of the cigarettes without damaging cigarette samples, so that the nondestructive testing of the falling rate of the cigarettes is realized.

3. The first position where the cigarette ends fall can only be obtained by measuring the cigarette end falling rate by using the end falling detection device, and the segmentation method model can obtain all possible end falling positions of the whole cigarette, so that the position precision can reach 1mm, and the method is favorable for pertinently taking measures to improve the distribution of the cut tobacco and reduce the end falling rate.

4. The method can predict the falling rate of cigarettes with different single weights and different tobacco shred distributions, can predict the falling rate by newly establishing a training set and modeling again even if the raw and auxiliary materials of the cigarettes are changed, and has wide application range.

Drawings

FIG. 1 is a schematic diagram of a segmentation model.

Whether a cigarette falls off or not is influenced by the density rho and the change rate rho ', in a rectangular coordinate system of a horizontal coordinate rho' and a vertical coordinate rho, the straight line rho is A multiplied by rho '+ B to divide an upper area and a lower area, a point with coordinates of (rho', rho) is positioned at a point with an equivalent condition of rho > A multiplied by rho '+ B above the straight line, and an equivalent condition of rho < A multiplied by rho' + B below the straight line. The point located in the area above the straight line is judged as not falling, and the point located in the area below the straight line is judged as falling.

FIG. 2 shows the comparison between the predicted and actual head-falling rates of different single-weight samples.

FIG. 3 is a comparison between the predicted value and the measured value of the percentage of the falling ends of tobacco shred distribution samples.

Detailed Description

Embodiments of the present invention will be described in further detail below with reference to the drawings of the embodiments of the present invention, and it should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The cigarette site density is measured according to YC/T476-2013 microwave method for measuring cigarette tobacco density.

The invention adopts the following indexes to evaluate the accuracy of the prediction model:

1: 1 line of predicted value and measured value. The method can visually check the simulation performance of the model.

Root Mean Square Error (RMSE), Normalized Root Mean Square Error (NRMSE) between predicted and measured values. RMSE and NRMSE may reflect relative and absolute errors between predicted and measured values.

Determining the coefficient (R)²) And a consistency index (D index). R²And the D index may reflect the consistency between the predicted value and the measured value, with values closer to 1 indicating higher consistency between the predicted value and the measured value.

The RMSE, NRMSE and D indices are calculated as follows:

in the formula: yi and Xi are respectively an analog value and an actual measurement value; x is the average value of the measured data; n is the sample volume. The simulation value is the average value of the head falling rates predicted by a group of samples according to the five steps of the invention, and the actual measurement value is the actual head falling rate obtained by measuring the group of samples by using a head falling detection instrument. n is the number of samples.