CN107392399A - A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA - Google Patents
A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA Download PDFInfo
- Publication number
- CN107392399A CN107392399A CN201710760920.7A CN201710760920A CN107392399A CN 107392399 A CN107392399 A CN 107392399A CN 201710760920 A CN201710760920 A CN 201710760920A CN 107392399 A CN107392399 A CN 107392399A
- Authority
- CN
- China
- Prior art keywords
- cigarette
- svm
- iaga
- value
- sensory quality
- 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 71
- 230000001953 sensory effect Effects 0.000 title claims abstract description 35
- 230000003044 adaptive effect Effects 0.000 title claims abstract description 30
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012360 testing method Methods 0.000 claims abstract description 43
- 239000000203 mixture Substances 0.000 claims abstract description 23
- 239000000126 substance Substances 0.000 claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 14
- 238000005457 optimization Methods 0.000 claims abstract description 9
- 238000010276 construction Methods 0.000 claims abstract description 4
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 36
- 229910052757 nitrogen Inorganic materials 0.000 claims description 18
- SNICXCGAKADSCV-JTQLQIEISA-N (-)-Nicotine Chemical compound CN1CCC[C@H]1C1=CC=CN=C1 SNICXCGAKADSCV-JTQLQIEISA-N 0.000 claims description 11
- 229960002715 nicotine Drugs 0.000 claims description 11
- SNICXCGAKADSCV-UHFFFAOYSA-N nicotine Natural products CN1CCCC1C1=CC=CN=C1 SNICXCGAKADSCV-UHFFFAOYSA-N 0.000 claims description 11
- 102000004169 proteins and genes Human genes 0.000 claims description 9
- 108090000623 proteins and genes Proteins 0.000 claims description 9
- 239000003513 alkali Substances 0.000 claims description 7
- 230000006978 adaptation Effects 0.000 claims description 6
- 230000035772 mutation Effects 0.000 claims description 6
- 239000003205 fragrance Substances 0.000 claims description 5
- WFZUGQHZHVPZHZ-UHFFFAOYSA-N [N].N1=CC=CC(=C1)C1N(C)CCC1 Chemical compound [N].N1=CC=CC(=C1)C1N(C)CCC1 WFZUGQHZHVPZHZ-UHFFFAOYSA-N 0.000 claims description 4
- XKMRRTOUMJRJIA-UHFFFAOYSA-N ammonia nh3 Chemical compound N.N XKMRRTOUMJRJIA-UHFFFAOYSA-N 0.000 claims description 4
- 150000008442 polyphenolic compounds Chemical class 0.000 claims description 4
- 235000013824 polyphenols Nutrition 0.000 claims description 4
- 241000208125 Nicotiana Species 0.000 abstract description 19
- 235000002637 Nicotiana tabacum Nutrition 0.000 abstract description 19
- 238000011156 evaluation Methods 0.000 abstract description 10
- 238000004458 analytical method Methods 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 39
- 238000012706 support-vector machine Methods 0.000 description 32
- 239000000779 smoke Substances 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000002068 genetic effect Effects 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 210000000697 sensory organ Anatomy 0.000 description 4
- 230000000391 smoking effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 102000008297 Nuclear Matrix-Associated Proteins Human genes 0.000 description 2
- 108010035916 Nuclear Matrix-Associated Proteins Proteins 0.000 description 2
- 230000006378 damage Effects 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 210000000299 nuclear matrix Anatomy 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 244000025254 Cannabis sativa Species 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 238000012356 Product development Methods 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 239000002585 base Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008303 genetic mechanism Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 239000002304 perfume Substances 0.000 description 1
- 230000035479 physiological effects, processes and functions Effects 0.000 description 1
- 230000010181 polygamy Effects 0.000 description 1
- 102000005912 ran GTP Binding Protein Human genes 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 235000013599 spices Nutrition 0.000 description 1
- 235000019505 tobacco product Nutrition 0.000 description 1
- 239000002023 wood Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06395—Quality analysis or management
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Physiology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Manufacturing & Machinery (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA, what is solved is the technical problem that can not meet that correct selection, the degree of accuracy of evaluation index characteristic value are low, by using including (1), according to SVM algorithm, it is Radial basis kernel function to select kernel function；(2) using improved adaptive GA-IAGA optimization Radial basis kernel function parameter, optimized parameter is obtained, is built according to optimized parameter and improves SVM algorithm model；(3) training dataset and test data set are defined, using improving SVM algorithm model construction forecasting system；(4) by the desired value input prediction system of at least two different evaluating cigarette quality, class test and combined test are carried out respectively, selects optimal index value；The desired value includes influenceing the chemical composition of Sensory Quality of Cigarette or the technical scheme of empirical value, preferably resolves the problem, available in the analysis prediction of tobacco business Sensory Quality of Cigarette.
Description
Technical field
The present invention relates to tobacco business Sensory Quality of Cigarette analysis prediction field, and in particular to one kind is based on improved genetic algorithms
The SVM Sensory Quality of Cigarette Forecasting Methodologies of method.
Background technology
Tobacco components and aesthetic quality are there is certain corresponding relation, during production of cigarettes, it is difficult to be directed to cigarette
The physical and chemical index of grass sets up effective mathematical modeling with Sensory Quality of Cigarette complex relationship, therefore in tobacco and its product
During new product development and product maintenance, mainly by brainstrust manually continuous smoking of smokeing panel test, mouthfeel, perfume (or spice) to cigarette
The evaluation criterions such as taste carry out the evaluation of cigarette quality.So far, because technical reason, most of tobacco business are main still
By smokeing panel test, expert assesses the quality of cigarette quality and the style of cigarette by sense organ, and what sensory evaluating smoking relied primarily on is to smoke panel test
The experience of smokeing panel test of expert, personal like, the physiology and mental condition of people, still one needs by long term accumulation and abundant warp for itself
The technology tested.But the assessment result of cigarette quality is often by the structure of knowledge of brainstrust, experience, mood, environment and personal like
Deng the influence of subjective factor, the reliability of assessment result is difficult to be guaranteed, and long campaigns smoke panel test work to the expert that smokes panel test
Health also have very big harm, therefore, be much directed to study tobacco quality in terms of scholars always search for cigarette
Relation between careless chemical composition and Sensory Quality of Cigarette.
To understand the problem of process subjectivity is strong, efficiency is low of smokeing panel test, the existing method using machine learning is to cigarette sense organ
Quality is evaluated, and makes every effort to the mapping ruler from a large amount of tobacco extracting data physical and chemical index and aesthetic quality, with auxiliary
Or completed instead of the tobacco expert that smokes panel test to the sense organ prediction and evaluation of new product.Following technical problem be present in prior art：First, nothing
Method meets the correct selection of evaluation index characteristic value；Second, it is unable to reach higher accuracy rate.
A kind of therefore it provides Sensory Quality of Cigarette that correct selection that disclosure satisfy that evaluation index characteristic value, accuracy rate are high
Forecasting Methodology is with regard to necessary.
The content of the invention
The technical problems to be solved by the invention are can not to meet evaluation index characteristic value just present in prior art
Really choose, the technical problem that the degree of accuracy is low.A kind of new SVM Sensory Quality of Cigarette prediction side based on improved adaptive GA-IAGA is provided
Method, the SVM Sensory Quality of Cigarette Forecasting Methodology based on improved adaptive GA-IAGA, which has, disclosure satisfy that evaluation index characteristic value just
Really choose, the characteristics of accuracy rate is high.
In order to solve the above technical problems, the technical scheme used is as follows：
A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA, the Forecasting Methodology include：
(1) according to SVM algorithm, it is Radial basis kernel function to select kernel function；
(2) using improved adaptive GA-IAGA optimization Radial basis kernel function parameter, optimized parameter is obtained, is built according to optimized parameter
Improve SVM algorithm model；
(3) training dataset and test data set are defined, using improving SVM algorithm model construction forecasting system；
(4) by the desired value input prediction system of at least two different evaluating cigarette quality, class test is carried out respectively
And combined test, select optimal index value；The desired value includes the chemical composition or empirical value for influenceing Sensory Quality of Cigarette.
The operation principle of the present invention：The present invention is calculated using algorithm of support vector machine, the i.e. SVM of improved adaptive GA-IAGA optimization
For method come the Sensory Quality of Cigarette forecast model that builds, the thought of wherein improved adaptive GA-IAGA optimization SVM algorithm is to pass through improvement
Genetic algorithm finds two optimal parameter error penalty terms and nuclear parameter.Model, Ran Houtong are built using optimized parameter
The chemical composition combination for crossing different algorithms and different influence Sensory Quality of Cigarette carries out contrast experiment, by relatively more correct
Rate, obtains accuracy highest algorithm and chemical composition, finally draws the optimal algorithm of prediction Sensory Quality of Cigarette and influences to roll up
The main chemical compositions of cigarette aesthetic quality.
Most of prior art is all as test data set, because as people are asked health using single characteristic value
The attention of topic, the quality problems of cigarette also receive much concern, and strengthen seeming to the cigarette quality of researching and analysing and improve of cigarette quality
It is so urgent and important, the single chemical composition to influence cigarette quality is accurate to what is judged as Main Basiss
Property has much room for improvement, and is subject to the summary of experience of people for a long time as test data set, can be to forecast analysis cigarette quality
There is the change of matter.Experience index according to the expert in terms of research cigarette to decision cigarette quality, including schmuck value, sugared nitrogen
Than, sugared alkali ratio, nicotine value, fragrance value and values of nitrogen might, the different content of these experience desired values also determine the quality of cigarette.Its
In, schmuck value is the ratio of sugar and protein in cigarette, and its ratio is higher to illustrate that sugar content is higher in cigarette, protein content
Relatively low, its corresponding cigarette quality class is higher, but nor explanation schmuck value is the higher the better, but have it is optimal suitable
Scope, this just needs and other content values are mutually coordinated, it is necessary to which test of many times is suitable to find one between this composition
Ratio.
Although the further investigation of the progress and researcher recently as science and technology, so far can't be complete
The quality of cigarette is represented with the content of chemical composition entirely, so needing to study influence of the main chemical composition to cigarette quality
To be cigarette composition as scientific basis and theoretical foundation.
By repeatedly to influence, the main chemical compositions of cigarette quality, empirical value are individually tested and combination carries out experiment point
Analysis, obtained accuracy rate are compared, it was concluded that standard when main chemical compositions and empirical value are combined test
True rate highest, and the value drawn more has actual application value.
In such scheme, for optimization, further, the optimized parameter is optimal penalty factor item and optimal kernel function
Parameter.
Further, included in the step (2) by improved adaptive GA-IAGA to calculate optimized parameter：
(A) training data is determined, initial population scale, maximum algebraically, mutation probability, crossover probability, fitness letter are set
Number predetermined value, initializes t=0；
(B) coding generation initial population q (t), calculates each individual adaptation degree f (t)；
(C) fitness function value of each individual adaptation degree composition is more than fitness function predetermined value, while population in population
Middle individual fitness f (t) optimum values then perform step (F) when unchanged；
(D) t=t+1, two individuals are selected to be intersected the selection p (t) i.e. from p (t-1)；
(E) p (t) is randomly choosed and carries out mutation operation, produce population of future generation, perform step (B)；
(F) optimal penalty factor item and optimal kernel functional parameter are finally obtained.
Further, punishment parameter C constant interval is in the optimal penalty factor item：1-5, optimal kernel functional parameter
The width δ constant intervals of middle Radial basis kernel function are 0.01-2.
Further, by the desired value input prediction system of two kinds of different evaluating cigarette quality in the step (4).
Further, the desired value is a kind of includes nine indexs, nine indexs be total reducing sugar, total nitrogen, nicotine, protein,
Free nicotine amount, polyphenol content, protein nitrogen quantity, ammonia nitrogen amount and nicotine nitrogen quantity.
Further, the desired value is that the experience index of cigarette quality is determined in history predictive result, including Shi Muke
Value, sugared nitrogen ratio, sugared alkali ratio, nicotine value, fragrance value and values of nitrogen might.
Further, the training dataset and test data set are the test result in historic test results, train number
It is identical with the data bulk of test data set according to collecting.
SVM is built upon the universal learning machine device on the basis of small sample theory, and it is based on Statistical Learning Theory and structure wind
Danger minimize, seek optimal compromise between model complexity and learning ability according to limited sample information, with obtain compared with
Good generalization ability.Data-oriented collectionWherein xt∈RnIt is input vector, including characteristic value and label,
yt∈RnIt is output vector, N is sample number.Input data is exactly non-linearly mapped to a higher dimensional space by SVMs, and
Linear regression is carried out using structural risk minimization on higher dimensional space, so as to solve the nonlinear regression problem of original space.
Specifically：
(a)Wherein, xt∈Rn, yt∈Rn,It is unknown mappings function, b is biased, etIt is white noise
Sound, ωTIt is weight vector；
(b) it is theoretical according to statistical machine learning, ωTDetermined, therefore can be converted into by minimizing object function with b It is the complexity of Controlling model, C is error penalty factor, RempIt is control errors letter
Number, i.e. insensitive loss function.The SVMs of standard is to be used as loss function by the use of error ε；
(3) SVMs now can be converted into object function：Constraints：This is the quadratic programming problem of a Problem with Some Constrained Conditions, introduces Lagrangian：
Wherein, at∈ R are Lagrangians, and above-mentioned formula is carried out to seek partial derivative,
Obtained after abbreviationat=Cet,Obtain：
According to Mercer theorems, kernel functionWherein K (xt,xi) it is to meet Mercer conditions
Symmetric function, i.e.,：
Establish nuclear matrix equation Ωij=K (xi,xj), therefore whole SVMs can be converted into system of linear equations：
Wherein, α=[α1,.....,αN] it is Lagrangian, y=[y1,.....,yN] it is output vector, therefore prop up
The Function Estimation for holding vector machine is：
Radial basis kernel function is one and uses function of the vector as independent variable, can calculate one based on vector distance
Scalar.The function has the advantages that non-linear, parameter is relatively fewer and may map to the high dimension in space, the SVM in the present invention
Sensory Quality of Cigarette is predicted using Radial basis kernel function：
K(x,xi)=exp (- | | x-xi||2/σ2)。
After Radial basis kernel function is determined, the model has two parameters：Error penalty term and nuclear parameter.Herein using excellent
Change improved adaptive GA-IAGA to optimize the parameter of SVM models, predictablity rate is improved with this.Optimized parameter is applied into mould
In type, training dataset is tested in different combinations, and is verified with test data set.According to more every group of survey
The accuracy of examination, obtain optimal chemical composition combination.
Beneficial effects of the present invention：
Effect one, solving the problems, such as to smoke panel test, process subjectivity is strong, efficiency is low；
Effect two, Sensory Quality of Cigarette is evaluated using the method for machine learning, from a large amount of tobacco extracting datas
Physical and chemical index and the mapping ruler of aesthetic quality, with aid in or instead of tobacco smoke panel test expert complete it is pre- to the sense organ of new product
Test and appraisal valency, assess and lay a good foundation for Sensory Quality of Cigarette artificial intelligence from now on；
Effect three：Predictablity rate is high；
Effect four：It disclosure satisfy that the correct selection of evaluation index characteristic value.
Brief description of the drawings
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1, the SVM Sensory Quality of Cigarette Forecasting Methodology schematic flow sheets based on improved adaptive GA-IAGA.
Fig. 2, improved adaptive GA-IAGA is to the model parameter searching process schematic diagram.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that specific embodiment described herein is not used to limit only to explain the present invention
The fixed present invention.
Embodiment 1
The present embodiment provides a kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA, the prediction side
Method includes：
(1) according to SVM algorithm, it is Radial basis kernel function to select kernel function；
(2) using improved adaptive GA-IAGA optimization Radial basis kernel function parameter, optimized parameter is obtained, is built according to optimized parameter
Improve SVM algorithm model；
(3) training dataset and test data set are defined, using improving SVM algorithm model construction forecasting system；
(4) by the desired value input prediction system of at least two different evaluating cigarette quality, class test is carried out respectively
And combined test, select optimal index value；The desired value includes the chemical composition or empirical value for influenceing Sensory Quality of Cigarette.
Step (1) specifically includes：The determination of SVMs (SVM) parameter：SVM is built upon on the basis of small sample theory
Universal learning machine device, it is based on Statistical Learning Theory and structural risk minimization, answered according to limited sample information in model
Seek optimal compromise between polygamy and learning ability, to obtain preferable generalization ability.Data-oriented collection
Wherein xt∈RnIt is input vector, including characteristic value and label, yt∈RnIt is output vector, N is sample number.Supporting vector
Input data is exactly non-linearly mapped to a higher dimensional space by machine, and is carried out on higher dimensional space using structural risk minimization
Linear regression, so as to solve the nonlinear regression of original space：
Wherein, xt∈Rn, yt∈Rn,It is unknown mappings function, b is biased, etIt is white noise, ωTIt is weight vector.
Theoretical, the ω according to statistical machine learningTDetermined, therefore can be converted into by minimizing object function with b
It is the complexity of Controlling model, C is error penalty factor, RempControl errors function, i.e., insensitive damage
Lose function.The SVMs of standard is to be used as loss function by the use of error ε.SVMs now can be converted into：
Object function：
Constraints：Introduce Lagrangian：
Wherein, at∈ R are Lagrangians, and above-mentioned formula is carried out to seek partial derivative,
Obtained after abbreviationat=Cet,Have：
According to Mercer theorems, kernel functionWherein K (xt,xi) it is to meet Mercer conditions
Symmetric function, bring above formula into and obtain：
Establish nuclear matrix equation Ωij=K (xi,xj), therefore whole SVMs can be converted into following linear side
Journey group：
Wherein, α=[α1,.....,αN] it is Lagrangian, y=[y1,.....,
yN] it is output vector, therefore the Function Estimation of SVMs is：
Kernel function is chosen, and Radial basis kernel function is one and uses function of the vector as independent variable, can be based on to span
From calculating a scalar.The function has the advantages that non-linear, parameter is relatively fewer and may map to the high dimension in space, this
The Sensory Quality of Cigarette of embodiment SVMs is predicted using Radial basis kernel function：K(x,xi)=exp (- | | x-xi|
|2/σ2)；
Specifically, the optimized parameter is optimal penalty factor item and optimal kernel functional parameter.
Specifically, according to improved adaptive GA-IAGA come optimizing parameter, improved adaptive GA-IAGA is improved in genetic algorithm,
Genetic algorithm is the random search algorithm of the natural selection and natural genetic mechanism during a kind of reference biological evolution, is had very
Strong ability of searching optimum, and adaptively command deployment process, in the hope of optimal., should after Radial basis kernel function is determined
Model has two parameters：Error penalty term and nuclear parameter.The present embodiment is carried out using improved adaptive GA-IAGA to the parameter of SVM models
Optimization, predictablity rate is improved with this.Included in the step (2) by improved adaptive GA-IAGA to calculate optimized parameter：
(A) training data is determined, initial population scale, maximum algebraically, mutation probability, crossover probability, fitness letter are set
Number predetermined value, initializes t=0；
(B) coding generation initial population q (t), calculates each individual adaptation degree f (t)；
(C) fitness function value of each individual adaptation degree composition is more than fitness function predetermined value, while population in population
Middle individual fitness f (t) optimum values then perform step (F) when unchanged；
(D) t=t+1, two individuals are selected to be intersected the selection p (t) i.e. from p (t-1)；
(E) p (t) is randomly choosed and carries out mutation operation, produce population of future generation, perform step (B)；
(F) optimal penalty factor item and optimal kernel functional parameter are finally obtained.
Specifically, step (4) is by the desired value input prediction system of two kinds of different evaluating cigarette quality in the present embodiment
System.
A kind of in 2 kinds of desired values to include nine indexs, nine indexs are total reducing sugar, total nitrogen, nicotine, protein, free cigarette
Alkali number, polyphenol content, protein nitrogen quantity, ammonia nitrogen amount and nicotine nitrogen quantity.
Another desired value is that the experience index of cigarette quality is determined in history predictive result, including schmuck value, sugared nitrogen
Than, sugared alkali ratio, nicotine value, fragrance value and values of nitrogen might.
The present embodiment is according to the detection of the tobacco product chemical composition of nearly 3 years of certain tobacco group and aesthetic quality's smoking result one
Series data is adjusted 800 groups of data for being appropriate for experiment, wherein 400 groups are used as training dataset, 400 groups are in addition
Test data set.
The present embodiment carries out class test and combined test respectively using the index of two kinds of different evaluation qualities of tobacco.
Primary sources are to extract nine main indexs that quality of tobacco is determined in tobacco leaf chemical composition, respectively total reducing sugar, total nitrogen, cigarette
Alkali, protein, free nicotine amount, polyphenol content, protein nitrogen quantity, ammonia nitrogen amount and nicotine nitrogen quantity.Conventional according to tobacco company
Smoke panel test the result that expert is smoked panel test to different tobacco leaves, the chemical composition of corresponding different tobacco leaves is collected, as training
And test data set.Second class is the experience index to decision quality of tobacco according to the expert in terms of research tobacco leaf, including applies wood
Gram value, sugared nitrogen ratio, sugared alkali ratio, nicotine value, fragrance value and values of nitrogen might, different expert's smoking result value corresponding to its every group of data,
Using a data part for collection as training dataset, another part is as test data set.
Wherein, punishment parameter C constant interval is in optimal penalty factor item：1-5, radial direction base in optimal kernel functional parameter
The width δ constant intervals of kernel function are 0.01-2.
By the accuracy rate of contrast and experiment, to determine to predict that the selection of the experimental data of Sensory Quality of Cigarette is main
When chemical composition and experience indicator combination are as the one 15 assemblage characteristic vector tieed up, obtained accuracy rate highest is predicted.
Although the illustrative embodiment of the present invention is described above, in order to the technology of the art
Personnel are it will be appreciated that the present invention, but the present invention is not limited only to the scope of embodiment, to the common skill of the art
For art personnel, as long as long as in various change spirit and scope of the invention, all are equal using the innovation and creation of present inventive concept
In the row of protection.
Claims (8)
- A kind of 1. SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA, it is characterised in that：The Forecasting Methodology bag Include：(1) according to SVM algorithm, it is Radial basis kernel function to select kernel function；(2) using improved adaptive GA-IAGA optimization Radial basis kernel function parameter, optimized parameter is obtained, is built and improved according to optimized parameter SVM algorithm model；(3) training dataset and test data set are defined, using improving SVM algorithm model construction forecasting system；(4) by the desired value input prediction system of at least two different evaluating cigarette quality, class test and group are carried out respectively Test is closed, selects optimal index value；The desired value includes the chemical composition or empirical value for influenceing Sensory Quality of Cigarette.
- 2. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 1 based on improved adaptive GA-IAGA, its feature exist In：The optimized parameter is optimal penalty factor item and optimal kernel functional parameter.
- 3. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 2 based on improved adaptive GA-IAGA, its feature exist In：It is described to be included using improved adaptive GA-IAGA optimization Radial basis kernel function parameter：(A) training data is determined, sets initial population scale, maximum algebraically, mutation probability, crossover probability, fitness function pre- Definite value, initialize t=0；(B) coding generation initial population q (t), calculates each individual adaptation degree f (t)；(C) fitness function value of each individual adaptation degree composition is more than fitness function predetermined value in population, while individual in population Step (F) is then performed when fitness f (t) optimum values of body are unchanged；(D) t=t+1, two individuals are selected to be intersected the selection p (t) i.e. from p (t-1)；(E) p (t) is randomly choosed and carries out mutation operation, produce population of future generation, perform step (B)；(F) optimal penalty factor item and optimal kernel functional parameter are finally obtained.
- 4. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 3 based on improved adaptive GA-IAGA, its feature exist In：Punishment parameter C constant interval is in the optimal penalty factor item：1-5, Radial basis kernel function in optimal kernel functional parameter Width δ constant intervals be 0.01-2.
- 5. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 1 based on improved adaptive GA-IAGA, its feature exist In：The desired value of the different evaluating cigarette quality is two kinds.
- 6. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 5 based on improved adaptive GA-IAGA, its feature exist In：The desired value is a kind of to include nine indexs, and nine indexs are total reducing sugar, total nitrogen, nicotine, protein, free nicotine amount, polyphenol Content, protein nitrogen quantity, ammonia nitrogen amount and nicotine nitrogen quantity.
- 7. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 5 based on improved adaptive GA-IAGA, its feature exist In：The desired value is that the experience index of cigarette quality is determined in history predictive result, including schmuck value, sugared nitrogen ratio, sugared alkali Than, nicotine value, fragrance value and values of nitrogen might.
- 8. the SVM Sensory Quality of Cigarette Forecasting Methodologies according to claim 1 based on improved adaptive GA-IAGA, its feature exist In：The training dataset and test data set are the test result in historic test results, training dataset and test data The data bulk of collection is identical.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710760920.7A CN107392399A (en) | 2017-08-30 | 2017-08-30 | A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710760920.7A CN107392399A (en) | 2017-08-30 | 2017-08-30 | A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107392399A true CN107392399A (en) | 2017-11-24 |
Family
ID=60347904
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710760920.7A Pending CN107392399A (en) | 2017-08-30 | 2017-08-30 | A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107392399A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858714A (en) * | 2019-04-03 | 2019-06-07 | 杭州安脉盛智能技术有限公司 | Based on pipe tobacco quality inspection index prediction technique, the apparatus and system for improving neural network |
| CN112465013A (en) * | 2020-11-25 | 2021-03-09 | 平安科技（深圳）有限公司 | Method, device and equipment for predicting peak value of underwater slamming force of ocean flat-bottom structure |
| CN112906910A (en) * | 2019-12-04 | 2021-06-04 | 北京沃东天骏信息技术有限公司 | Test method, system, device and electronic equipment |
| CN113488113A (en) * | 2021-07-12 | 2021-10-08 | 浙江中烟工业有限责任公司 | Industrial use value identification method of redried strip tobacco |
| CN113609775A (en) * | 2021-08-09 | 2021-11-05 | 上海太太乐食品有限公司 | Quantitative forecasting method for salivation feeling sensory evaluation value of freshness of solid compound seasoning |
| CN113903458A (en) * | 2021-10-26 | 2022-01-07 | 北京大学第三医院（北京大学第三临床医学院） | Acute kidney injury early prediction method and device |
| CN114242183A (en) * | 2021-12-20 | 2022-03-25 | 甘肃烟草工业有限责任公司 | Cigarette smoke H value prediction model construction method based on tobacco chemistry convention |
| CN114401211A (en) * | 2022-01-17 | 2022-04-26 | 重庆邮电大学 | Test system and test method for accessing industrial wireless network equipment to IPv6 network |
| CN115437333A (en) * | 2022-11-07 | 2022-12-06 | 杭州安脉盛智能技术有限公司 | Sensory quality-based adjusting method, device, equipment and storage medium |
| CN116227974A (en) * | 2022-12-26 | 2023-06-06 | 中国农业科学院蜜蜂研究所 | Identification method for honey sensory and quality ratings |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1975705A (en) * | 2005-11-28 | 2007-06-06 | 颐中烟草(集团)有限公司 | Cigarette internal quality index extimating method based on regression function estimating SVM |
| RU2428083C1 (en) * | 2009-12-28 | 2011-09-10 | Государственное научное учреждение Всероссийский научно-исследовательский институт табака, махорки и табачных изделий Российской академии сельскохозяйственных наук (ГНУ ВНИИТТИ Россельхозакадемии) | Method for estimate of tobacco smoke inhalation effort degree and cigarette quality control |
| CN103217408A (en) * | 2013-05-14 | 2013-07-24 | 贵州大学 | Method for identifying two flue-cured tobaccos with different odor types in Guizhou |
| CN106295686A (en) * | 2016-08-01 | 2017-01-04 | 安徽农业大学 | The tea aroma sorting technique of parameter optimization support vector machine |
-
2017
- 2017-08-30 CN CN201710760920.7A patent/CN107392399A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1975705A (en) * | 2005-11-28 | 2007-06-06 | 颐中烟草(集团)有限公司 | Cigarette internal quality index extimating method based on regression function estimating SVM |
| RU2428083C1 (en) * | 2009-12-28 | 2011-09-10 | Государственное научное учреждение Всероссийский научно-исследовательский институт табака, махорки и табачных изделий Российской академии сельскохозяйственных наук (ГНУ ВНИИТТИ Россельхозакадемии) | Method for estimate of tobacco smoke inhalation effort degree and cigarette quality control |
| CN103217408A (en) * | 2013-05-14 | 2013-07-24 | 贵州大学 | Method for identifying two flue-cured tobaccos with different odor types in Guizhou |
| CN106295686A (en) * | 2016-08-01 | 2017-01-04 | 安徽农业大学 | The tea aroma sorting technique of parameter optimization support vector machine |
Non-Patent Citations (2)
| Title |
|---|
| 张喆: "烟叶主要化学成分与等级品质关系的研究", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
| 张圣男: "基于支持向量机的卷烟烟气指标预测", 《中国优秀硕士学位论文全文数据库工程科技Ⅰ辑》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858714A (en) * | 2019-04-03 | 2019-06-07 | 杭州安脉盛智能技术有限公司 | Based on pipe tobacco quality inspection index prediction technique, the apparatus and system for improving neural network |
| CN112906910A (en) * | 2019-12-04 | 2021-06-04 | 北京沃东天骏信息技术有限公司 | Test method, system, device and electronic equipment |
| CN112465013B (en) * | 2020-11-25 | 2024-02-02 | 平安科技（深圳）有限公司 | Peak value prediction method, device and equipment for water slamming force of ocean flat-bottom structure |
| CN112465013A (en) * | 2020-11-25 | 2021-03-09 | 平安科技（深圳）有限公司 | Method, device and equipment for predicting peak value of underwater slamming force of ocean flat-bottom structure |
| CN113488113A (en) * | 2021-07-12 | 2021-10-08 | 浙江中烟工业有限责任公司 | Industrial use value identification method of redried strip tobacco |
| CN113488113B (en) * | 2021-07-12 | 2024-02-23 | 浙江中烟工业有限责任公司 | Industrial use value identification method for redried strip tobacco |
| CN113609775A (en) * | 2021-08-09 | 2021-11-05 | 上海太太乐食品有限公司 | Quantitative forecasting method for salivation feeling sensory evaluation value of freshness of solid compound seasoning |
| CN113609775B (en) * | 2021-08-09 | 2024-04-23 | 上海太太乐食品有限公司 | Quantitative forecasting method for delicious salivation sense sensory evaluation score of solid compound seasoning |
| CN113903458A (en) * | 2021-10-26 | 2022-01-07 | 北京大学第三医院（北京大学第三临床医学院） | Acute kidney injury early prediction method and device |
| CN114242183A (en) * | 2021-12-20 | 2022-03-25 | 甘肃烟草工业有限责任公司 | Cigarette smoke H value prediction model construction method based on tobacco chemistry convention |
| CN114401211A (en) * | 2022-01-17 | 2022-04-26 | 重庆邮电大学 | Test system and test method for accessing industrial wireless network equipment to IPv6 network |
| CN114401211B (en) * | 2022-01-17 | 2023-05-12 | 重庆邮电大学 | Test system and test method for accessing industrial wireless network equipment to IPv6 network |
| CN115437333A (en) * | 2022-11-07 | 2022-12-06 | 杭州安脉盛智能技术有限公司 | Sensory quality-based adjusting method, device, equipment and storage medium |
| CN115437333B (en) * | 2022-11-07 | 2023-02-28 | 杭州安脉盛智能技术有限公司 | Sensory quality-based adjusting method, device, equipment and storage medium |
| CN116227974B (en) * | 2022-12-26 | 2024-01-30 | 中国农业科学院蜜蜂研究所 | Identification method for honey sensory and quality ratings |
| CN116227974A (en) * | 2022-12-26 | 2023-06-06 | 中国农业科学院蜜蜂研究所 | Identification method for honey sensory and quality ratings |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107392399A (en) | A kind of SVM Sensory Quality of Cigarette Forecasting Methodologies based on improved adaptive GA-IAGA | |
| Elhorst et al. | Spatial panel data models | |
| Dell et al. | What do we learn from the weather? The new climate-economy literature | |
| Heiri et al. | A chironomid-based Holocene summer air temperature reconstruction from the Swiss Alps | |
| Bodner et al. | A statistical approach to root system classification | |
| Wolf et al. | Forest biomass allometry in global land surface models | |
| CN107731304A (en) | A kind of prediction of height method and system | |
| CN109034960B (en) | Multi-attribute inference method based on user node embedding | |
| Pede | Diversity and regional economic growth: evidence from US counties | |
| CN108257675A (en) | Chronic obstructive pulmonary disease onset risk Forecasting Methodology, server and computer readable storage medium | |
| CN101419209A (en) | Cigarette sensing appraise and flume index immune neural net prediction method | |
| CN109002917A (en) | Total output of grain multidimensional time-series prediction technique based on LSTM neural network | |
| Bleaney et al. | Income inequality and growth—does the relationship vary with the income level? | |
| CN106776930A (en) | A kind of location recommendation method for incorporating time and geographical location information | |
| Chang | Simulation estimation of two‐tiered dynamic panel Tobit models with an application to the labor supply of married women | |
| Aladag et al. | Improving weighted information criterion by using optimization | |
| Galinato et al. | The short-run and long-run effects of corruption control and political stability on forest cover | |
| CN109472318A (en) | For the method and device of the machine learning model selected characteristic of building | |
| CN109583635A (en) | A kind of short-term load forecasting modeling method towards operational reliability | |
| CN107766573A (en) | Method of Commodity Recommendation, device, equipment and storage medium based on data processing | |
| Lian et al. | Mutual reinforcement of academic performance prediction and library book recommendation | |
| Tejera et al. | Description and codification of Miscanthus× giganteus growth stages for phenological assessment | |
| Martinez et al. | A study of Mexican free-tailed bat chirp syllables: Bayesian functional mixed models for nonstationary acoustic time series | |
| CN111340132A (en) | Machine olfaction mode identification method based on DA-SVM | |
| Kleshchenko et al. | Estimation of winter wheat yield using the principal component analysis based on the integration of satellite and ground information |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171124 |