CN105954451A - Rapid cigarette type distinguishing method based on electronic nose full chromatographic data - Google Patents

Rapid cigarette type distinguishing method based on electronic nose full chromatographic data Download PDF

Info

Publication number
CN105954451A
CN105954451A CN201610398897.7A CN201610398897A CN105954451A CN 105954451 A CN105954451 A CN 105954451A CN 201610398897 A CN201610398897 A CN 201610398897A CN 105954451 A CN105954451 A CN 105954451A
Authority
CN
China
Prior art keywords
cigarette
brand
odor type
vector machine
class
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.)
Granted
Application number
CN201610398897.7A
Other languages
Chinese (zh)
Other versions
CN105954451B (en
Inventor
吴君章
韩冰
陈翠玲
汪军霞
古君平
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China Tobacco Guangdong Industrial Co Ltd
Original Assignee
China Tobacco Guangdong Industrial Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China Tobacco Guangdong Industrial Co Ltd filed Critical China Tobacco Guangdong Industrial Co Ltd
Priority to CN201610398897.7A priority Critical patent/CN105954451B/en
Publication of CN105954451A publication Critical patent/CN105954451A/en
Application granted granted Critical
Publication of CN105954451B publication Critical patent/CN105954451B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N30/00—Investigating or analysing materials by separation into components using adsorption, absorption or similar phenomena or using ion-exchange, e.g. chromatography or field flow fractionation
    • G01N30/02—Column chromatography

Landscapes

  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
  • Manufacturing Of Cigar And Cigarette Tobacco (AREA)

Abstract

The invention relates to a rapid cigarette type distinguishing method based on electronic nose full chromatographic data. According to the method, through the tiny difference between full chromatographic data of cigarettes of different types, a support vector machine model is established for distinguishing. The experiment proves that the accuracy of the distinguishing result of the method is high, and the requirements of cigarette distinguishing can be met.

Description

Medicated cigarette type Quick methods based on Electronic Nose whole chromatogram data
Technical field
The present invention relates to Nicotiana tabacum L. detection field, more particularly, to a kind of Medicated cigarette classes based on Electronic Nose whole chromatogram data Type Quick method.
Background technology
The research of Electronic Nose starts from the phase early 1980s, and it is by the volatility gas having in object Body, Organic substance etc. trap, and then the data trapped are carried out statistical analysis.It uses a series of sensor to imitate Olfactory sensation, it is possible to detect and distinguish the abnormal smells from the patient of complex sample, has advantage with low cost, widely used.Nearly ten years, gone out Show a large amount of use Electronic Nose and carried out the report of every research, be included in the application during tobacco quality is evaluated.End 2008, entirely The supplier of world's Electronic Nose commercial product has reached 18, such as France Alpha MOS, the Cyrano sciences of the U.S. Deng.Most of Electronic Nose are sensor type, and unfortunately, the Electronic Nose of sensor type is susceptible to intoxicating phenomenon.
The HERACLES Flash GC type Electronic Nose that France Alpha MOS produces, its built-in Trap, can be greatly improved Detection sensitivity；Using post sheath heating technique, heating rate reaches as high as 25 DEG C/s；Use the principle of gas chromatogram, configure two Chromatographic column and two fid detectors that root polarity is different gather data, and its post footpath is 0.1mm, has high theoretical tray Number.
Summary of the invention
The present invention solves a difficult problem for above prior art, it is provided that a kind of Medicated cigarette classes based on Electronic Nose whole chromatogram data Type Quick method, the method is utilized the fine difference between the whole chromatogram data of dissimilar Medicated cigarette, is supported by structure Vector machine model makes a distinction differentiation.It is demonstrated experimentally that the accuracy rate that the method that the present invention provides differentiates is higher, it is possible to meet volume The requirement that cigarette differentiates.
For realizing above goal of the invention, the technical scheme is that
A kind of Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data, is used for differentiating whether Medicated cigarette belongs to certain One brand, then on the basis of differentiating that Medicated cigarette belongs to a certain brand, differentiates the valency class belonging to Medicated cigarette and odor type, described Method of discrimination comprises the following steps:
One, brand judges the stage
S11. the Medicated cigarette of a certain brand is referred to as A brand cigarette, uses Electronic Nose to gather A brand cigarette, other brands volume The whole chromatogram data of cigarette, and the whole chromatogram data of collection are equally divided into k part, making (k-1) part is training data, and 1 part is test Data；
S12. supporting vector machine model is built:
f ( x ) = sgn { Σ i = 1 m α i * y i K ( x i · x ) + b * } - - - ( 1 )
In formula (1), yiRepresenting sample class, sample class includes positive sample and negative sample, wherein the value of positive sample be+ 1, the value of negative sample is-1；K(xiX) representing kernel function, m represents total sample number, xiBeing the vector representing sample, x represents all The average vector of sample, α* iAnd b*For model parameter, sgn is output function；
The most then (k-1) part training data is used successively supporting vector machine model to be trained, the mistake being trained The whole chromatogram data of Cheng Zhong, A brand cigarette as positive sample, the whole chromatogram data of other brand cigarettes as negative sample,；
S14. use the supporting vector machine model trained that the cigarette brand belonging to test data is differentiated, it is judged that its Whether belong to A brand cigarette, and export differentiation result；
The most described step S13, S14 repeat k time, during performing at k time, and k part whole chromatogram data conduct successively Test data, remaining whole chromatogram data is then as training data；
S16. based on the accuracy rate carrying out differentiating for k time, the parameter of supporting vector machine model is optimized；
S17. input the whole chromatogram data of Medicated cigarette to be discriminated, then use the supporting vector machine model through optimizing to judge Whether it belongs to A brand: if the output valve of supporting vector machine model is+1, the Medicated cigarette representing to be discriminated belongs to A brand, then perform Step S21 and step S4；If the output valve of supporting vector machine model is-1, then it represents that Medicated cigarette to be discriminated belongs to other brands, Now perform step S4；
Two, to the differentiation of valency class belonging to Medicated cigarette to be discriminated
S21. set A brand cigarette and include m kind valency class, respectively numbered valency class 1, valency class 2, valency class 3 ...., valency class m, make The whole chromatogram data of the A brand cigarette of various valency class are gathered respectively by Electronic Nose；
S22. the whole chromatogram data using the A brand cigarette of m kind valency class build according to the method for step S11～S16 to be supported Vector machine model, with the A of valency class i during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of brand cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of its partial valence class are negative sample, and i's is initial Value is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If support vector machine mould Type output+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of valency class i, now performs step S31 and step S4；If supporting Vector machine model output-1, then it represents that Medicated cigarette to be discriminated is not belonging to the A brand cigarette of valency class i, now performs step S23；
S23. i=i+1 repeated execution of steps S22 are made；
Three, to the differentiation of odor type belonging to Medicated cigarette to be discriminated
S31. set A brand cigarette and include n kind odor type, respectively numbered odor type 1, odor type 2, odor type 3 ... odor type n, use Electronic Nose gathers the whole chromatogram data of the A brand cigarette of various odor type respectively；
S32. the whole chromatogram data using the A brand cigarette of n kind odor type build according to the method for step S11～S16 to be supported Vector machine model, with the A of odor type p during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of brand cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of remaining odor type are negative sample, and p's is initial Value is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If support vector machine mould Type output+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of odor type p, performs step S4；If supporting vector machine model is defeated Go out-1, then it represents that Medicated cigarette to be discriminated is not belonging to the A brand cigarette of odor type p, now performs step S23；
S33. p=p+1 repeated execution of steps S32 are made；
S4. output differentiates result.
In such scheme, by build supporting vector machine model and utilize the whole chromatogram data of Medicated cigarette to be trained and Test so that the parameter of supporting vector machine model can be with various brand cigarettes, valency class Medicated cigarette, the whole chromatogram data of odor type Medicated cigarette Set up corresponding contact, closer in the practical situation of Medicated cigarette, the therefore accuracy rate of the differentiation result of the method that the present invention provides Higher, it is possible to meet the requirement that Medicated cigarette differentiates.
Preferably, described A brand cigarette includes three kinds of valency classes, and respectively one class, two classes and three classes, then to volume to be discriminated The process that valency class belonging to cigarette differentiates is specific as follows:
S101. use Electronic Nose gather respectively a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes complete Chromatographic data；
S102. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step The method of rapid S11～S16 builds supporting vector machine model, wherein according to the method for step S13 to the support vector machine built It is positive sample with the whole chromatogram data of a class A brand cigarette during model training, two class A brand cigarettes, three class A brands volumes The whole chromatogram data of cigarette are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram number of Medicated cigarette to be discriminated According to；If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to a class A brand cigarette, now perform step S31 and Step S4；If supporting vector machine model output-1, then it represents that Medicated cigarette to be discriminated belongs to two class A brand cigarettes or three class A brands volume Cigarette, now performs step S103；
S103. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step The method of rapid S11～S16 builds supporting vector machine model, wherein according to the method for step S13 to the support vector machine built It is positive sample with the whole chromatogram data of two class A brand cigarettes during model training, a class A brand cigarette, three class A brands volumes The whole chromatogram data of cigarette are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram number of Medicated cigarette to be discriminated According to；If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to two class A brand cigarettes, now perform step S31 and Step S4；If supporting vector machine model output-1, then it represents that Medicated cigarette to be discriminated belongs to three class A brand cigarettes, now performs step S104；
S104. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step The method of rapid S11～S16 builds supporting vector machine model, wherein according to the method for step S13 to the support vector machine built It is positive sample with the whole chromatogram data of three class A brand cigarettes during model training, a class A brand cigarette, two class A brands volumes The whole chromatogram data of cigarette are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram number of Medicated cigarette to be discriminated According to；If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to three class A brand cigarettes, now perform step S31 and Step S4.
Preferably, A brand cigarette has five kinds of odor types, and respectively I odor type, II odor type, III odor type, IV odor type and V are fragrant Type, then the detailed process differentiated odor type belonging to Medicated cigarette to be discriminated is as follows:
S201. Electronic Nose is used to gather the whole chromatogram data gathering five kinds of odor type A brand cigarettes respectively respectively；
S202. I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand volume are used Cigarette, the whole chromatogram data of V odor type A brand cigarette build supporting vector machine model according to the method for step S11～S16, wherein press According to step S13 method to during the supporting vector machine model training built with the whole chromatogram number of I odor type A brand cigarette According to for positive sample, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette Whole chromatogram data are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated； If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to I odor type, now performs step S4；If support vector machine Model output-1, then it represents that Medicated cigarette to be discriminated belongs to II odor type, III odor type, IV odor type or V odor type, now performs step S203；
S203. I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand volume are used Cigarette, the whole chromatogram data of V odor type A brand cigarette build supporting vector machine model according to the method for step S11～S16, wherein press According to step S13 method to during the supporting vector machine model training built with the whole chromatogram of II odor type A brand cigarette Data are positive sample, I odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette Whole chromatogram data are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated； If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to II odor type, now performs step S4；If support vector machine Model output-1, then it represents that Medicated cigarette to be discriminated belongs to III odor type, IV odor type or V odor type, now performs step S204；
S204. I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand volume are used Cigarette, the whole chromatogram data of V odor type A brand cigarette build supporting vector machine model according to the method for step S11～S16, wherein press According to step S13 method to during the supporting vector machine model training built with the whole chromatogram of III odor type A brand cigarette Data are positive sample, I odor type A brand cigarette, II odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette Whole chromatogram data are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated； If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to III odor type, now performs step S4；If support vector machine Model output-1, then it represents that Medicated cigarette to be discriminated belongs to IV odor type or V odor type, now performs step S205；
S205. I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand volume are used Cigarette, the whole chromatogram data of V odor type A brand cigarette build supporting vector machine model according to the method for step S11～S16, wherein press According to step S13 method to during the supporting vector machine model training built with the whole chromatogram of IV odor type A brand cigarette Data are positive sample, I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, V odor type A brand cigarette Whole chromatogram data are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated； If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to IV odor type, now performs step S4；If support vector machine Model output-1, then it represents that Medicated cigarette to be discriminated belongs to V odor type, now performs step S206；
S206. I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand volume are used Cigarette, the whole chromatogram data of V odor type A brand cigarette build supporting vector machine model according to the method for step S11～S16, wherein press According to step S13 method to during the supporting vector machine model training built with the whole chromatogram of V odor type A brand cigarette Data are positive sample, I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette Whole chromatogram data are negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated； If supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to V odor type, now performs step S4.
Preferably, in described step S11, S21, S31, after collecting whole chromatogram data, need whole chromatogram data are entered Row normalized.
Preferably, the procedural representation of described normalized is as follows:
x i ( j ) n e w = x i ( j ) - x ( j ) min x ( j ) m a x - x ( j ) min
Wherein xiJ () represents the jth chromatographic data value of the whole chromatogram data of Medicated cigarette type i, described Medicated cigarette type includes Brand, valency class or odor type, x (j)minWith x (j)maxRepresent jth chromatographic data in the whole chromatogram data of Medicated cigarette type i respectively Minima and maximum.
Preferably, described supporting vector machine model runs in Matlab environment.
Preferably, described step S11, S21, S31 use Flash GC type Electronic Nose to gather the whole chromatogram data of Medicated cigarette.Excellent Selection of land, described Flash GC type Electronic Nose is when gathering Medicated cigarette whole chromatogram data, and it is as follows that it arranges condition: initial temperature: 45 DEG C；Eventually Temperature: 270 DEG C；Programming rate: 2 DEG C of s-1；Sample introduction needle scavenging period: 90s；Injector temperature: 210 DEG C；FID temperature: 260 DEG C； TRAP initial temperature: 40 DEG C；TRAP final temperature: 260 DEG C；Sample introduction needle temperature: 110 DEG C；Incubation temperature: 90 DEG C；Brooding time: 1200s；Catch Collection time: 50s；Sample size: 500 μ L.
Compared with prior art, the invention has the beneficial effects as follows:
The method that the present invention provides utilizes the fine difference between the whole chromatogram data of dissimilar Medicated cigarette, props up by building Hold vector machine model to make a distinction differentiation.It is demonstrated experimentally that the accuracy rate of the differentiation result of the method for present invention offer is higher, energy Enough meet the requirement that Medicated cigarette differentiates.
Accompanying drawing explanation
Fig. 1 is the flow chart of Medicated cigarette method of discrimination.
Fig. 2 is ROC and PRC curve chart.
Fig. 3 is the visual effect figure that A brand cigarette differentiates with non-A brand cigarette.
Fig. 4 is the differentiation procedure chart to A brand cigarette odor type.
Fig. 5 is the visual effect figure differentiating A brand cigarette odor type.
Fig. 6 is the flow chart differentiating each valency class Medicated cigarette of A brand board.
The visual effect figure that I odor type A brand three class Medicated cigarette is differentiated by Fig. 7.
Fig. 8 is the visual effect figure differentiating III odor type one class and three class A brand cigarettes.
Fig. 9 is the visual effect figure differentiating two kinds of Medicated cigarette of IV odor type A16 and A19.
Figure 10 is the visual effect figure differentiating V odor type one class and two class A brand cigarettes.
Detailed description of the invention
Accompanying drawing being merely cited for property explanation, it is impossible to be interpreted as the restriction to this patent；
Below in conjunction with drawings and Examples, the present invention is further elaborated.
Embodiment 1
The Medicated cigarette type Quick method that the present invention provides is used for differentiating whether Medicated cigarette belongs to a certain brand, is then sentencing On the basis of other Medicated cigarette belongs to a certain brand, the valency class belonging to Medicated cigarette and odor type are differentiated, comprise the following steps:
One, brand judges the stage
S11. the Medicated cigarette of a certain brand is referred to as A brand cigarette, uses Electronic Nose to gather A brand cigarette, other brands volume The whole chromatogram data of cigarette, and the whole chromatogram data of collection are equally divided into k part, making (k-1) part is training data, and 1 part is test Data；
S12. supporting vector machine model is built:
f ( x ) = s g n { Σ i = 1 m α i * y i K ( x i · x ) + b * } - - - ( 1 )
In formula (1), yiRepresenting sample class, sample class includes positive sample and negative sample, wherein the value of positive sample be+ 1, the value of negative sample is-1；K(xiX) representing kernel function, m represents total sample number, xiBeing the vector representing sample, x represents all The average vector of sample, α* iAnd b*For model parameter, sgn is output function；
The most then (k-1) part training data is used successively supporting vector machine model to be trained, the mistake being trained The whole chromatogram data of Cheng Zhong, A brand cigarette as positive sample, the whole chromatogram data of other brand cigarettes as negative sample,；
S14. use the supporting vector machine model trained that the cigarette brand belonging to test data is differentiated, it is judged that its Whether belong to A brand cigarette, and export differentiation result；
The most described step S13, S14 repeat k time, during performing at k time, and k part whole chromatogram data conduct successively Test data, remaining whole chromatogram data is then as training data；
S16. based on the accuracy rate carrying out differentiating for k time, the parameter of supporting vector machine model is optimized；
S17. input the whole chromatogram data of Medicated cigarette to be discriminated, then use the supporting vector machine model through optimizing to judge Whether it belongs to A brand: if the output valve of supporting vector machine model is+1, the Medicated cigarette representing to be discriminated belongs to A brand, then perform Step S21 and step S4；If the output valve of supporting vector machine model is-1, then it represents that Medicated cigarette to be discriminated belongs to other brands, Now perform step S4；
Two, to the differentiation of valency class belonging to Medicated cigarette to be discriminated
S21. set A brand cigarette and include m kind valency class, respectively numbered valency class 1, valency class 2, valency class 3 ...., valency class m, make The whole chromatogram data of the A brand cigarette of various valency class are gathered respectively by Electronic Nose；
S22. the whole chromatogram data using the A brand cigarette of m kind valency class build according to the method for step S11～S16 to be supported Vector machine model, with the A of valency class i during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of brand cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of its partial valence class are negative sample, and i's is initial Value is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If support vector machine mould Type output+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of valency class i, now performs step S31 and step S4；If supporting Vector machine model output-1, then it represents that Medicated cigarette to be discriminated is not belonging to the A brand cigarette of valency class i, now performs step S23；
S23. i=i+1 repeated execution of steps S22 are made；
Three, to the differentiation of odor type belonging to Medicated cigarette to be discriminated
S31. set A brand cigarette and include n kind odor type, respectively numbered odor type 1, odor type 2, odor type 3 ... odor type n, use Electronic Nose gathers the whole chromatogram data of the A brand cigarette of various odor type respectively；
S32. the whole chromatogram data using the A brand cigarette of n kind odor type build according to the method for step S11～S16 to be supported Vector machine model, with the A of odor type p during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of brand cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of remaining odor type are negative sample, and p's is initial Value is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If support vector machine mould Type output+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of odor type p, performs step S4；If supporting vector machine model is defeated Go out-1, then it represents that Medicated cigarette to be discriminated is not belonging to the A brand cigarette of odor type p, now performs step S23；
S33. p=p+1 repeated execution of steps S32 are made；
S4. output differentiates result.
Embodiment 2
The present embodiment is on the basis of embodiment 1, and the Medicated cigarette of selected a collection of different brands, valency class and odor type has been carried out specifically Experiment.The cigarette brand that this experiment uses includes A brand, B brand, C brand, D brand, E brand, F brand and G product Board.And carrying out valency class, odor type is when judging, mainly uses the different price class of A brand, the Medicated cigarette of odor type, A brand cigarette bag Having included three kinds of valency classes, be respectively divided into a class, two classes and three classes, in the present embodiment, a class, two classes, three class A brand cigarettes are respectively It is expressed as A1, A2, A3；A brand cigarette includes five kinds of odor types, respectively I odor type, II odor type, III odor type, IV odor type and V Odor type.This experiment is to A brand and non-A brand cigarette, the Medicated cigarette of the various odor type of A brand, the Medicated cigarette of A brand various valency class, I perfume Type A brand one class, two classes and three class Medicated cigarette, II odor type A brand cigarette, III odor type one class and three class A brand cigarettes, IV odor type Two kinds of Medicated cigarette of A16 and A19, V odor type one class and two class A brand cigarettes all carry out concrete discriminating experiment.
This experiment uses Flash GC type Electronic Nose to gather the whole chromatogram data of Medicated cigarette, setting of Flash GC type Electronic Nose Put condition as follows: initial temperature: 45 DEG C；Final temperature: 270 DEG C；Programming rate: 2 DEG C of s-1；Sample introduction needle scavenging period: 90s；Injection port temperature Degree: 210 DEG C；FID temperature: 260 DEG C；TRAP initial temperature: 40 DEG C；TRAP final temperature: 260 DEG C；Sample introduction needle temperature: 110 DEG C；Hatching temperature Degree: 90 DEG C；Brooding time: 1200s；Trap time: 50s；Sample size: 500 μ L.
Below for the concrete condition of experiment:
(1) A brand and the differentiation of non-A brand cigarette
Parameter such as table 1 institutes such as the degree of accuracy that structure supporting vector machine model A non-to A brand brand cigarette differentiates Show, it is thus achieved that the degree of accuracy of 93.28%, the sensitivity of 96.18%, the specificity of 83.64%, the accuracy rate of 95.12% and The geneva correlation coefficient of 0.8089.ROC and PRC curve is as in figure 2 it is shown, area is respectively 0.9865 and 0.9952.Result shows, The supporting vector machine model set up can differentiate A brand and non-A brand cigarette well.A brand cigarette and non-A brand cigarette The visual effect differentiated is as shown in Figure 3.
The accuracy rate that table 1A brand-non-A brand cigarette differentiates
(2) differentiation of odor type Medicated cigarette various to A brand
Differentiation process to A brand cigarette odor type, the most as shown in Figure 4.Build supporting vector machine model to A brand cigarette I, the differentiation degree of accuracy of II, III, IV, V odor type is as shown in table 2.92.95%, 89.78% differentiate degree of accuracy respectively:, 100%, 57.50% and 71.67%.Show that supporting vector machine model can differentiate the A brand of I, II, III, V odor type well Medicated cigarette.Wherein IV odor type A brand cigarette number of samples is 120, and wherein 51 samples are supported vector machine model and are determined as V perfume Type A brand cigarette.V odor type A brand cigarette number of samples is 180, and wherein 51 samples are identified as IV odor type A brand cigarette. This is owing to IV odor type and V odor type odor type are closer to, and causes chromatographic data the most similar, it is desirable to complete for this two class Complete the most more difficulty.Visual effect based on the differentiation of A brand cigarette odor type is as shown in Figure 5.
Table 2A brand cigarette odor type differentiates result
(3) differentiation of valency class Medicated cigarette various to A brand
1) differentiation of valency class Medicated cigarette various to A brand
The present embodiment, based on the flow chart shown in Fig. 6, is classified research to each valency class of A brand cigarette.Such as table 3 institute Show, supporting vector machine model to A brand cigarette one class and two classes, three classes, two classes and a class, three classes and three classes and a class, three The differentiation degree of accuracy of class price is 100%, shows that supporting vector machine model can very well differentiate the A brand of different price class Medicated cigarette.
Table 3 different price class differentiates result
2) to A brand one class, two classes and the differentiation of three class Medicated cigarette
The present embodiment has carried out concrete differentiation to each trade mark Medicated cigarette in A brand one class Medicated cigarette, it determines result such as table 4 Shown in.A11, A12, A13, A14, A15, A16, A17, A18, A19 are a class Medicated cigarette, and its accuracy differentiated is all 100%, Show that the method that the present invention provides can differentiate a class A brand cigarette well.
Table 4 one class A brand cigarette differentiates result
The present embodiment uses supporting vector machine model to be differentiated each trade mark Medicated cigarette of A brand two class Medicated cigarette, A21, A22 and A23 are A brand two class Medicated cigarette, and it differentiates that result is as shown in table 5.The differentiation accuracy of A21, A22 and A23 is all It is 100%, shows that the supporting vector machine model built can be good at distinguishing each trade mark Medicated cigarette of A brand two class Medicated cigarette.
Table 5 two class A brand cigarette differentiates result
The present embodiment uses supporting vector machine model to be differentiated each trade mark Medicated cigarette of A brand three class Medicated cigarette, A31, A32, A34, A35, A36, A37, A38, A39, A310 are A brand three class Medicated cigarette, it determines the confusion matrix of result such as table 6 Shown in.The differentiation accuracy of A31, A32, A34, A35, A36, A37, A38, A39, A310 is 100% respectively, and the differentiation of A33 Accuracy is 98.33%.Show each trade mark Medicated cigarette that the method proposed can extraordinary distinguish in A brand three class Medicated cigarette.
The differentiation result of table 6 three class A brand cigarette
A31/ sample number 60 0 0 0 0 0 0 0 0 0
A32/ sample number 0 60 0 0 0 0 0 0 0 0
A33/ sample number 0 1 59 0 0 0 0 0 0 0
A34/ sample number 0 0 0 60 0 0 0 0 0 0
A35/ sample number 0 0 0 0 60 0 0 0 0 0
A36/ sample number 0 0 0 0 0 60 0 0 0 0
A37/ sample number 0 0 0 0 0 0 30 0 0 0
A38/ sample number 0 0 0 0 0 0 0 30 0 0
A39/ sample number 0 0 0 0 0 0 0 0 60 0
A310/ sample number 0 0 0 0 0 0 0 0 0 30
A31 A32 A33 A34 A35 A36 A37 A38 A39 A310
3) differentiation to I odor type A brand cigarette
The differentiation degree of accuracy of I odor type A brand one class, two classes and three classes is all by the supporting vector machine model that the present invention provides 100%, show that the method proposed can be good at distinguishing I odor type three class A brand cigarette.And it is pre-to I odor type A brand three class Medicated cigarette The visual effect surveyed is as shown in Figure 7.
4) differentiation to II odor type A brand cigarette
II odor type A brand cigarette has also been carried out differentiating research by the present invention, and A31, A34, A35 and A36 are II odor type A product Board Medicated cigarette, it determines the confusion matrix of result is as shown in table 7.Supporting vector machine model is accurate to the differentiation of A31, A34, A35 and A36 Rate is all 100%.For A33, only one sample is erroneously identified as A34, it was predicted that precision is 98.33%.These results indicate that Supporting vector machine model can be good at identifying II odor type A brand cigarette.
The differentiation result of table 7 II odor type A brand cigarette
A31/ sample number 60 0 0 0 0
A33/ sample number 0 59 1 0 0
A34/ sample number 0 0 60 0 0
A35/ sample number 0 0 0 60 0
A36/ sample number 0 0 0 0 60
A31 A33 A34 A35 A36
5) to III odor type one class and the differentiation of three class A brand cigarettes
In the present embodiment, also having carried out III odor type one class and three class A brand cigarettes differentiating research, its visual effect figure is such as Shown in Fig. 8, and differentiate that result is as shown in table 8.The supporting vector machine model prediction essence to III odor type one class and three class A brand cigarettes Degree, Sensitivity and Specificity etc. are all 100.0%, and geneva correlation coefficient is 1.00.Show that supporting vector machine model can be accurate Ground identifies III odor type A brand one class and three class Medicated cigarette.
Table 8 III odor type one class and three class A brand cigarettes differentiate result
6) differentiation to IV two kinds of Medicated cigarette of odor type A16 and A19
In the present embodiment, two kinds of Medicated cigarette of IV odor type A16 and A19 have been carried out differentiating research by supporting vector machine model, and it is sentenced The visual effect of other result is as shown in Figure 9.It can be seen that supporting vector machine model can accurately identify IV odor type A16 and A19 Two class brand cigarettes, indicate the effectiveness of current method.
7) to V odor type one class and the differentiation of two class A brand cigarettes
In the present embodiment, V odor type one class and two class A brand cigarettes have been carried out differentiating research by supporting vector machine model, its Differentiate the visual effect of result as shown in Figure 10.It can be seen that supporting vector machine model can accurately identify V odor type one class With two class Medicated cigarette, indicate the effectiveness of current method.
Differentiation result from each form above and figure is it can be seen that use whole chromatogram data to be obtained in that higher Predict the outcome, this is because each type Medicated cigarette is respectively provided with the characteristic fingerprint chromatograph of oneself.Chromatographic peak area is used with other Method compare, use whole chromatogram data can catch the fine difference of chromatographic data between dissimilar Medicated cigarette, it is possible to Produce higher differentiation result.
Obviously, the above embodiment of the present invention is only for clearly demonstrating example of the present invention, and is not right The restriction of embodiments of the present invention.For those of ordinary skill in the field, the most also may be used To make other changes in different forms.Here without also cannot all of embodiment be given exhaustive.All at this Any amendment, equivalent and the improvement etc. made within the spirit of invention and principle, should be included in the claims in the present invention Protection domain within.

Claims (8)

1. Medicated cigarette type Quick methods based on Electronic Nose whole chromatogram data, are used for differentiating whether Medicated cigarette belongs to a certain Brand, then on the basis of differentiating that Medicated cigarette belongs to a certain brand, differentiates the valency class belonging to Medicated cigarette and odor type, its feature It is: described method of discrimination comprises the following steps:
One, brand judges the stage
S11. the Medicated cigarette of a certain brand is referred to as A brand cigarette, uses Electronic Nose to gather A brand cigarette, other brand cigarettes Whole chromatogram data, and the whole chromatogram data of collection are equally divided into k part, making (k-1) part is training data, and 1 part is test data；
S12. supporting vector machine model is built:
f ( x ) = sgn { Σ i = 1 m α i * y i K ( x i · x ) + b * } - - - ( 1 )
In formula (1), yiRepresenting sample class, sample class includes positive sample and negative sample, and wherein the value of positive sample is+1, negative The value of sample is-1；K(xiX) representing kernel function, m represents total sample number, xiBeing the vector representing sample, x represents all samples Average vector, α* iAnd b*For model parameter, sgn is output function；
The most then (k-1) part training data is used successively supporting vector machine model to be trained, during being trained, The whole chromatogram data of A brand cigarette as positive sample, the whole chromatogram data of other brand cigarettes as negative sample,；
S14. use the supporting vector machine model trained that the cigarette brand belonging to test data is differentiated, it is judged that whether it Belong to A brand cigarette, and export differentiation result；
The most described step S13, S14 repeat k time, and during performing at k time, k part whole chromatogram data are successively as test Data, remaining whole chromatogram data is then as training data；
S16. based on the accuracy rate carrying out differentiating for k time, the parameter of supporting vector machine model is optimized；
S17. input the whole chromatogram data of Medicated cigarette to be discriminated, then use and judge that it is through the supporting vector machine model optimized The no A brand that belongs to: if the output valve of supporting vector machine model is+1, the Medicated cigarette representing to be discriminated belongs to A brand, then perform step S21 and step S4；If the output valve of supporting vector machine model is-1, then it represents that Medicated cigarette to be discriminated belongs to other brands, now Perform step S4；
Two, to the differentiation of valency class belonging to Medicated cigarette to be discriminated
S21. set A brand cigarette and include m kind valency class, respectively numbered valency class 1, valency class 2, valency class 3 ...., valency class m, make electricity consumption Sub-nose gathers the whole chromatogram data of the A brand cigarette of various valency class respectively；
S22. the whole chromatogram data using the A brand cigarette of m kind valency class build according to the method for step S11～S16 supports vector Machine model, with the A brand of valency class i during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of Medicated cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of its partial valence class are negative sample, and the initial value of i is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If supporting vector machine model is defeated Go out+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of valency class i, now performs step S31 and step S4；If supporting vector Machine model output-1, then it represents that Medicated cigarette to be discriminated is not belonging to the A brand cigarette of valency class i, now performs step S23；
S23. i=i+1 repeated execution of steps S22 are made；
Three, to the differentiation of odor type belonging to Medicated cigarette to be discriminated
S31. set A brand cigarette and include n kind odor type, respectively numbered odor type 1, odor type 2, odor type 3 ... odor type n, use electronics Nose gathers the whole chromatogram data of the A brand cigarette of various odor type respectively；
S32. the whole chromatogram data using the A brand cigarette of n kind odor type build according to the method for step S11～S16 supports vector Machine model, with the A brand of odor type p during wherein training the supporting vector machine model of structure according to the method for step S13 The whole chromatogram data of Medicated cigarette are positive sample, and the whole chromatogram data of the A brand cigarette of remaining odor type are negative sample, and the initial value of p is 1；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If supporting vector machine model is defeated Go out+1, then it represents that Medicated cigarette to be discriminated belongs to the A brand cigarette of odor type p, perform step S4；If supporting vector machine model output-1, The Medicated cigarette then representing to be discriminated is not belonging to the A brand cigarette of odor type p, now performs step S23；
S33. p=p+1 repeated execution of steps S32 are made；
S4. output differentiates result.
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 1, its feature exists In: described A brand cigarette includes three kinds of valency classes, respectively one class, two classes and three classes, then sentence valency class belonging to Medicated cigarette to be discriminated Other process is specific as follows:
S101. Electronic Nose is used to gather a class A brand cigarette, two class A brand cigarettes, the whole chromatogram of three class A brand cigarettes respectively Data；
S102. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step S11 ～the method for S16 builds supporting vector machine model, wherein instruct to the supporting vector machine model built according to the method for step S13 Be positive sample with the whole chromatogram data of a class A brand cigarette during white silk, two class A brand cigarettes, three class A brand cigarettes complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to a class A brand cigarette, now performs step S31 and step S4；If supporting vector machine model output-1, then it represents that Medicated cigarette to be discriminated belongs to two class A brand cigarettes or three class A brand cigarettes, this Shi Zhihang step S103；
S103. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step S11 ～the method for S16 builds supporting vector machine model, wherein instruct to the supporting vector machine model built according to the method for step S13 Be positive sample with the whole chromatogram data of two class A brand cigarettes during white silk, a class A brand cigarette, three class A brand cigarettes complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to two class A brand cigarettes, now performs step S31 and step S4；If supporting vector machine model output-1, then it represents that Medicated cigarette to be discriminated belongs to three class A brand cigarettes, now performs step S104；
S104. use a class A brand cigarette, two class A brand cigarettes, three class A brand cigarettes whole chromatogram data according to step S11 ～the method for S16 builds supporting vector machine model, wherein instruct to the supporting vector machine model built according to the method for step S13 Be positive sample with the whole chromatogram data of three class A brand cigarettes during white silk, a class A brand cigarette, two class A brand cigarettes complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to three class A brand cigarettes, now performs step S31 and step S4。
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 2, its feature exists In: A brand cigarette has five kinds of odor types, and respectively I odor type, II odor type, III odor type, IV odor type and V odor type, then to be discriminated The detailed process that odor type belonging to Medicated cigarette differentiates is as follows:
S201. Electronic Nose is used to gather the whole chromatogram data gathering five kinds of odor type A brand cigarettes respectively respectively；
S202. use I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, The whole chromatogram data of V odor type A brand cigarette according to step S11～S16 method build supporting vector machine model, wherein according to The method of step S13 to during the supporting vector machine model training built with the whole chromatogram data of I odor type A brand cigarette For positive sample, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to I odor type, now performs step S4；If support vector machine mould Type output-1, then it represents that Medicated cigarette to be discriminated belongs to II odor type, III odor type, IV odor type or V odor type, now performs step S203；
S203. use I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, The whole chromatogram data of V odor type A brand cigarette according to step S11～S16 method build supporting vector machine model, wherein according to The method of step S13 to during the supporting vector machine model training built with the whole chromatogram number of II odor type A brand cigarette According to for positive sample, I odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to II odor type, now performs step S4；If support vector machine mould Type output-1, then it represents that Medicated cigarette to be discriminated belongs to III odor type, IV odor type or V odor type, now performs step S204；
S204. use I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, The whole chromatogram data of V odor type A brand cigarette according to step S11～S16 method build supporting vector machine model, wherein according to The method of step S13 to during the supporting vector machine model training built with the whole chromatogram number of III odor type A brand cigarette According to for positive sample, I odor type A brand cigarette, II odor type A brand cigarette, IV odor type A brand cigarette, V odor type A brand cigarette complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to III odor type, now performs step S4；If support vector machine mould Type output-1, then it represents that Medicated cigarette to be discriminated belongs to IV odor type or V odor type, now performs step S205；
S205. use I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, The whole chromatogram data of V odor type A brand cigarette according to step S11～S16 method build supporting vector machine model, wherein according to The method of step S13 to during the supporting vector machine model training built with the whole chromatogram number of IV odor type A brand cigarette According to for positive sample, I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, V odor type A brand cigarette complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to IV odor type, now performs step S4；If support vector machine mould Type output-1, then it represents that Medicated cigarette to be discriminated belongs to V odor type, now performs step S206；
S206. use I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette, The whole chromatogram data of V odor type A brand cigarette according to step S11～S16 method build supporting vector machine model, wherein according to The method of step S13 to during the supporting vector machine model training built with the whole chromatogram number of V odor type A brand cigarette According to for positive sample, I odor type A brand cigarette, II odor type A brand cigarette, III odor type A brand cigarette, IV odor type A brand cigarette complete Chromatographic data is negative sample；Then the supporting vector machine model after optimizing inputs the whole chromatogram data of Medicated cigarette to be discriminated；If Supporting vector machine model output+1, then it represents that Medicated cigarette to be discriminated belongs to V odor type, now performs step S4.
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 1, its feature exists In: in described step S11, S21, S31, after collecting whole chromatogram data, need whole chromatogram data are normalized.
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 4, its feature exists In: the procedural representation of described normalized is as follows:
x i ( j ) n e w = x i ( j ) - x ( j ) min x ( j ) m a x - x ( j ) min
Wherein xiJ () represents the jth chromatographic data value of whole chromatogram data of Medicated cigarette type i, described Medicated cigarette type include brand, Valency class or odor type, x (j)minWith x (j)maxRepresent the minima of jth chromatographic data in the whole chromatogram data of Medicated cigarette type i respectively And maximum.
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 1, its feature exists In: described supporting vector machine model runs in Matlab environment.
7. according to the Medicated cigarette type Quick methods based on Electronic Nose whole chromatogram data described in any one of claim 1～6, It is characterized in that: described step S11, S21, S31 use Flash GC type Electronic Nose to gather the whole chromatogram data of Medicated cigarette.
Medicated cigarette type Quick method based on Electronic Nose whole chromatogram data the most according to claim 6, its feature exists In: described Flash GC type Electronic Nose is when gathering Medicated cigarette whole chromatogram data, and it is as follows that it arranges condition: initial temperature: 45 DEG C；Final temperature: 270℃；Programming rate: 2 DEG C of s-1；Sample introduction needle scavenging period: 90s；Injector temperature: 210 DEG C；FID temperature: 260 DEG C；TRAP Initial temperature: 40 DEG C；TRAP final temperature: 260 DEG C；Sample introduction needle temperature: 110 DEG C；Incubation temperature: 90 DEG C；Brooding time: 1200s；During trapping Between: 50s；Sample size: 500 μ L.
CN201610398897.7A 2016-06-06 2016-06-06 Cigarette type Quick method based on the panchromatic modal data of electronic nose Active CN105954451B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610398897.7A CN105954451B (en) 2016-06-06 2016-06-06 Cigarette type Quick method based on the panchromatic modal data of electronic nose

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610398897.7A CN105954451B (en) 2016-06-06 2016-06-06 Cigarette type Quick method based on the panchromatic modal data of electronic nose

Publications (2)

Publication Number Publication Date
CN105954451A true CN105954451A (en) 2016-09-21
CN105954451B CN105954451B (en) 2017-12-26

Family

ID=56907950

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610398897.7A Active CN105954451B (en) 2016-06-06 2016-06-06 Cigarette type Quick method based on the panchromatic modal data of electronic nose

Country Status (1)

Country Link
CN (1) CN105954451B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107807199A (en) * 2017-10-31 2018-03-16 华东理工大学 A kind of electronic nose instrument and tobacco and tobacco product aesthetic quality's detection method
CN107843695A (en) * 2017-10-31 2018-03-27 华东理工大学 Tobacco and the electronic nose instrument evaluation method of tobacco product aesthetic quality
CN109781870A (en) * 2017-11-13 2019-05-21 贵州中烟工业有限责任公司 A method of identifying flue-cured tobacco flavor types

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007002677A2 (en) * 2005-06-24 2007-01-04 The Board Of Trustees Of The Leland Stanford Junior University Methods and compositions for diagnosis and monitoring of atherosclerotic cardiovascular disease
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
CN104634744A (en) * 2014-12-19 2015-05-20 云南省烟草公司昆明市公司 Method for measuring tobacco leaf maturity based on reflection spectrum
CN104820077A (en) * 2015-05-29 2015-08-05 广东中烟工业有限责任公司 Method and device for recognizing cigarette

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007002677A2 (en) * 2005-06-24 2007-01-04 The Board Of Trustees Of The Leland Stanford Junior University Methods and compositions for diagnosis and monitoring of atherosclerotic cardiovascular disease
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
CN104634744A (en) * 2014-12-19 2015-05-20 云南省烟草公司昆明市公司 Method for measuring tobacco leaf maturity based on reflection spectrum
CN104820077A (en) * 2015-05-29 2015-08-05 广东中烟工业有限责任公司 Method and device for recognizing cigarette

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
K.BRUDZEWSKI等: "Differential electronic nose and support vector machine for fast recognition of tobacco", 《EXPERT SYSTEMS WITH APPILICATIONS》 *
SCHOLKOPF,B., &SMOLA,A: "《Learning with kernels》", 28 February 2006, CAMBRIDGE, MA:MIT PRESS *
吴君章等: "电子鼻在烟草行业中的研究与应用进展", 《分析测试学报》 *
吴君章等: "电子鼻***预处理方法和模式识别技术概述", 《广东化工》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107807199A (en) * 2017-10-31 2018-03-16 华东理工大学 A kind of electronic nose instrument and tobacco and tobacco product aesthetic quality's detection method
CN107843695A (en) * 2017-10-31 2018-03-27 华东理工大学 Tobacco and the electronic nose instrument evaluation method of tobacco product aesthetic quality
CN107843695B (en) * 2017-10-31 2018-08-28 华东理工大学 The electronic nose instrument evaluation method of tobacco and tobacco product aesthetic quality
CN109781870A (en) * 2017-11-13 2019-05-21 贵州中烟工业有限责任公司 A method of identifying flue-cured tobacco flavor types

Also Published As

Publication number Publication date
CN105954451B (en) 2017-12-26

Similar Documents

Publication Publication Date Title
Olsen et al. The anatomy of single cell mass cytometry data
CN105954451A (en) Rapid cigarette type distinguishing method based on electronic nose full chromatographic data
CN109540978A (en) Odor identification equipment
Oliva et al. Next-generation of instrumental odour monitoring system (IOMS) for the gaseous emissions control in complex industrial plants
JP2020089363A (en) Apparatus and method for studying effect of land change on fish diversity based on edna
CN110109543B (en) c-VEP identification method based on tested migration
Macias et al. Detection of piperonal emitted from polymer controlled odor mimic permeation systems utilizing Canis familiaris and solid phase microextraction–ion mobility spectrometry
Huang et al. Scent profiling of Cymbidium ensifolium by electronic nose
CN103558354A (en) Water toxicity analysis method based on biologic omics integrated technology
CN112466399A (en) Method for predicting mutagenicity of chemicals through machine learning algorithm
Liu et al. DeepSniffer: A meta-learning-based chemiresistive odor sensor for recognition and classification of aroma oils
CN105651868A (en) A method of screening a marker of renal toxicity caused by aristolochic acid by utilizing cell metabolic profiling in vitro
CN101110095B (en) Method for batch detecting susceptibility gene of common brain disease
Zeng et al. MSTracer: A Machine Learning Software Tool for Peptide Feature Detection from Liquid Chromatography–Mass Spectrometry Data
CN105738503A (en) Method for quickly predicting fatty acid content of walnuts based on electronic nose
CN110349633B (en) Method for screening radiation biomarkers and predicting radiation dose based on radiation response biological pathway
CN102798683B (en) Universal total-component quantitative analysis method of gas chromatography-mass spectrometry
CN102854237A (en) On-site intelligent fast real-time detector of residue of beta-adrenalin receptor agonist adopting ion mobility spectrometry
CN102636588A (en) Method for discriminating white spirits by using quartz crystal oscillator electronic nose
CN104792898A (en) Quality analysis method for tobacco essence and spice
Sweeney et al. Considerations for the inclusion of metabarcoding data in the plant protection product risk assessment of the soil microbiome
CN106124700A (en) A kind of band is from the Electronic Nose non-targeted interference Gas Distinguishing Method expressed
Ong et al. Inferring regulatory networks from time series expression data and relational data via inductive logic programming
US20240079083A1 (en) Using bipartite networks to determine interactions between analytes and chemical treatments
Kit et al. E-nose herbs recognition system based on Artificial Neural Network technique

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant