CN117272239A - Cigarette ash packing performance prediction method based on random forest algorithm - Google Patents
Cigarette ash packing performance prediction method based on random forest algorithm Download PDFInfo
- Publication number
- CN117272239A CN117272239A CN202311154051.5A CN202311154051A CN117272239A CN 117272239 A CN117272239 A CN 117272239A CN 202311154051 A CN202311154051 A CN 202311154051A CN 117272239 A CN117272239 A CN 117272239A
- Authority
- CN
- China
- Prior art keywords
- ash
- cigarette
- performance
- prediction
- cigarette paper
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 235000019504 cigarettes Nutrition 0.000 title claims abstract description 267
- 238000012856 packing Methods 0.000 title claims abstract description 97
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000007637 random forest analysis Methods 0.000 title claims abstract description 46
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 30
- 239000000463 material Substances 0.000 claims abstract description 35
- 230000001419 dependent effect Effects 0.000 claims abstract description 12
- 230000035699 permeability Effects 0.000 claims description 38
- FKNQFGJONOIPTF-UHFFFAOYSA-N Sodium cation Chemical compound [Na+] FKNQFGJONOIPTF-UHFFFAOYSA-N 0.000 claims description 17
- 229910001415 sodium ion Inorganic materials 0.000 claims description 17
- NPYPAHLBTDXSSS-UHFFFAOYSA-N Potassium ion Chemical compound [K+] NPYPAHLBTDXSSS-UHFFFAOYSA-N 0.000 claims description 16
- 229910001414 potassium ion Inorganic materials 0.000 claims description 16
- KRKNYBCHXYNGOX-UHFFFAOYSA-K Citrate Chemical compound [O-]C(=O)CC(O)(CC([O-])=O)C([O-])=O KRKNYBCHXYNGOX-UHFFFAOYSA-K 0.000 claims description 14
- 229910019142 PO4 Inorganic materials 0.000 claims description 14
- 239000010452 phosphate Substances 0.000 claims description 14
- MUBZPKHOEPUJKR-UHFFFAOYSA-N Oxalic acid Chemical compound OC(=O)C(O)=O MUBZPKHOEPUJKR-UHFFFAOYSA-N 0.000 claims description 11
- NBIIXXVUZAFLBC-UHFFFAOYSA-K phosphate Chemical compound [O-]P([O-])([O-])=O NBIIXXVUZAFLBC-UHFFFAOYSA-K 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 10
- 238000011002 quantification Methods 0.000 claims description 9
- 239000011248 coating agent Substances 0.000 claims description 8
- 238000000576 coating method Methods 0.000 claims description 8
- 238000005336 cracking Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000013461 design Methods 0.000 abstract description 4
- 238000003908 quality control method Methods 0.000 abstract description 3
- 238000012827 research and development Methods 0.000 abstract description 2
- 239000000123 paper Substances 0.000 description 131
- 238000009423 ventilation Methods 0.000 description 36
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 16
- 229910052799 carbon Inorganic materials 0.000 description 16
- 230000000052 comparative effect Effects 0.000 description 14
- 238000004458 analytical method Methods 0.000 description 10
- -1 phosphate radical Chemical class 0.000 description 8
- 238000012549 training Methods 0.000 description 7
- 238000003066 decision tree Methods 0.000 description 6
- 238000005259 measurement Methods 0.000 description 5
- VTYYLEPIZMXCLO-UHFFFAOYSA-L Calcium carbonate Chemical compound [Ca+2].[O-]C([O-])=O VTYYLEPIZMXCLO-UHFFFAOYSA-L 0.000 description 4
- 238000012163 sequencing technique Methods 0.000 description 4
- 241000208125 Nicotiana Species 0.000 description 3
- 235000002637 Nicotiana tabacum Nutrition 0.000 description 3
- 238000002485 combustion reaction Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 229910000019 calcium carbonate Inorganic materials 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 239000002245 particle Substances 0.000 description 2
- 239000000779 smoke Substances 0.000 description 2
- QTBSBXVTEAMEQO-UHFFFAOYSA-M Acetate Chemical compound CC([O-])=O QTBSBXVTEAMEQO-UHFFFAOYSA-M 0.000 description 1
- 239000002253 acid Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003763 carbonization Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012938 design process Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004255 ion exchange chromatography Methods 0.000 description 1
- ZLNQQNXFFQJAID-UHFFFAOYSA-L magnesium carbonate Chemical compound [Mg+2].[O-]C([O-])=O ZLNQQNXFFQJAID-UHFFFAOYSA-L 0.000 description 1
- 229910000021 magnesium carbonate Inorganic materials 0.000 description 1
- 239000001095 magnesium carbonate Substances 0.000 description 1
- 235000006408 oxalic acid Nutrition 0.000 description 1
- 239000011087 paperboard Substances 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- BITYAPCSNKJESK-UHFFFAOYSA-N potassiosodium Chemical compound [Na].[K] BITYAPCSNKJESK-UHFFFAOYSA-N 0.000 description 1
- 150000003254 radicals Chemical class 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000391 smoking effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/34—Paper
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Manufacturing Of Cigar And Cigarette Tobacco (AREA)
Abstract
The invention relates to a prediction method of cigarette ash packing performance based on a random forest algorithm. The prediction method comprises the following steps: s1, taking cigarettes as modeling samples, and detecting auxiliary material parameters, physical indexes and ash wrapping performance of the modeling samples; s2, taking the auxiliary material parameters and physical indexes obtained in the step S1 as independent variables, taking the ash wrapping performance as a dependent variable, and constructing a cigarette ash wrapping performance prediction model by adopting a random forest algorithm; s3, detecting auxiliary material parameters and physical indexes of the unknown sample, and inputting the auxiliary material parameters and physical indexes into a cigarette ash packing performance prediction model to obtain the ash packing performance of the unknown sample. The prediction method can accurately predict the ash packing performance of the cigarettes and analyze the importance of the auxiliary material parameters and physical indexes of the cigarettes, and can provide important reference basis for the design, research and development and quality control of cigarette products.
Description
Technical Field
The invention relates to the technical field of cigarette ash packing performance detection, in particular to a method for predicting cigarette ash packing performance based on a random forest algorithm.
Background
With the continuous development of the tobacco industry, consumers pay more and more attention to the cigarette ash wrapping performance, the quality of the cigarette ash wrapping performance directly influences the smoking experience of the consumers, and meanwhile, the cigarette ash wrapping performance is an important index for evaluating the comprehensive quality of cigarettes. The indexes for evaluating the ash wrapping performance of the cigarettes comprise whiteness value, cracking rate, ash shrinkage, carbon line uniformity, carbon line width, ash holding capacity and the like. If the cigarette ash packing performance can be accurately predicted in advance in the cigarette design process, the cigarette ash packing performance can be better controlled, and the method has important significance for guiding the digital design of cigarette products. For the prediction of the ash packing performance of cigarettes, related researches are also carried out in the industry, for example, a Chinese patent named as a method for quantitatively predicting the ash packing effect in the combustion process of cigarette paper in advance constructs a quantitative relation linear model of the crack rate, the ash shrinkage rate, the carbon line uniformity, the carbon line width and the calcium carbonate particle coverage area through the calcium carbonate particle coverage area; li Huan et al (see literature (influence of cigarette paper parameters on cigarette ash holding capacity, (2021)) construct a linear prediction model of cigarette paper ration, air permeability, magnesium carbonate addition, sodium-potassium ratio of combustion improver, acid radical type and combustion improver consumption on ash holding capacity of cigarette package; xu Yanran (see the literature of the influence of tobacco shred and cigarette machine parameters on the cigarette ash coating performance (2020)) establishes a linear prediction model of tobacco shred and cigarette machine parameters on ash coating values, gray values, whiteness values, cone heights and carbonization ring widths through multiple linear regression analysis.
The prediction methods are all based on linear fitting models, and the linear fitting models have the defects of low accuracy, easiness in overfitting, large influence of noise and the like, so that the application of the prediction methods in cigarette ash wrapping performance is limited.
Disclosure of Invention
The invention aims to solve the problems that the existing models for predicting the cigarette ash packing performance are mostly linear fitting models, are low in accuracy, easy to overfit, greatly affected by noise and the like, and provides a prediction method for the cigarette ash packing performance based on a random forest algorithm.
The above object of the present invention is achieved by the following technical solutions:
a prediction method of cigarette ash packing performance based on a random forest algorithm comprises the following steps:
s1, taking cigarettes as modeling samples, and detecting auxiliary material parameters, physical indexes and ash wrapping performance of the modeling samples;
s2, taking the auxiliary material parameters and physical indexes obtained in the step S1 as independent variables, taking the ash wrapping performance as a dependent variable, and constructing a cigarette ash wrapping performance prediction model by adopting a random forest algorithm;
s3, detecting auxiliary material parameters and physical indexes of an unknown sample, and inputting the auxiliary material parameters and physical indexes into a cigarette ash packing performance prediction model to obtain the ash packing performance of the unknown sample;
the auxiliary material parameters in the step S1 and the step S3 are tipping paper air permeability, cigarette paper phosphate content, cigarette paper oxalate content, cigarette paper citrate content, cigarette paper sodium ion content, cigarette paper potassium ion content, cigarette paper air permeability and cigarette paper quantification, and the physical indexes are filter tip air permeability, cigarette paper air permeability, total air permeability, cigarette weight, open suction resistance and closed suction resistance;
the ash coating performance in the steps S1, S2 and S3 is whiteness value, cracking rate or ash holding force.
Compared with a linear regression fit model, the prediction model constructed by the random forest algorithm has special advantages, the random forest algorithm can process data with very high dimensionality and is not used for selecting characteristic variables, after model training is completed, the sequencing result of the characteristic variables can be given, and the method is superior to the linear fit model in anti-interference and anti-overfitting aspects.
The inventor of the invention finds that the random forest algorithm is used for predicting the cigarette ash wrapping performance, the independent variable is critical to select, the influence on different dependent variables is different, and the two influence the accuracy of the model, thereby influencing the applicability of the random forest algorithm. The invention uses specific auxiliary material parameters and physical indexes of the cigarettes as independent variables, adopts a random forest algorithm to construct a cigarette ash packing performance prediction model, and can accurately predict the specific ash packing performance (whiteness value, cracking rate or ash holding force) of unknown samples based on the cigarette ash packing performance prediction model. In addition, the prediction method can accurately sort the importance of the auxiliary material parameters and physical indexes of the cigarettes to the ash packing performance of the specific cigarettes. Therefore, the prediction method can accurately predict the ash packing performance (whiteness value, cracking rate or ash holding force) of the cigarettes, analyze the importance of auxiliary material parameters and physical indexes of the cigarettes, and provide important reference basis for the design, development and quality control of cigarette products.
In the invention, the following components are added: the ventilation rate of the filter tip, the ventilation rate of the cigarette paper, the total ventilation rate, the weight of the cigarette, the open suction resistance and the closed suction resistance can be measured according to GB/T22838.5-2009 measurement of physical properties of the cigarette and the filter stick;
the air permeability of tipping paper and the air permeability of cigarette paper can be measured according to GB/T23227-2008 'measurement of air permeability of cigarette paper, forming paper, tipping paper and materials with directional air-permeable belts';
the phosphate radical content of the cigarette paper, the oxalate radical content of the cigarette paper, the citrate radical content of the cigarette paper, the sodium ion content of the cigarette paper and the potassium ion content of the cigarette paper can be measured according to YC/T275-2008 ion chromatography for measuring citrate ions, phosphate radical ions and acetate ions in the cigarette paper;
the quantitative determination of the cigarette paper can be determined according to GB/T451.2 paper and paperboard quantitative determination;
the ash packing performance can be measured according to CN 217846213U, CN 114965462A, a real-time tracking and detecting method for ash holding capacity of cigarette ash, and influence of cigarette paper characteristics on static ash packing performance of cigarettes (DOI: 10.16135/j. Issn 1002-0861.2019.0093).
Preferably, the number of modeling samples in step S1 is greater than 56.
More preferably, the number of modeling samples in step S1 is 56 to 150.
Preferably, the random forest algorithm in step S2 is a random forest classification method or a random forest regression method.
Preferably, the specific process in step S2 is: taking the auxiliary material parameters and the physical indexes obtained in the step S1 as independent variables, taking the ash wrapping performance as dependent variables, introducing the auxiliary material parameters, the physical indexes and the auxiliary material indexes into an IBM SPSS model 18.0 data processing system, selecting a random forest algorithm, and setting construction parameters to finish the construction of the cigarette ash wrapping performance prediction model.
More preferably, the build parameters include the number of models, sample size, or minimum number of children.
Further preferably, the number of the models is 100 to 150, the sample size is 0.9 to 1.0, and the minimum number of the sub-nodes is 1 to 5.
Preferably, the step S2 further includes a step of performing internal prediction by using the cigarette ash packing performance prediction model after the cigarette ash packing performance prediction model is constructed.
More preferably, the specific process of the step of adopting the cigarette ash packing performance prediction model to perform internal prediction is as follows: and adopting a cigarette ash packing performance prediction model to conduct internal prediction on the ash packing performance of the modeling sample.
Preferably, the ash packing performance in steps S1, S2 and S3 is the cracking rate.
Compared with the prior art, the invention has the beneficial effects that:
the prediction method can accurately predict the ash coating performance (whiteness value, cracking rate or ash holding force) of the cigarettes, analyze the importance of auxiliary material parameters and physical indexes of the cigarettes, and provide important reference basis for the design, research and development and quality control of cigarette products.
Drawings
FIG. 1 is a graph comparing measured values of the ash packing crack rate with predicted values inside the model for the modeling sample of example 1.
Fig. 2 is a graph of the residual results of the measured values of the ash inclusion breach rate and the model internal prediction values of the modeling sample of example 1.
FIG. 3 is a graph comparing measured values of the ash inclusion values with predicted values inside the model for the modeled samples of example 2.
Fig. 4 is a graph of the residual results of the measured values of the inclusion gray scale values and the model internal predictions of the modeled samples of example 2.
Fig. 5 is a graph comparing the measured values of the ash holding force of the modeling sample of example 3 with the predicted values inside the model.
Fig. 6 is a graph of the residual results of the measured values of the ash holding force and the model internal prediction values of the modeling sample of example 3.
Fig. 7 is a display view of raw data (part) of the modeling sample detection of the examples and comparative examples.
Fig. 8 is a display view of raw data (part) of the modeling sample detection of the examples and comparative examples.
Detailed Description
The present invention will be described in further detail with reference to the following specific examples for the purpose of illustration and not limitation, and various modifications may be made within the scope of the present invention as defined by the appended claims.
Example 1
The embodiment provides a method for predicting cigarette ash packing performance (ash packing breach rate) based on a random forest algorithm, which comprises the following steps:
1. taking 56 cigarettes as modeling samples, and detecting auxiliary material parameters and physical indexes (including filter tip ventilation rate, cigarette paper ventilation rate, total ventilation rate, cigarette weight, open smoke resistance, closed smoke resistance, tipping paper air permeability, cigarette paper phosphate content, cigarette paper oxalate content, cigarette paper citrate content, cigarette paper sodium ion content, cigarette paper potassium ion content, cigarette paper air permeability and cigarette paper quantification, 14 in total) of each modeling sample.
2. Introducing the index detected in the step 1 into an IBM SPSS model 18.0 data processing system, wherein the auxiliary material parameters and the physical indexes are taken as independent variables (14 in total), and the ash packing split rate is taken as the dependent variable; and selecting a random forest regression method, setting parameters including the number of models to be constructed, the sample size, the maximum node number (wherein the number of models is set to 100, the sample size is set to 1.0, the minimum node number is set to 5) and the like, and constructing a cigarette ash packing performance prediction model. Model internal predictions were made for the ash packing breach rate for 56 modeled samples.
3. And taking unknown samples of 5 cigarettes, and respectively measuring the ventilation rate of the filter tip, the ventilation rate of the cigarette paper, the total ventilation rate, the weight of the cigarette, the open suction resistance, the closed suction resistance, the air permeability of tipping paper, the phosphate content of the cigarette paper, the oxalate content of the cigarette paper, the citrate content of the cigarette paper, the sodium ion content of the cigarette paper, the potassium ion content of the cigarette paper, the air permeability of the cigarette paper and the quantification of the cigarette paper. Then, predicting the breach rate of unknown samples of 5 cigarettes based on the cigarette ash packing performance prediction model obtained in the step 2; the results are shown in Table 1 below. Table 2 is an analysis of the cigarette ash packing breach rate prediction results for unknown samples.
TABLE 1 ash packing breach rate prediction results for unknown samples
| Unknown sample number | Predicted value/% | Measured value/% | Residual/% |
| LK-1 | 23.52 | 23.97 | 0.45 |
| LK-2 | 13.46 | 13.68 | 0.22 |
| LK-3 | 20.42 | 20.57 | 0.16 |
| LK-4 | 19.43 | 20.48 | 1.06 |
| LK-5 | 19.09 | 19.61 | 0.52 |
TABLE 2 analysis of cigarette ash packing breach rate prediction results for unknown samples
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Mean square error |
| -1.055 | -0.155 | -0.480 | 0.480 | 0.356 | 0.332 |
As can be seen from tables 1 and 2, the prediction residual errors of the ash packing breach rate of the cigarettes are within 1.5%, and the average absolute error is only 0.480%, which indicates that the constructed ash packing performance prediction model of the cigarettes can accurately pack the ash breach rate of unknown samples.
In step 2, the process of model internal prediction of the ash packing breach rate of 56 modeling samples is as follows:
and (3) adopting the cigarette ash packing performance prediction model in the step (2) to conduct internal prediction on the ash packing split rate of 56 modeling samples, comparing the predicted value with the actual measured value, calculating residual error, minimum error, maximum error, average absolute error, standard deviation, linear correlation and mean square error, and verifying the prediction capability of the model. The results are shown in Table 3, FIG. 1, FIG. 2 and Table 4.
TABLE 3 prediction and actual measurement values and residuals within cigarette ash packing performance prediction model for modeling samples
TABLE 4 analysis of the results of the comparison of the predicted and measured values within the cigarette ash packing performance prediction model for modeling samples
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Linear correlation | Mean square error |
| -3.461 | 4.091 | 0.114 | 1.376 | 1.697 | 0.959 | 2.843 |
As can be seen from tables 3 and 4, the linear correlation fitting degree R of the predicted value and the measured value in the cigarette ash packing performance prediction model of the modeling sample is 0.959, which indicates that the constructed cigarette ash packing performance prediction model has better prediction capability on the ash packing breach rate.
In addition, the embodiment also analyzes the importance of the auxiliary material parameters and physical indexes of the cigarettes on the ash packing split rate, and the process is as follows:
the total number of cigarette samples of the training model was set to 56 and the number of random forest decision trees constructed was 100. When each decision tree splits at a node, the mode that the entropy of 14 independent variables in the training model is maximized is used as the optimal splitting mode. And calculating the overall entropy increase of each characteristic independent variable in the random forest model, and sequencing the characteristic independent variables according to the overall entropy increase, wherein the more important the overall entropy increase is.
The importance ranking results of the cigarette auxiliary parameters and physical indexes are shown in table 5 (only auxiliary parameters and physical indexes of 10 before importance ranking are given).
TABLE 5 importance ranking of cigarette auxiliary parameters and physical indicators
| Variable(s) | Importance value | Importance ranking |
| Sodium ion content of cigarette paper | 126.37 | 1 |
| Cigarette paper potassium ion content | 121.21 | 2 |
| Quantitative cigarette paper | 51.48 | 3 |
| Cigarette paper citrate content | 40.67 | 4 |
| Air permeability of cigarette paper | 38.95 | 5 |
| Ventilation rate of filter tip | 38.72 | 6 |
| Cigarette paper phosphate radical content | 38.27 | 7 |
| Ventilation rate of cigarette paper | 22.74 | 8 |
| Total ventilation rate | 16.41 | 9 |
| Air permeability of tipping paper | 13.17 | 10 |
From table 5, the importance ranking of the cigarette auxiliary parameters and physical indexes to the ash packing breach rate in the random forest model is as follows: the sodium ion content of the cigarette paper is larger than the potassium ion content of the cigarette paper, the quantitative content of the cigarette paper is larger than the citrate content of the cigarette paper, the air permeability of the cigarette paper is larger than the ventilation rate of the filter tip is larger than the phosphate content of the cigarette paper, the ventilation rate of the cigarette paper is larger than the total ventilation rate of the cigarette paper, and the air permeability of the tipping paper is larger than the ventilation rate of the tipping paper.
Example 2
The embodiment provides a method for predicting cigarette ash wrapping performance (ash wrapping whiteness value) based on a random forest algorithm, which comprises the following steps:
1. taking 56 cigarettes as modeling samples, and detecting auxiliary material parameters and physical indexes (including filter tip ventilation rate, cigarette paper ventilation rate, total ventilation rate, cigarette weight, open suction resistance, closed suction resistance, tipping paper air permeability, cigarette paper phosphate content, cigarette paper oxalate content, cigarette paper citrate content, cigarette paper sodium ion content, cigarette paper potassium ion content, cigarette paper air permeability and cigarette paper quantification, 14 in total) of each modeling sample.
2. Introducing the index detected in the step 1 into an IBM SPSS model 18.0 data processing system, wherein auxiliary material parameters and physical indexes are taken as independent variables (14 in total), and the gray scale value is taken as a dependent variable; and selecting a random forest regression method, setting parameters including the number of models to be constructed, the sample size, the maximum node number (wherein the number of the models is set to 100, the sample size is set to 1.0, the minimum node number is set to 5) and the like, and constructing a cigarette ash inclusion performance prediction model to carry out model internal prediction on ash inclusion whiteness values of 56 modeling samples.
3. And taking unknown samples of 5 cigarettes, and respectively measuring the ventilation rate of the filter tip, the ventilation rate of the cigarette paper, the total ventilation rate, the weight of the cigarette, the open suction resistance, the closed suction resistance, the air permeability of tipping paper, the phosphate content of the cigarette paper, the oxalate content of the cigarette paper, the citrate content of the cigarette paper, the sodium ion content of the cigarette paper, the potassium ion content of the cigarette paper, the air permeability of the cigarette paper and the quantification of the cigarette paper. Then, based on the cigarette ash coating performance prediction model obtained in the step 2, predicting whiteness values of unknown samples of 5 cigarettes; the results are shown in Table 6 below. Table 7 is an analysis of the predicted cigarette packet ash values for unknown samples.
Table 6 prediction results of the ash inclusion values for unknown samples
| Unknown sample number | Predicted value/% | Measured value/% | Residual/% |
| BD-1 | 60.36 | 58.76 | -1.60 |
| BD-2 | 60.79 | 61.74 | 0.95 |
| BD-3 | 55.04 | 54.42 | -0.62 |
| BD-4 | 60.36 | 59.42 | -0.94 |
| BD-5 | 58.70 | 57.65 | -1.05 |
Table 7 analysis of the prediction results of the ash-containing whiteness values of cigarettes for unknown samples
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Mean square error |
| -0.948 | 1.602 | -0.654 | 1.033 | 0.962 | 1.168 |
As can be seen from tables 6 and 7, the prediction residual errors of the ash packing whiteness values of the cigarettes are within 1.6%, and the average absolute error is only 1.033%, which indicates that the constructed ash packing performance prediction model of the cigarettes can accurately measure the ash packing whiteness values of unknown samples.
In step 2, the process of model internal prediction of the ash packing breach rate of 56 modeling samples is as follows:
and (3) adopting the cigarette ash inclusion performance prediction model in the step (2) to internally predict ash inclusion whiteness values of 56 modeling samples, comparing the predicted values with actual measured values, calculating residual errors, minimum errors, maximum errors, average absolute errors, standard deviations, linear correlations and mean square errors, and verifying the prediction capability of the model. The results are shown in tables 8, 3, 4 and 9.
Table 8 modeling sample cigarette packet ash level prediction model internal prediction and actual measurement values and residuals
TABLE 9 analysis of the results of the comparison of the predicted and measured values within the cigarette packet ash values prediction model for the modeled samples
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Linear correlation | Mean square error |
| -2.3 | 4.031 | -0.146 | 0.779 | 1.069 | 0.936 | 1.144 |
As can be seen from tables 8 and 9, R of the predicted value and the measured value in the cigarette ash packing value prediction model of the modeling sample is 0.936, which indicates that the constructed cigarette ash packing performance prediction model has better prediction capability on the ash packing value.
In addition, the importance of the auxiliary material parameters and the physical indexes of the cigarettes on the ash packing whiteness value is analyzed, and the process is as follows:
the total number of cigarette samples of the training model was set to 56 and the number of random forest decision trees constructed was 100. When each decision tree splits at a node, the mode that the entropy of 14 independent variables in the training model is maximized is used as the optimal splitting mode. And calculating the overall entropy increase of each characteristic independent variable in the random forest model, and sequencing the characteristic independent variables according to the overall entropy increase, wherein the more important the overall entropy increase is.
The importance ranking results of the cigarette auxiliary parameters and physical indexes are shown in table 10 (only auxiliary parameters and physical indexes of 10 before importance ranking are given).
TABLE 10 importance ranking of cigarette auxiliary parameters and physical indicators
| Variable(s) | Importance value | Importance ranking |
| Cigarette paper citrate content | 36.72 | 1 |
| Air permeability of cigarette paper | 18.48 | 2 |
| Quantitative cigarette paper | 15.19 | 3 |
| Cigarette paper potassium ion content | 14.91 | 4 |
| Sodium ion content of cigarette paper | 12.58 | 5 |
| Cigarette paper phosphate radical content | 5.81 | 6 |
| Open suction resistor | 5.72 | 7 |
| Ventilation rate of filter tip | 5.12 | 8 |
| Total ventilation rate | 5.09 | 9 |
| Air permeability of tipping paper | 4.50 | 10 |
From table 10, the importance ranking of the cigarette auxiliary parameters and physical indexes to the ash packing whiteness value in the random forest model is as follows: cigarette paper citrate content > cigarette paper air permeability > cigarette paper quantitative content > cigarette paper potassium ion content > cigarette paper sodium ion content > cigarette paper phosphate content > open draw resistance > filter tip air permeability > total air permeability > tipping paper air permeability.
Example 3
The embodiment provides a method for predicting cigarette ash packing performance (ash packing holding force) based on a random forest algorithm, which comprises the following steps:
1. taking 56 cigarettes as modeling samples, and detecting auxiliary material parameters and physical indexes (including filter tip ventilation rate, cigarette paper ventilation rate, total ventilation rate, cigarette weight, open suction resistance, closed suction resistance, tipping paper air permeability, cigarette paper phosphate content, cigarette paper oxalate content, cigarette paper citrate content, cigarette paper sodium ion content, cigarette paper potassium ion content, cigarette paper air permeability and cigarette paper quantification, 14 in total) of each modeling sample.
2. Introducing the index detected in the step 1 into an IBM SPSS model 18.0 data processing system, wherein auxiliary material parameters and physical indexes are taken as independent variables (14 in total), and the ash holding force of the bag ash is taken as a dependent variable; and selecting a random forest regression method, setting parameters including the number of models to be constructed, the sample size, the maximum node number (wherein the number of models is set to 100, the sample size is set to 1.0, the minimum node number is set to 5) and the like, and constructing a cigarette ash packing performance prediction model. Model internal predictions were made for the ash holding power of the ash pack for 56 modeled samples.
3. And taking unknown samples of 5 cigarettes, and respectively measuring the ventilation rate of the filter tip, the ventilation rate of the cigarette paper, the total ventilation rate, the weight of the cigarette, the open suction resistance, the closed suction resistance, the air permeability of tipping paper, the phosphate content of the cigarette paper, the oxalate content of the cigarette paper, the citrate content of the cigarette paper, the sodium ion content of the cigarette paper, the potassium ion content of the cigarette paper, the air permeability of the cigarette paper and the quantification of the cigarette paper. Then, based on the cigarette ash packing performance prediction model obtained in the step 2, predicting the ash holding force of unknown samples of 5 cigarettes; the results are shown in Table 11 below. Table 12 is an analysis of the cigarette ash holding capacity predictions for unknown samples.
Table 11 ash holding capacity prediction results for unknown samples
| Unknown sample number | Predicted value/mm | Measured value/mm | Residual error/mm |
| CH-1 | 33.89 | 32.49 | -1.40 |
| CH-2 | 34.50 | 34.94 | 0.44 |
| CH-3 | 38.38 | 37.04 | -1.34 |
| CH-4 | 31.93 | 34.22 | 2.29 |
| CH-5 | 31.67 | 30.11 | -1.56 |
Cigarette ash holding capacity prediction result analysis of unknown sample in Table 12
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Mean square error |
| -2.29 | 1.563 | -0.314 | 1.407 | 1.669 | 2.327 |
As can be seen from tables 11 and 12, the prediction residual errors of the ash holding force of the cigarette ash are all within 2.5mm, and the average absolute error is only 1.407mm, which indicates that the constructed ash holding force of the cigarette ash can be accurately measured by the ash holding performance prediction model of the cigarette ash.
In step 2, the process of model internal prediction for the ash holding force of the ash wrapping of 56 modeling samples is as follows:
and (3) adopting the cigarette ash packing performance prediction model in the step (2) to internally predict the ash packing holding power of 56 modeling samples, comparing the predicted value with the measured value, calculating residual error, minimum error, maximum error, average absolute error, standard deviation, linear correlation and mean square error, and verifying the prediction capability of the model. The results are shown in Table 13, FIG. 5, FIG. 6 and Table 14.
Table 13 modeling sample cigarette ash packing performance prediction model internal prediction value and actual measurement value and residual error
/>
Table 14 analysis of the results of the comparison of the predicted and measured values within the cigarette ash packing performance prediction model of the modeled samples
| Minimum error | Maximum error | Average error | Average absolute error | Standard deviation of | Linear correlation | Mean square error |
| -3.645 | 2.199 | -0.196 | 0.784 | 1.047 | 0.947 | 1.115 |
As can be seen from tables 13 and 14, the R of the predicted value and the measured value in the cigarette ash packing performance prediction model of the modeling sample is 0.947, which indicates that the constructed cigarette ash packing performance prediction model has better prediction capability on ash packing holding power.
In addition, the embodiment also analyzes the importance of auxiliary material parameters and physical indexes of the cigarettes on the ash holding force of the wrapping ash, and the process is as follows:
the total number of cigarette samples of the training model was set to 56 and the number of random forest decision trees constructed was 100. When each decision tree splits at a node, the mode that the entropy of 14 independent variables in the training model is maximized is used as the optimal splitting mode. And calculating the overall entropy increase of each characteristic independent variable in the random forest model, and sequencing the characteristic independent variables according to the overall entropy increase, wherein the more important the overall entropy increase is.
The importance ranking results of the cigarette auxiliary parameters and physical indexes are shown in table 15 (only auxiliary parameters and physical indexes of 10 before importance ranking are given).
TABLE 15 importance ranking of cigarette auxiliary parameters and physical indicators
| Variable(s) | Importance value | Importance ranking |
| Cigarette paper citrate content | 35.46 | 1 |
| Sodium ion content of cigarette paper | 29.35 | 2 |
| Cigarette paper potassium ion content | 27.04 | 3 |
| Ventilation rate of cigarette paper | 18.81 | 4 |
| Quantitative cigarette paper | 13.24 | 5 |
| Open suction resistor | 6.81 | 6 |
| Ventilation rate of filter tip | 6.48 | 7 |
| Air permeability of cigarette paper | 2.88 | 8 |
| Oxalic acid radical content of cigarette paper | 2.80 | 9 |
| Total ventilation rate | 2.06 | 10 |
From table 15, the importance ranking of the cigarette auxiliary parameters and physical indexes on the ash holding force of the ash in the random forest model is as follows: cigarette paper citrate content & gt, cigarette paper sodium ion content & gt cigarette paper potassium ion content & gt, cigarette paper ventilation rate & gt, and cigarette paper basis weight & gt open draw resistance & gt filter tip ventilation rate & gt cigarette paper air permeability & gt cigarette paper oxalate content & gt total ventilation rate.
Comparative example 1
This comparative example provides a method for predicting cigarette ash packing performance (ash packing breach rate) based on a random forest algorithm, which is basically the same as the prediction method of example 1, except that: in the step 2, the independent variables do not comprise the sodium ion content of the cigarette paper, the potassium ion content of the cigarette paper and the quantification of the cigarette paper, namely 11 independent variables are all included.
The cigarette ash packing performance prediction model of the comparative example is adopted to conduct internal prediction on the ash packing split rate of 56 modeling samples, the predicted value is compared with the actual measured value, residual error, minimum error, maximum error, average absolute error, standard deviation, linear correlation and mean square error are calculated, and the prediction capability of the model is verified. And finally, calculating to obtain a linear correlation fitting degree R of 0.849 of the internal predicted value and the measured value, wherein the linear correlation fitting degree R is smaller than 0.9 which is required by statistics, which shows that the lack of partial independent variables (such as sodium ion content of cigarette paper) can cause poor prediction capability of a cigarette ash packing performance prediction model constructed by a random forest algorithm on the ash packing breach rate.
Comparative example 2
The present comparative example provides a prediction method of cigarette ash packing performance (ash shrinkage) based on a random forest algorithm, which is basically the same as the prediction method of example 1, except that: in the step 1, the detected ash packing performance is the ash shrinkage rate; in step 2, the dependent variable is the ash shrinkage.
The cigarette ash packing performance prediction model of the comparative example is adopted to conduct internal prediction on the ash shrinkage of 56 modeling samples, the predicted value is compared with the actual measured value, residual error, minimum error, maximum error, average absolute error, standard deviation, linear correlation and mean square error are calculated, and the prediction capability of the model is verified. And finally, calculating to obtain the linear correlation fitting degree R of the internal predicted value and the measured value which is 0.875 and smaller than 0.9 required by statistics, wherein the linear correlation fitting degree R is indicated that the prediction capability of the cigarette ash packing performance prediction model constructed by a random forest algorithm on the ash shrinkage is poor and is not suitable for the prediction of the ash shrinkage.
Comparative example 3
This comparative example provides a prediction method of cigarette ash packing performance (carbon line uniformity) based on a random forest algorithm, which is basically the same as the prediction method of example 1, except that: in the step 1, the detected ash coating performance is carbon line uniformity; in step 2, the dependent variable is carbon line uniformity.
The carbon line uniformity of 56 modeling samples is internally predicted by adopting the cigarette ash packing performance prediction model of the comparative example, a predicted value is compared with an actual measured value, residual errors, minimum errors, maximum errors, average absolute errors, standard deviations, linear correlations and mean square errors are calculated, and the prediction capability of the model is verified. And finally, calculating to obtain the linear correlation fitting degree R of the internal predicted value and the measured value which is 0.887 and smaller than 0.9 required by statistics, wherein the linear correlation fitting degree R is shown that the prediction capability of the cigarette ash packing performance prediction model constructed by a random forest algorithm on the uniformity of the carbon line is poor and is not suitable for the prediction of the uniformity of the carbon line.
Comparative example 4
This comparative example provides a prediction method of cigarette ash packing performance (carbon line width) based on a random forest algorithm, which is basically the same as that of example 1, except that: in the step 1, the detected ash coating performance is the width of the carbon wire; in step 2, the dependent variable is the carbon line width.
The carbon line widths of 56 modeling samples of the cigarette ash packing performance prediction model of the comparative example are internally predicted, the predicted value is compared with the actual measured value, residual errors, minimum errors, maximum errors, average absolute errors, standard deviations, linear correlations and mean square errors are calculated, and the prediction capacity of the model is verified. And finally, calculating to obtain a linear correlation fitting degree R of the internal predicted value and the measured value which is 0.879 and smaller than 0.9 required by statistics, wherein the linear correlation fitting degree R is shown that a cigarette ash packing performance prediction model constructed by a random forest algorithm has poor prediction capability on the width of the carbon line and is not suitable for the prediction of the width of the carbon line.
It is to be understood that the above examples of the present invention are provided by way of illustration only and not by way of limitation of the embodiments of the present invention. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the invention are desired to be protected by the following claims.
Claims (10)
1. A prediction method of cigarette ash packing performance based on a random forest algorithm is characterized by comprising the following steps:
s1, taking cigarettes as modeling samples, and detecting auxiliary material parameters, physical indexes and ash wrapping performance of the modeling samples;
s2, taking the auxiliary material parameters and physical indexes obtained in the step S1 as independent variables, taking the ash wrapping performance as a dependent variable, and constructing a cigarette ash wrapping performance prediction model by adopting a random forest algorithm;
s3, detecting auxiliary material parameters and physical indexes of an unknown sample, and inputting the auxiliary material parameters and physical indexes into a cigarette ash packing performance prediction model to obtain the ash packing performance of the unknown sample;
the auxiliary material parameters in the step S1 and the step S3 are tipping paper air permeability, cigarette paper phosphate content, cigarette paper oxalate content, cigarette paper citrate content, cigarette paper sodium ion content, cigarette paper potassium ion content, cigarette paper air permeability and cigarette paper quantification, and the physical indexes are filter tip air permeability, cigarette paper air permeability, total air permeability, cigarette weight, open suction resistance and closed suction resistance;
the ash coating performance in the steps S1, S2 and S3 is whiteness value, cracking rate or ash holding force.
2. The prediction method according to claim 1, wherein the number of modeling samples in step S1 is greater than 56.
3. The method according to claim 2, wherein the number of modeling samples in step S1 is 56 to 150.
4. The prediction method according to claim 1, wherein the random forest algorithm in step S2 is a random forest classification method or a random forest regression method.
5. The prediction method according to claim 1, wherein the specific process in step S2 is: taking the auxiliary material parameters and the physical indexes obtained in the step S1 as independent variables, taking the ash wrapping performance as dependent variables, introducing the auxiliary material parameters, the physical indexes and the auxiliary material indexes into an IBM SPSS model 18.0 data processing system, selecting a random forest algorithm, and setting construction parameters to finish the construction of the cigarette ash wrapping performance prediction model.
6. The prediction method of claim 5, wherein the build parameters include a number of models, a sample size, or a minimum number of sub-nodes.
7. The prediction method according to claim 6, wherein the number of models is 100 to 150, the sample size is 0.9 to 1.0, and the minimum number of sub-nodes is 1 to 5.
8. The prediction method according to claim 1, wherein after the construction of the cigarette ash packing performance prediction model in step S2, the method further comprises a step of performing internal prediction using the cigarette ash packing performance prediction model.
9. The method of claim 8, wherein the step of performing the internal prediction using the cigarette ash packing performance prediction model comprises the steps of: and adopting a cigarette ash packing performance prediction model to conduct internal prediction on the ash packing performance of the modeling sample.
10. The method according to claim 1, wherein the ash packing property in steps S1, S2 and S3 is a crack rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311154051.5A CN117272239A (en) | 2023-09-06 | 2023-09-06 | Cigarette ash packing performance prediction method based on random forest algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311154051.5A CN117272239A (en) | 2023-09-06 | 2023-09-06 | Cigarette ash packing performance prediction method based on random forest algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117272239A true CN117272239A (en) | 2023-12-22 |
Family
ID=89216951
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311154051.5A Pending CN117272239A (en) | 2023-09-06 | 2023-09-06 | Cigarette ash packing performance prediction method based on random forest algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117272239A (en) |
-
2023
- 2023-09-06 CN CN202311154051.5A patent/CN117272239A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108061691B (en) | Method for evaluating quality stability of reconstituted tobacco between batches | |
| CN107563672A (en) | A kind of method based on analytic hierarchy process (AHP) structure cigarette primary processing Environmental Evaluation Model | |
| CN110726645B (en) | Method for evaluating stability of axial density distribution of cigarette tobacco shreds | |
| CN108732127B (en) | Method for detecting mixing proportion of each component in cut tobacco | |
| CN101710073A (en) | Method for detecting physical and chemical indexes of reconstituted tobacco by near infrared spectrum detection paper making method | |
| CN105608334B (en) | A method of evaluation cigarette auxiliary material parameter influence degree | |
| CN110196265B (en) | Method for evaluating distribution stability of cut tobacco in cigarette branches in batch based on distribution probability | |
| CN106248617A (en) | Based near infrared tobacco tar detection method | |
| CN113190967B (en) | Multi-objective screening-based fine cigarette material combination method | |
| CN103892435B (en) | Method for preparing tobacco powder raw materials and stabilizing quality of reconstituted tobacco | |
| CN105628646B (en) | A kind of online tar prediction of cigarette and method for early warning | |
| CN113295813B (en) | Detection method for material liquid application uniformity | |
| CN116343944A (en) | Cigarette auxiliary material parameter and physical index and main stream smoke component influence prediction method | |
| CN104596976A (en) | Method for determining protein of paper-making reconstituted tobacco through ear infrared reflectance spectroscopy technique | |
| CN117272239A (en) | Cigarette ash packing performance prediction method based on random forest algorithm | |
| CN108982766B (en) | Method for determining dense end index of cigarette and application | |
| CN110887809B (en) | Method for measuring stem content in tobacco shreds based on near infrared spectrum technology | |
| CN110333177B (en) | Cigarette segmented suction resistance nondestructive rapid detection evaluation method based on cigarette paper air permeability stability | |
| CN107884360B (en) | Cigarette paper combustion improver detection method | |
| CN104090082A (en) | Method for forecasting pH value of full smoke of cigarette | |
| CN109540728B (en) | Method for predicting tar release amount and/or carbon monoxide release amount in cigarette | |
| CN111077103A (en) | Method for measuring content of glyceryl triacetate | |
| CN112526009A (en) | Method for measuring water content of heated cigarette core material based on water activity | |
| CN102253192B (en) | Quality detection method for tobacco leaves and application thereof | |
| CN111680260A (en) | Method for evaluating cigarette physical quality by efficacy function method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |