CN108133313A - A kind of artificial intelligence sensory evaluation flavour of food products system and its construction method - Google Patents
A kind of artificial intelligence sensory evaluation flavour of food products system and its construction method Download PDFInfo
- Publication number
- CN108133313A CN108133313A CN201711358141.0A CN201711358141A CN108133313A CN 108133313 A CN108133313 A CN 108133313A CN 201711358141 A CN201711358141 A CN 201711358141A CN 108133313 A CN108133313 A CN 108133313A
- Authority
- CN
- China
- Prior art keywords
- data
- food
- artificial
- evaluation
- flavour
- 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
Links
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
- G06Q10/06—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
-
- 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
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- Pathology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Analytical Chemistry (AREA)
- Immunology (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
- Fats And Perfumes (AREA)
Abstract
The present invention provides a kind of artificial intelligence sensory evaluation flavour of food products system and its construction method, including ingredient organoleptic detection system, computer intelligence analysis system and food smell database, the ingredient organoleptic detection system is made of GC MS and GC O, GC MS are used to obtain the odor characteristics initial data of food, and GC O are used to obtain the artificial sense evaluation data of food smell；For being pre-processed and feature extraction to odor characteristics initial data and artificial sense evaluation data, dimensionality reduction, cluster analysis are carried out to initial data for the computer intelligence analysis system；The food smell database is the odor characteristics initial data that is obtained by computer to organoleptic detection system and artificial sense evaluation data are fitted analysis, the smell initial data of foundation and manually evaluates the food volatile ingredient that data are combined and is mutually related Various types of data.Instrumental Analysis and human body evaluation are combined into one by the present invention, and the drawbacks of not only overcoming the one-sidedness of existing flavour of food products detection method and need artificial repeatability detection, but also evaluation result is more comprehensively reliable.
Description
Technical field
It is particularly a kind of to be used to evaluate based on artificial intelligence structure the present invention relates to a kind of system for evaluating flavour of food products
The system of flavour of food products.
Background technology
As the source and power of food industry original innovation, flavour of food products（Smell）It is the important spy for reflecting food quality
Sign, quick, thoroughly evaluating and its material base and changing rule are worth furtheing investigate.In China, especially Guangdong Province, sense organ
Flavor evaluation has long history, particularly production, processing and the stream in the tradition such as white wine, tealeaves, tobacco hobby property food
Application in siphunculus reason is particularly extensive.The main experienced three stages of flavor sensory evaluation technique, i.e.,：Starting is judged from manager；
Main body, multi-crossed disciplines and application, sensory evaluation activity standard are judged by professional sensory evaluation group；Organoleptic analysis and physics and chemistry
Analysis is combined, apparatus measures ancillary sensory evaluation.And after entering 21st century, with information technology, life science, instrument
The development of analytical technology, sense organ science and technology and multiple subject crossings show as the development trend of man-machine integration.Wherein, it is electric
The sensory sensor technology of the simulation people such as sub- nose, electronic tongues is developed." a kind of smell as disclosed in Chinese patent
Simulator and olfactory analog method "（02111936.5）And " portable electronic pen and preparation method thereof " provides for simulation people
Sensory sensor technology.
It cannot be good however, either artificial sense is judged, electronic nose of people's sense organ etc. is still simulated in Instrumental Analysis
Meet the needs of development of food industry.Although artificial sense evaluation can carry out thoroughly evaluating to flavour of food products, with food
Product production scale, Extensive Development only carry out artificial sense evaluation by professional, it is difficult to reach the mesh of control product quality
, while personnel psychology and physiological effect are judged in artificial sense evaluation, human error can hardly be avoided, and sensory evaluation can not obtain
Knowing influences the material base of flavour of food products；Although the assessment techniques such as chemical analysis and Instrumental Analysis can be to the main allusion quotation in food
Type flavor characteristic substance accurate quantitative analysis, but be difficult to carry out comprehensive flavor evaluation as evaluating professional；Electronic nose
Although can analyze the smell of special component, the development of sensor technology is limited to, to micro constitutent, complicated
Smell is difficult to make satisfied analysis and evaluation, lacks self study and comprehensive analysis ability.
In recent years, the application development of artificial intelligence technology is swift and violent, the Alpha Go that especially google companies release, into one
Step pushes the application of artificial intelligence technology to peak, and every profession and trade introduces artificial intelligence technology one after another.The technology of Alpha Go is abundant
Embody the fast development of the artificial intelligence technologys with self study feature such as depth learning technology, enhancing learning art.It faces
This trend if the self study technology that field introduces artificial intelligence can be evaluated in flavour of food products, preferably merges multi-source, multi-modal
Data, including artificial sense evaluation data, Instrumental Analysis data and image data etc., by the comprehensive of artificial sense evaluation with
The accuracy of Instrumental Analysis is combined, and has not only overcome the limitation of the electronic nose and electronic tongues based on sensor technology, but also promoted
The objectivity of artificial sense evaluation and comprehensive, novel self study " artificial intelligence sense organ " system of structure evaluation flavour of food products
System, this is not only the inexorable trend of technology development and application demand, will also have important scientific value and application prospect.
Invention content
The purpose of the present invention is to provide a kind of artificial intelligence sensory evaluation flavour of food products system, use analytical instrument and
Human body evaluates flavour of food products, and evaluation data are deep in living things feature recognition and physical properties of food analysis field by combined calculation machine
Degree study, forms the autonomous evaluation model of food smell, builds artificial intelligence sensory evaluation flavour of food products（Smell）System, to
Evaluate the flavor of volatile food.
To achieve the above object, embodiment of the present invention is：A kind of artificial intelligence sensory evaluation flavour of food products system, packet
Ingredient organoleptic detection system, computer intelligence analysis system and food smell database are included,
The ingredient organoleptic detection system is made of GC-MS and GC-O, and GC-MS is used to obtain the odor characteristics initial data of food
（The chromatography and mass spectrometric data of each volatile component content of food and chemical constitution）, GC-O is used to obtain the artificial sense of food smell
Official evaluates data（Odour intensity and odor characteristics）；
The computer intelligence analysis system, for the odor characteristics initial data and artificial sense obtained to organoleptic detection system
Evaluation data are pre-processed and feature extraction, and dimensionality reduction, cluster analysis are carried out to data；
The food smell database is that the odor characteristics obtained to organoleptic detection system by Artificial intelligent algorithm are original
Data and artificial sense evaluation data are fitted analysis, reject error dot, find association and the rule of data, the smell of foundation
The food volatile ingredient that initial data and artificial evaluation data are combined is mutually related Various types of data.
As specific embodiment, the computer intelligence analysis system uses multi-modal, multitask deep learning frame
Structure M3TDN, it is described it is multi-modal refer to the multi-modal of data, including chromatographic data, mass spectrometric data and food appearance image data, institute
Multiple targets that multitask refers to by while learns are stated, including flavor type, flavor intensity, flavor amplitude.
As specific embodiment, the food smell database includes the Chromatographic information of volatile ingredient, mass spectrum
Information, odor characteristic, aroma strength, substance image.
The present invention also provides a kind of construction method of above system, including：
（1）Food samples are entered into GC, after being detached via capillary column, outflow component is split valve and is divided into two-way, enters all the way
Chemical detector（FID or MS）, obtain the odor characteristics initial data being made of chromatography and mass spectrum initial data（Instant food is respectively waved
The chromatography and mass spectrometric data of hair property component content and chemical constitution）, another way enters sniff mouth by dedicated transfer line
（O）, obtain artificial sense evaluation data；
（2）Data are evaluated to odor characteristics initial data and artificial sense by computer software（Content, aroma strength and fragrance
Type etc.）Analysis is fitted, rejects error dot, association and the rule of data are found, so as to establish odor characteristics initial data
The food smell database being combined with artificial sense evaluation data；
（3）Odor characteristics initial data in food smell database and artificial sense evaluation data are normalized and quantified
Processing, obtains higher-dimension flavor data, and be limited Boltzmann machine using depth carries out feature extraction to the higher-dimension flavor data；
（4）Higher-dimension multi-source characteristic after feature extraction is trained, builds multi-modal, multitask deep learning framework
M3TDN on the basis of feature extraction, is respectively adopted multilayer LSTM and multilayer is limited Boltzmann machine training scoring model, from
And establish the artificial intelligence flavor evaluation system of food.
As specific embodiment, step（1）In, the method that the artificial evaluation data obtain, which is that sample enters, smells
Mouth is visited, manual knob electronic signal record is carried out while being smelt by people's nose news, record data result forms spectrum by computer disposal
Figure or transitional information data.
Alternatively embodiment, step（3）In, the method for the feature extraction can also be, by food data
The odor characteristics initial data obtained by the use of machine decision, artificial interpretation mode and artificial sense in library evaluate data as feature
The input data of processing carries out feature-extraction analysis using LDA, PCA study classic algorithm.
The verification of the present invention（It uses）Method is to select in food smell database and passed through after the sample pretreatment of same type
Chromatography and mass spectrum initial data are obtained after crossing GC-MS Instrumental Analysis detection, chromatography and mass spectrum initial data are imported the present invention's
Artificial intelligence flavor evaluation system, system will automatically generate flavor evaluation to the food.
Instrumental Analysis and human body evaluation are combined into one by the present invention, obtain more scientific and effective evaluation method, both
It can reflect direct feel of the human body to flavour of food products, and embody the advantage that analytical instrument obtains accurate detection data.It is not only
The drawbacks of overcoming the one-sidedness of existing flavour of food products detection method and needing artificial repeatability detection, and evaluation result is more complete
Face is reliable.
Description of the drawings
Fig. 1 is artificial intelligence sensory evaluation flavour of food products system construction drawing provided by the invention；
Fig. 2 is research approach flow diagram of the present invention；
Fig. 3 is the relationship of the GC spectrograms peak area and ethanol content in the embodiment of the present invention 3；
Fig. 4 is the scoring of ethanol solution fragrance and relation with contents figure in the embodiment of the present invention 3.
Specific embodiment
Drawings and examples are given below, the present invention will be described in detail.
Embodiment 1：Artificial intelligence sensory evaluation flavour of food products system and its and construction method
As shown in Figure 1, a kind of artificial intelligence sensory evaluation flavour of food products system, by ingredient organoleptic detection system, computer intelligence
Analysis system and food smell database composition, the ingredient organoleptic detection system is by GC-MS（Gaschromatographic mass spectrometry detector）
And GC-O（Gas-chromatography olfactometry instrument）Composition, GC-MS are used to obtain the gas that food samples are made of chromatography and mass spectrometric data
Taste feature initial data, GC-O are used to obtain the smell artificial sense evaluation data of food samples；The computer intelligence analysis
System, odor characteristics initial data and artificial sense the evaluation data for being obtained to organoleptic detection system carry out pretreatment and spy
Sign extraction carries out dimensionality reduction, cluster analysis to initial data；The food smell database is by Artificial intelligent algorithm pair
Odor characteristics initial data and artificial sense the evaluation data that organoleptic detection system obtains are fitted analysis, reject error dot,
Find association and the rule of data, the food volatile ingredient phase that the smell initial data of foundation and artificial evaluation data are combined
The Various types of data of mutual correlation, Chromatographic information, Information in Mass Spectra, odor characteristic, aroma strength including food samples volatile ingredient,
Substance image of food samples etc..
The construction method of above-mentioned artificial intelligence sensory evaluation flavour of food products system is as follows：
1st, food smell database is established.
The odour component in raw-food material is carried out using suitable pre-treatment means and extraction separation method pre-treatment and
Enrichment；Establish the GC-MS-O systems of smell artificial sense evaluation（Gas chromatography-mass spectrum-olfactometry system）：I.e. by sample into
Enter GC, after being detached via capillary column, outflow component is split valve and is divided into two-way, all the way into chemical detector（FID or MS）,
Schemed by TIC（Molecular ion peak）And mass spectrogram（Fragment ion peak）Each volatile ingredient is qualitatively judged, then passes through standard items
Internal standard method is precisely quantitative to volatile ingredient, obtains evaluation result of the analytical instrument to food smell.By compound characteristic collection of illustrative plates
Each substance chromatography and mass spectrum initial data, i.e. odor characteristics initial data are converted to, using the gas as computer to predetermined substance
Taste identifies data；Another way enters sniff mouth by dedicated transfer line, by profession hear fragrant teacher to the volatility isolated into
Divide and carry out smelling detection, be carried out at the same time manual knob electronic signal record, record the odor characteristic and aroma strength of each ingredient, obtain
Obtain artificial evaluation data of the human body to food smell.By computer software to the content of tie substance, aroma strength and
Type of aroma is fitted analysis, rejects error dot, finds association and the rule of data, and the odor characteristics that instrument is detected are original
Data and artificial sense evaluation data are combined as the comprehensive basic data of artificial intelligence sensory evaluation flavour of food products system, establish
Food smell database.
2nd, the pretreatment of computer odor data and feature extraction and the foundation of evaluation model
（3）Deep learning scoring model is built.Smell higher-dimension multi-source characteristic is trained, existing CNN,
On the basis of LSTM even depth learning models, a kind of new multi-modal, multitask deep learning framework M is built3TDN.It will be in spy
On the basis of sign extraction, multilayer LSTM is respectively adopted and multilayer is limited Boltzmann machine structure training scoring model.Wherein, depth
LSTM is employed in structure and allows for different material ingredient in the flavour characteristic interdependence of formation；It is multi-modal to refer to data
It is multi-modal, including chromatographic data, mass spectrometric data and food appearance image data etc.；Multitask refers to by while learns flavor class
Multiple targets such as type, flavor intensity, flavor amplitude.
Embodiment 2：Using artificial intelligence sensory evaluation flavour of food products system to different Storing-year Xinhui tangerine peels into sector-style
Taste is evaluated.
Dried orange peel has the saying of " old long person is good ", volatile aroma protrudes, and with storage from ancient times as qi-regulating Chinese medicine material
The extension in time, the content and odor characteristic of volatile ingredient can change, by the number for establishing dried orange peel volatile ingredient
According to library, dried orange peel smell artificial intelligence appraisement system is built, the dried orange peel in different year, the place of production is played and differentiates effect.In addition, evaluation
System can be associated with out the flavor of counter sample, save labour by the comparison to dried orange peel sample GC-MS data, more smart
The flavor data of sample is known accurately.
（1）Data acquire
As shown in Figure 1, research dried orange peel odoring substance, selects the Xinhui tangerine peel of different year as experimental raw, dried orange peel time packet
Include, nineteen eighty-two, 1997,1999,2000,2001,2004,2009,2010 and 2012 in 1980 in 1978
Year.When preparing sample, ignore its time, and to each sample number into spectrum 1-11.With beveller by sample comminution, each sample weighs
2 g, 75 μm of CAR/PDMS syringe needles, 30 min of Headspace-solid phase microextraction at 90 DEG C parse 10 min in injection port.
GC-MS selects DB-5 nonpolar elastic quartz capillary columns, and temperature program is that injector temperature is 260 DEG C, flow velocity
For 1.0ml/min, carrier gas is high-purity helium（＞ 99.999%）；70 DEG C of initial temperature is kept 2 min, is raised to the rate of 4 DEG C/min
210 DEG C, keep 10 min, split ratio 50:1.Mass Spectrometry Conditions：Ion source is 230 DEG C, electron energy 70eV, MS level four bars temperature
Spend is 150 DEG C.
Volatile ingredient is after gas-chromatography post separation, into mass spectrum end, and by multiple mass spectral analysis after, obtain each
The volatile ingredient type of time Xinhui tangerine peel, appearance time and relative amount and including chromatography and mass spectrum initial data.Data
Such as table 1.
（2）Data prediction and comparison
Data preparation be can recognize that into form into computer, the volatile ingredient that each time dried orange peel and all dried orange peels include is compiled
Number, such as dried orange peel data number in 1978 is 1978000, and the sample gas-chromatography initial data on record schemes each point including TIC
Time and intensity, the corresponding peak area of each chromatographic peak and peak height.The raw mass spectrum data of all samples appearance substance are converged
Always, include the mass-to-charge ratio and intensity of each fragment ion of the substance.It is determined by initial data qualitative to dried orange peel volatile ingredient
It is quantitative, food smell artificial intelligence evaluation system is imported data to, obtains the fragrance information of dried orange peel volatile ingredient.As a result such as table
2。
（3）Flavor evaluation is proofreaded
Wen Xiangshi carries out GC-O detections, subjective appreciation dried orange peel flavor to dried orange peel volatile ingredient.Method is as follows：
Each 2 g of time sample is taken, grinder crushes, and 30min, 75 μm of CAR/PDMS extraction syringe needle extractions are adsorbed at 90 DEG C
30min, injection port parsing 10min.
Chromatographic condition：It is injector temperature is 260 that chromatographic column, which is DB-5 nonpolar elastic quartz capillary column temperature programmings,
DEG C, flow velocity 1.0ml/min, carrier gas is high-purity helium（＞ 99.999%）；70 DEG C of column initial temperature keeps 2min, with 4 DEG C/min's
Rate rises to 210 DEG C, keeps 10min, split ratio 50:1.
Mass Spectrometry Conditions：Ion source is 230 DEG C, and electron energy 70eV, MS level four bars temperature is 150 DEG C.
Olfactometry device：ODP heating temperatures are 270 DEG C, and make-up gas flow 50mL/min, humidifier is in smelling exit
Make-up gas is humidified to reduce injury of the dry gas to schneiderian membrance.The person of distinguishing is smelt with training early period to be tested, and smells the person of distinguishing
The time for smelling smell, aromatic property and aroma strength are recorded in olfactometry mouth.Table 4 hears fragrant teacher and artificial intelligence for profession
The comparison of evaluation system evaluation result hears the accuracy of fragrant teacher as 100% using profession.
Embodiment 3：Flavor evaluation is carried out to the ethanol solution of various concentration.
（1）Establish sensory evaluation data library and flavor intelligent Evaluation model：
At 20 DEG C of room temperature, take 100 ml volumetric flasks of 21 clean drieds, it is accurate respectively measure 0,5,10,15,20,25,30,
35th, 40,45,50,55,60,65,70,75,80,85,90,95, the absolute ethyl alcohol of 100ml is placed in 100 ml volumetric flasks, is added in
Water constant volume, be made a concentration of 0%, 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%,
75%th, 80%, 85%, 90%, 95%, 100% totally 21 ethanol solution standard samples.
By 21 ethanol solution standard samples, to hear the Wen Xiangshi of fragrant paper method training subjective appreciation group, determine preliminary
Artificial sense evaluation.Method is as follows with reference to GB/T 22366-2008：A kind of ethanol solution of concentration is identical together with two
Other concentration ethanol solution standard samples are supplied to Wen Xiangshi, Wen Xiang to select concentration in being an apprentice of different from other samples of two together
Product.Training in this way hears fragrant teacher 2 weeks, and news perfume (or spice) teacher is made to be familiar with discriminating power.
21 ethanol solution standard samples are separately entered into GC, after being detached via capillary column, outflow component is split valve point
Into two-way, all the way into chemical detector MS, another way enters sniff mouth by dedicated transfer line, is smelt simultaneously by Wen Xiangshi news
It is carried out at the same time manual knob electronic signal record.GC-MS conditions are as follows：
Chromatographic condition：Chromatographic column is DB-1 type polarity fused-silica capillary columns.Temperature programming is that injector temperature is 250 DEG C,
Flow velocity is 1.0ml/min, and carrier gas is high-purity helium（＞ 99.999%）；50 DEG C of column initial temperature keeps 1min, with the speed of 15 DEG C/min
Rate rises to 110 DEG C, keeps 1min, split ratio 8:1.
Mass Spectrometry Conditions：230 DEG C of interface temperature, ion source electron energy 70eV, 1.0 kV mass numbers of electron multiplier high pressure
Scanning range 20 ~ 100,0.2 s of sampling rate.
Olfactometry device：ODP heating temperatures are 150 DEG C, and make-up gas flow 50mL/min, humidifier is in smelling exit
Make-up gas is humidified to reduce injury of the dry gas to schneiderian membrance.
Record data result（Aroma characteristic, intensity）, fragrance information data that will hear fragrant teacher's smelling and obtain analyzes with reference to GC
In derived data, establish ingredient and relative amount relational database（Such as table 5, Fig. 3）, build flavor intelligent Evaluation model（Such as
Table 6, Fig. 4）.
Scoring setting：Ethyl alcohol fragrance is divided into nothing（0-10）, light perfume (or spice)（10-20）, faint scent（20-30）, giving off a strong fragrance（30-40）、
It is pungent（40-50）, pungent stimulation（50-60）6 oder levels, aroma strength scoring section is 0-60.
（2）Verify sensory evaluation model
Prepare the ethyl alcohol sample to be tested of 5 random concentration：It is hearing outside the fragrant teacher visual field, is taking the 100ml graduated cylinders of a clean dried, with
Meaning pours into the absolute ethyl alcohol of any amount, writes down absolute ethyl alcohol volume, moves into 100 ml volumetric flasks of clean dried, with water constant volume,
It writes down but does not mark its ethanol content, be repeated 5 times, each solution numbers are given with digital 1-5.
Wen Xiangshi is evaluated：5 every part of samples are taken into three repetitions, all samples sequence are upset, with alphabetical a-o again
Label transfers to Wen Xiang Shi Duli smellings, describes flavor type and intensity, writes down result.
5 ethyl alcohol samples to be tested are entered into GC-MS analyses and GC-O analyses, method are as follows：
Chromatographic condition：Chromatographic column is DB-1 type polarity fused-silica capillary columns.Temperature programming is that injector temperature is 250 DEG C,
Flow velocity is 1.0ml/min, and carrier gas is high-purity helium（＞ 99.999%）；50 DEG C of column initial temperature keeps 1min, with the speed of 15 DEG C/min
Rate rises to 110 DEG C, keeps 1min, split ratio 8:1.
Mass Spectrometry Conditions：230 DEG C of interface temperature, ion source electron energy 70eV, 1.0 kV mass numbers of electron multiplier high pressure
Scanning range 20 ~ 100,0.2 s of sampling rate.
Olfactometry device：ODP heating temperatures are 150 DEG C, and make-up gas flow 50mL/min, humidifier is in smelling exit
Make-up gas is humidified to reduce injury of the dry gas to schneiderian membrance.
By initial data derived in GC-MS, it is transferred to established flavor intelligent Evaluation model and calculates, obtain ethyl alcohol
The data of content and aroma strength.As a result such as table 7.
Claims (6)
1. a kind of artificial intelligence sensory evaluation flavour of food products system, which is characterized in that including ingredient organoleptic detection system, computer
Intelligent analysis system and food smell database,
The ingredient organoleptic detection system is made of GC-MS and GC-O, and GC-MS is used to obtain the odor characteristics original number of food
According to GC-O is used to obtain the artificial sense evaluation data of food smell（Odour intensity and odor characteristics）；
The computer intelligence analysis system, for the odor characteristics initial data and artificial sense obtained to organoleptic detection system
Evaluation data are pre-processed and feature extraction, and dimensionality reduction, cluster analysis are carried out to initial data；
The food smell database is that the odor characteristics obtained to organoleptic detection system by Artificial intelligent algorithm are original
Data and artificial sense evaluation data are fitted analysis, reject error dot, find association and the rule of data, the smell of foundation
The food volatile ingredient that initial data and artificial evaluation data are combined is mutually related Various types of data.
2. artificial intelligence sensory evaluation flavour of food products system according to claim 1, which is characterized in that the Computing Intelligence
Energy analysis system uses multi-modal, multitask deep learning framework M3TDN, it is described it is multi-modal refer to the multi-modal of data, including
Chromatographic data, mass spectrometric data and food appearance image data, multiple targets that the multitask refers to by while learns, including flavor
Type, flavor intensity, flavor amplitude.
3. artificial intelligence sensory evaluation flavour of food products system according to claim 1, which is characterized in that the food smell
Database includes Chromatographic information, Information in Mass Spectra, odor characteristic, aroma strength, the substance image of volatile ingredient.
4. a kind of construction method of artificial intelligence sensory evaluation flavour of food products system, which is characterized in that including：
（1）Food samples are entered into GC, after being detached via capillary column, outflow component is split valve and is divided into two-way, enters all the way
Chemical detector（FID or MS）, the odor characteristics initial data being made of chromatography and mass spectrum initial data is obtained, another way passes through
Dedicated transfer line enters sniff mouth（O）, obtain artificial sense evaluation data；
（2）Data are evaluated to odor characteristics initial data and artificial sense by computer software（Content, aroma strength and fragrance
Type etc.）Analysis is fitted, rejects error dot, association and the rule of data are found, so as to establish odor characteristics initial data
The food smell database being combined with artificial sense evaluation data；
（3）Odor characteristics initial data in food smell database and artificial sense evaluation data are normalized and quantified
Processing, obtains higher-dimension flavor data, and be limited Boltzmann machine using depth carries out feature extraction to the higher-dimension flavor data；
（4）Higher-dimension multi-source characteristic after feature extraction is trained, builds multi-modal, multitask deep learning framework
M3TDN on the basis of feature extraction, is respectively adopted multilayer LSTM and multilayer is limited Boltzmann machine training scoring model, from
And establish the artificial intelligence flavor evaluation system of food.
5. the construction method of artificial intelligence sensory evaluation flavour of food products system according to claim 4, which is characterized in that step
Suddenly（1）In, the method that the artificial evaluation data obtain is that sample enters sniff mouth, is carried out while being smelt by people's nose news manual
Knob electronic signal records, and record data result forms spectrogram or transitional information data by computer disposal.
6. the construction method of artificial intelligence sensory evaluation flavour of food products system according to claim 4, which is characterized in that step
Suddenly（3）In, the method for the feature extraction can also will be obtained in food database with machine decision, artificial interpretation mode
Input data of odor characteristics initial data and artificial sense the evaluation data arrived as characteristic processing, learns using LDA, PCA
Classic algorithm carries out feature-extraction analysis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2018/000423 WO2019114052A1 (en) | 2017-11-12 | 2018-12-17 | Artificial intelligence-based food flavor sensory evaluation system and establishment method therefor |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2017111106728 | 2017-11-12 | ||
| CN201711110672 | 2017-11-12 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108133313A true CN108133313A (en) | 2018-06-08 |
| CN108133313B CN108133313B (en) | 2021-07-20 |
Family
ID=62390497
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711358141.0A Active CN108133313B (en) | 2017-11-12 | 2017-12-17 | Artificial intelligent sensory evaluation food flavor system and construction method thereof |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN108133313B (en) |
| WO (1) | WO2019114052A1 (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109374802A (en) * | 2018-09-06 | 2019-02-22 | 北京工商大学 | The differentiation of identity-based identification characteristic flavor on basis component is cut up to cheese grade method for distinguishing |
| CN109409635A (en) * | 2018-08-27 | 2019-03-01 | 中化现代农业有限公司 | The processing of agricultural product sensorial data, acquisition methods and device, system |
| CN109709221A (en) * | 2018-12-14 | 2019-05-03 | 江苏恒顺醋业股份有限公司 | The lookup analysis method of smell substance in a kind of fermentation vinegar liquid based on GC-MS |
| WO2019114052A1 (en) * | 2017-11-12 | 2019-06-20 | 华南农业大学 | Artificial intelligence-based food flavor sensory evaluation system and establishment method therefor |
| CN110082458A (en) * | 2019-05-31 | 2019-08-02 | 北京工商大学 | It cuts up to cheese volatile materials and the recognition methods of organoleptic quality level dependencies |
| CN110596330A (en) * | 2019-10-16 | 2019-12-20 | 粤海永顺泰(广州)麦芽有限公司 | Special malt quality evaluation method |
| CN111260199A (en) * | 2020-01-10 | 2020-06-09 | 江西理工大学 | Artificial intelligent sensory evaluation food flavor system and construction method thereof |
| CN111474287A (en) * | 2020-03-19 | 2020-07-31 | 杭州憶盛医疗科技有限公司 | Computer-aided system and method for analyzing composition components of medicine |
| CN111695825A (en) * | 2020-06-16 | 2020-09-22 | 内蒙古蒙牛乳业（集团）股份有限公司 | Food flavor detection method and device and electronic equipment |
| CN111693661A (en) * | 2020-06-11 | 2020-09-22 | 上海交通大学 | Intelligent sense organ based dry-cured ham flavor evaluation method |
| CN112858650A (en) * | 2021-01-13 | 2021-05-28 | 上海应用技术大学 | Analysis research method for improving automobile leather smell based on sigma-tau intensity method |
| CN113396324A (en) * | 2019-02-12 | 2021-09-14 | 曾田香料株式会社 | Method for evaluating volatile component and apparatus for the same |
| CN113504316A (en) * | 2021-05-31 | 2021-10-15 | 江苏中烟工业有限责任公司 | Method for evaluating main body fragrance of Maillard reaction product by utilizing modular fragrance functional group |
| CN114324716A (en) * | 2021-08-30 | 2022-04-12 | 华南农业大学 | Method for measuring degree of oxidative rancidity of infant formula rice flour and application of method |
| CN114441667A (en) * | 2020-11-02 | 2022-05-06 | 株式会社岛津制作所 | Odor evaluation device, odor evaluation method, and odor evaluation sample adjustment device |
| CN114487234A (en) * | 2022-02-11 | 2022-05-13 | 云南中烟工业有限责任公司 | Method for simultaneously predicting moisture content and smoke characteristic index judgment value in tobacco drying process based on HS-GC-MS |
| CN115047091A (en) * | 2022-04-06 | 2022-09-13 | 中华全国供销合作总社济南果品研究所 | Analysis method for aroma components in dried red date slices |
| CN115754179A (en) * | 2022-11-07 | 2023-03-07 | 佛山市海天(高明)调味食品有限公司 | Oyster juice detection method |
| WO2024038419A1 (en) * | 2022-08-19 | 2024-02-22 | The Live Green Group, Inc. | Plant-only cheese flavoring replacement system in food products |
| CN118604275A (en) * | 2024-08-08 | 2024-09-06 | 中国农业科学院农产品加工研究所 | Directional flavor editing method and system for food characteristic flavor preservation |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11636870B2 (en) | 2020-08-20 | 2023-04-25 | Denso International America, Inc. | Smoking cessation systems and methods |
| US11932080B2 (en) | 2020-08-20 | 2024-03-19 | Denso International America, Inc. | Diagnostic and recirculation control systems and methods |
| US11760170B2 (en) | 2020-08-20 | 2023-09-19 | Denso International America, Inc. | Olfaction sensor preservation systems and methods |
| US11760169B2 (en) | 2020-08-20 | 2023-09-19 | Denso International America, Inc. | Particulate control systems and methods for olfaction sensors |
| US11828210B2 (en) | 2020-08-20 | 2023-11-28 | Denso International America, Inc. | Diagnostic systems and methods of vehicles using olfaction |
| US12017506B2 (en) | 2020-08-20 | 2024-06-25 | Denso International America, Inc. | Passenger cabin air control systems and methods |
| US11813926B2 (en) | 2020-08-20 | 2023-11-14 | Denso International America, Inc. | Binding agent and olfaction sensor |
| US11881093B2 (en) | 2020-08-20 | 2024-01-23 | Denso International America, Inc. | Systems and methods for identifying smoking in vehicles |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010082684A1 (en) * | 2009-01-13 | 2010-07-22 | Kao Corporation | Fragrance composition |
| CN102135526A (en) * | 2010-12-10 | 2011-07-27 | 中国农业科学院农产品加工研究所 | Method for extracting characteristic flavor substance of roasted peanuts |
| CN102323385A (en) * | 2011-06-03 | 2012-01-18 | 上海应用技术学院 | Method for measuring smell threshold of volatile flavor compound and application thereof |
| CN104316660A (en) * | 2014-10-27 | 2015-01-28 | 四川省农业科学院农产品加工研究所 | Method for rapidly detecting flavor and grade of thick broad-bean sauce and product tracing |
| CN104316635A (en) * | 2014-10-28 | 2015-01-28 | 石河子大学 | Method for rapidly identifying flavor and quality of fruits |
| CN106874948A (en) * | 2017-02-08 | 2017-06-20 | 武汉海卓科科技有限公司 | A kind of black smelly water automatic identification and appraisal procedure |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103105440A (en) * | 2012-11-06 | 2013-05-15 | 天津科技大学 | Method for identifying volatile flavor substance in bolete |
| CN103487558B (en) * | 2013-07-30 | 2016-10-12 | 中国标准化研究院 | A kind of method detecting exceptional sample during the pattern recognition analysis applying intelligent sensory signal to carry out tea leaf quality |
| CN108133313B (en) * | 2017-11-12 | 2021-07-20 | 华南农业大学 | Artificial intelligent sensory evaluation food flavor system and construction method thereof |
-
2017
- 2017-12-17 CN CN201711358141.0A patent/CN108133313B/en active Active
-
2018
- 2018-12-17 WO PCT/CN2018/000423 patent/WO2019114052A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010082684A1 (en) * | 2009-01-13 | 2010-07-22 | Kao Corporation | Fragrance composition |
| CN102135526A (en) * | 2010-12-10 | 2011-07-27 | 中国农业科学院农产品加工研究所 | Method for extracting characteristic flavor substance of roasted peanuts |
| CN102323385A (en) * | 2011-06-03 | 2012-01-18 | 上海应用技术学院 | Method for measuring smell threshold of volatile flavor compound and application thereof |
| CN104316660A (en) * | 2014-10-27 | 2015-01-28 | 四川省农业科学院农产品加工研究所 | Method for rapidly detecting flavor and grade of thick broad-bean sauce and product tracing |
| CN104316635A (en) * | 2014-10-28 | 2015-01-28 | 石河子大学 | Method for rapidly identifying flavor and quality of fruits |
| CN106874948A (en) * | 2017-02-08 | 2017-06-20 | 武汉海卓科科技有限公司 | A kind of black smelly water automatic identification and appraisal procedure |
Cited By (27)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019114052A1 (en) * | 2017-11-12 | 2019-06-20 | 华南农业大学 | Artificial intelligence-based food flavor sensory evaluation system and establishment method therefor |
| CN109409635B (en) * | 2018-08-27 | 2021-08-20 | 中化现代农业有限公司 | Agricultural product sensory data processing and acquiring method, device and system |
| CN109409635A (en) * | 2018-08-27 | 2019-03-01 | 中化现代农业有限公司 | The processing of agricultural product sensorial data, acquisition methods and device, system |
| CN109374802A (en) * | 2018-09-06 | 2019-02-22 | 北京工商大学 | The differentiation of identity-based identification characteristic flavor on basis component is cut up to cheese grade method for distinguishing |
| CN109374802B (en) * | 2018-09-06 | 2020-07-07 | 北京工商大学 | Method for distinguishing cut cheese grade based on identity characteristic flavor component |
| CN109709221A (en) * | 2018-12-14 | 2019-05-03 | 江苏恒顺醋业股份有限公司 | The lookup analysis method of smell substance in a kind of fermentation vinegar liquid based on GC-MS |
| CN113396324B (en) * | 2019-02-12 | 2024-04-16 | 曾田香料株式会社 | Method for evaluating volatile component and device for the method |
| CN113396324A (en) * | 2019-02-12 | 2021-09-14 | 曾田香料株式会社 | Method for evaluating volatile component and apparatus for the same |
| CN110082458A (en) * | 2019-05-31 | 2019-08-02 | 北京工商大学 | It cuts up to cheese volatile materials and the recognition methods of organoleptic quality level dependencies |
| CN110082458B (en) * | 2019-05-31 | 2022-05-03 | 北京工商大学 | Method for identifying correlation between cheddar cheese volatile substances and sensory quality grade |
| CN110596330A (en) * | 2019-10-16 | 2019-12-20 | 粤海永顺泰(广州)麦芽有限公司 | Special malt quality evaluation method |
| CN111260199A (en) * | 2020-01-10 | 2020-06-09 | 江西理工大学 | Artificial intelligent sensory evaluation food flavor system and construction method thereof |
| CN111474287A (en) * | 2020-03-19 | 2020-07-31 | 杭州憶盛医疗科技有限公司 | Computer-aided system and method for analyzing composition components of medicine |
| CN111693661A (en) * | 2020-06-11 | 2020-09-22 | 上海交通大学 | Intelligent sense organ based dry-cured ham flavor evaluation method |
| CN111695825A (en) * | 2020-06-16 | 2020-09-22 | 内蒙古蒙牛乳业（集团）股份有限公司 | Food flavor detection method and device and electronic equipment |
| CN114441667A (en) * | 2020-11-02 | 2022-05-06 | 株式会社岛津制作所 | Odor evaluation device, odor evaluation method, and odor evaluation sample adjustment device |
| CN112858650B (en) * | 2021-01-13 | 2023-08-18 | 上海应用技术大学 | Analysis and research method for improving automobile leather smell based on sigma-tau strength method |
| CN112858650A (en) * | 2021-01-13 | 2021-05-28 | 上海应用技术大学 | Analysis research method for improving automobile leather smell based on sigma-tau intensity method |
| CN113504316A (en) * | 2021-05-31 | 2021-10-15 | 江苏中烟工业有限责任公司 | Method for evaluating main body fragrance of Maillard reaction product by utilizing modular fragrance functional group |
| CN113504316B (en) * | 2021-05-31 | 2023-08-29 | 江苏中烟工业有限责任公司 | Method for evaluating main fragrance of Maillard reaction product by utilizing functional groups |
| CN114324716A (en) * | 2021-08-30 | 2022-04-12 | 华南农业大学 | Method for measuring degree of oxidative rancidity of infant formula rice flour and application of method |
| CN114487234A (en) * | 2022-02-11 | 2022-05-13 | 云南中烟工业有限责任公司 | Method for simultaneously predicting moisture content and smoke characteristic index judgment value in tobacco drying process based on HS-GC-MS |
| CN115047091A (en) * | 2022-04-06 | 2022-09-13 | 中华全国供销合作总社济南果品研究所 | Analysis method for aroma components in dried red date slices |
| WO2024038419A1 (en) * | 2022-08-19 | 2024-02-22 | The Live Green Group, Inc. | Plant-only cheese flavoring replacement system in food products |
| CN115754179A (en) * | 2022-11-07 | 2023-03-07 | 佛山市海天(高明)调味食品有限公司 | Oyster juice detection method |
| CN118604275A (en) * | 2024-08-08 | 2024-09-06 | 中国农业科学院农产品加工研究所 | Directional flavor editing method and system for food characteristic flavor preservation |
| CN118604275B (en) * | 2024-08-08 | 2024-10-25 | 中国农业科学院农产品加工研究所 | Directional flavor editing method and system for food characteristic flavor preservation |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2019114052A1 (en) | 2019-06-20 |
| CN108133313B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108133313A (en) | A kind of artificial intelligence sensory evaluation flavour of food products system and its construction method | |
| US10895560B2 (en) | Electronic nose instrument for sensory quality evaluation of tobacco and tobacco product | |
| Dressler et al. | Sociophonological methods in the study of sociolinguistic variation in Viennese German | |
| CN101504395A (en) | Method for controlling quality of flavoring essence for tobacco by gas-phase fingerprint pattern | |
| CN110047516A (en) | A kind of speech-emotion recognition method based on gender perception | |
| Chastrette | Classification of odors and structure-odor relationships | |
| CN106909572A (en) | A kind of construction method and device of question and answer knowledge base | |
| Vincs et al. | Snapshots of complexity: using motion capture and principal component analysis to reconceptualise dance | |
| CN103336090A (en) | Support vector machine (SVM) method for predicting cigarette sensory quality indicators | |
| Tian et al. | Development of a flavour fingerprint by GC‐MS and GC‐O combined with chemometric methods for the quality control of Korla pear (Pyrus serotina Reld) | |
| Böck et al. | Intraindividual and interindividual multimodal emotion analyses in human-machine-interaction | |
| CN105954366A (en) | Quality monitoring and control method for tobacco essence perfume | |
| CN106404884A (en) | Method for quickly evaluating quality consistency of flavors and fragrances of volatile cigarettes by HS-IMR-MS | |
| MacIntyre et al. | Pushing the envelope: Evaluating speech rhythm with different envelope extraction techniques | |
| Prasetyawan et al. | Odor Reproduction Technology Using a Small Set of Odor Components | |
| Santos et al. | Pilot study for validation and differentiation of alveolar and esophageal air | |
| Santos et al. | Instrumentation for differentiation of exhaled air | |
| CN202794093U (en) | Device for fast detecting quality of baked food based on bionic olfaction | |
| Rey et al. | Episodic memory and recognition are influenced by cues’ sensory modality: comparing odours, music and faces using virtual reality | |
| Bansleben et al. | Chemometric tools for identification of volatile aroma-active compounds in oregano | |
| CN110782917B (en) | Poetry reciting style classification method and system | |
| Manetta et al. | Verbal representation of fragrances: Dependence on specific task | |
| Chen et al. | Do speech features for detecting cognitive load depend on specific languages? | |
| Zheng et al. | The Extraction Method of Emotional Feature Based on Children's Spoken Speech | |
| Stone et al. | Techniques for sensory evaluation of food odors |
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 | ||
| CB03 | Change of inventor or designer information |
Inventor after: Cao Yong Inventor after: Wen Linfeng Inventor after: Fu Jiangyan Inventor after: Liu Zhan Inventor after: Hu Yuanshu Inventor after: Qiu zhirou Inventor after: Chang Huiyou Inventor after: Chen Haoran Inventor after: Sun Shengwei Inventor after: Hu Yongjun Inventor after: Chen Tong Inventor after: Liu Guo Inventor after: Kan Qixin Inventor before: Cao Yong Inventor before: Qiu zhirou Inventor before: Chang Huiyou Inventor before: Chen Haoran Inventor before: Sun Shengwei Inventor before: Hu Yongjun Inventor before: Chen Tong |
|
| CB03 | Change of inventor or designer information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |