CN1595164A - Method for verifying original brandy made in French - Google Patents
Method for verifying original brandy made in French Download PDFInfo
- Publication number
- CN1595164A CN1595164A CN 200410024379 CN200410024379A CN1595164A CN 1595164 A CN1595164 A CN 1595164A CN 200410024379 CN200410024379 CN 200410024379 CN 200410024379 A CN200410024379 A CN 200410024379A CN 1595164 A CN1595164 A CN 1595164A
- Authority
- CN
- China
- Prior art keywords
- brandy
- acid
- ethyl
- network
- french
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Investigating Or Analyzing Non-Biological Materials By The Use Of Chemical Means (AREA)
Abstract
It is a detection method for French original brandy, which is the following: to use the French brandy sample as sample space to form a competitive reverse transmit human neural net input matrix composed of fore reverse transmit neural net and back competitive function layer; to use the identity matrix with order of the brandy order as training aim and to use the brandy sample input matrix and stimulate input matrix with noise to train the connection weigh and establish the net connection; to input the vector composed of organic acid , volatile component and metal element contents in the brandy into the established neural net, wherein, the most outstanding neutral element succeeds and is displayed as result and check the final detect result of the train aim.
Description
Technical field
The present invention relates to a kind of France and originate in the brandy authentication method.
Background technology
Along with the demand of home market to foreign wine day by day improves, China is just becoming the principal market of external brandy (grape) wine big country.When foreign wine entered domestic consumption market one after another, quality problems also became this field hot topic.Because great number market profit and domestic consumer be to the psychology of following blindly of foreign wine, dragons and fishes jumbled together in import foreign wine market, very different, and more the someone adulterates, pretends to be the original producton location famous brand of wine to cheat the consumer, seeks exorbitant profit.Particularly the end of the nineties in last century, incidents such as sensational ox blood cleanser scandal, the shoddy fraud scandal of method national wine allowed the noble image of foreign wine have a greatly reduced quality to the beginning of this century, and some consumers of China trust crisis occurred to foreign wine.Therefore, set up the reliable imported France brandy authentication method of a cover, to safeguarding market order, protecting consumer's interests significant.
In the most flourishing France of brandy (grape) wine industry, the classification systems in 1855 that followed hundreds of years are adopted in evaluation and classification to brandy, promptly come the brandy of various brands, kind is carried out classification, evaluation, but this method is not accurate enough and effective by sample wine record and the scoring of commenting the wine teacher.For preventing fake wine, the investigator that fake products hits in France wishes that also a kind of new technology replaces this not scientific approach of present use.Some cluster analysis technology commonly used, as principal component analysis (PCA), K-Means cluster, Hierarchical cluster, Discriminant cluster analysis, can only make rough identification to true and false, the true and false place of production of brandy, classification results is not accurate enough reliable, the statistics of shortage method guarantees, has common theoretical defects.
Along with the application of Modern Analytical Instrument at the brandy analysis field, people are having new progress aspect brandy flavor characteristic and the evaluation identification, work embodies a concentrated reflection of the searching to brandy wine taste characteristic component, attempt by finding one or more characteristic peaks, definite and content analysis by to characteristic peak reaches the brandy flavor characteristic and analyzes and true and false evaluation purpose.Facts have proved that this thinking also is impracticable, main because: it almost is impossible 1, describing complex system by one or more indexs; 2, one or more indexs that directly obtain are because instability or artificial interference cause uncertainty to identification or evaluation; 3, be difficult to select effective several indexs to describe to complex system.
In the analytical chemistry field, artificial neural network has been brought into play effect in the pattern classification of handling complex sample as aspects such as crops place of production identifications.The potato place of production, the North America recognition system of comparatively successful application such as United States Department of Agriculture's exploitation, this cover is presided over the recognition system (U.S. Patent number 632.4531) of exploitation and is measured by the plasma emission spectrum of 18 kinds of trace meter content of potato sample by Anderson people such as (2001), the detection of protein and peptide and various active enzyme, sample with 562 potato places of production, North America is done recognition objective and training set, sets up the place of production, effective artificial neural network potato North America recognition system.But do not see useful artificial neural network evaluation French brandy so far as yet.
Summary of the invention
The purpose of this invention is to provide a kind of France and originate in the brandy authentication method, it can overcome the above-mentioned shortcoming of prior art.
A kind of France originates in the brandy authentication method, it is characterized in that with the known multiple organic acid of French brandy sample, multiple volatile components and multiple metal element content be the variable space, is the input matrix that the sample point space constitutes competitive back-propagation artificial neural network with known French brandy sample; This competitiveness reverse transmittance nerve network is made of front end reverse transmittance nerve network and rear end competitive function layer, with the order of known French brandy sample is that the unit matrix of order is the network training target, with known French brandy sample input matrix and simulate noisy vocal input matrix and network is connected weights train, set up network and connect; The neural network that the vector input that the organic acid of French brandy to be identified, volatile components and metal element content are constituted is set up, after the conversion of counterpropagation network and competitive function layer, the most significant neuron is won and is as a result of showed, and the title of checking the French brandy of training objective correspondence obtains final qualification result.
Advantage of the present invention is to identify accurately that France originates in the true and false of brandy.
Description of drawings
Accompanying drawing 1 is a global procedures synoptic diagram of the present invention.
Accompanying drawing 2 is for identifying the used back-propagation artificial neural network structural representation of French brandy.
Embodiment
Present embodiment is example explanation the present invention with French brandy Hennessy VSOP and Remy Martin XO, and global procedures of the present invention as shown in Figure 1.
At first measure metal element content, organic acid content, the volatile components content of known French brandy sample respectively with atomic absorption, high performance liquid chromatography, gas chromatography-mass spectrography; Metallic element in described Hennessy VSOP and the Remy Martin XO is arsenic, antimony, barium, selenium, lead, copper, nickel, cadmium, cobalt, zinc, potassium, sodium, lithium, tin, calcium, manganese, magnesium, iron, chromium, mercury, and measurement result sees Table 1; Organic acid in described Hennessy VSOP and the Remy Martin XO is malic acid, succinic acid, citric acid, tartrate, lactic acid, and measurement result sees Table 1; Volatile components in described Hennessy VSOP and the Remy Martin XO is 2 furan carboxyaldehyde, n-hexyl alcohol, ethyl hexanoate, phenylethyl alcohol, diethyl succinate, sad, ethyl caprilate, capric acid, ethyl caprate, dodecoic acid, ethyl laurate, 4-hydroxyl-3,5-dimethoxy-benzaldehyde, tetradecanoic acid, ethyl myristate, hexadecanoic acid, ethyl palmitate, ethyl linoleate, docosane, hexacosane, squalane, measurement result sees Table 1.With above-mentioned data is the variable space, is the input matrix that the sample point space constitutes competitive back-propagation artificial neural network with known French brandy sample.
Table 1.
The plain component poem of the pavilion Buddhist nun of metal unit number of people horse mg/L VSOP XO | Volatile components pavilion Buddhist nun number of people mg/L poem horse VSOP XO | Organic acid Hennessy Remy Martin component VSOP XO g/L |
Cadmium 0.047 0.105 cobalt 0.082 0.052 chromium 0.280 0.110 iron 1.302 0.392 manganese 0.080 0.015 manganese 0.140 0.087 bronze medal 1.287 0.767 magnesium 1.199 0.764 calcium 2.752 3.160 tin 11.053 6.333 lithiums 0.150 0.187 sodium 3.057 2.191 potassium 0.001 0.001 antimony 0.073 0.210 arsenic 0.001 0.001 barium 0.077 0.039 plumbous 0.031 0.036 selenium, 0.013 0.006 zinc, 0.828 0.187 mercury 0.004 0.004 | Furfural 15.14 45.65 benzyl carbinols 1.66 7.34 4-hydroxyls-3,5-dimethoxy-benzaldehyde 5.65 31.12 n-hexyl alcohols 5.28 4.84 ethyl hexanoates 2.70 3.03 ethyl caprilates 12.76 19.60 sad 63.41 111.08 ethyl caprates, 0.53 23.28 capric acid, 72.13 0.10 tetradecylic acids, 2.85 2.56 hexadecylic acids, 26.80 44.50 diethyl succinates, 0.47 0.7g ethyl myristate, 0.05 0.10 ethyl palmitate, 0.20 0.20 ethyl linoleate, 0.62 0.60 ethyl laurate, 0.02 8.54 docosane, 0.71 0.99 hexacosane, 0.92 0.99 dodecoic acid, 21.11 0.10 saualanes 0.87 0.76 | Malic acid 0.0777 0.0751 succinic acids 0.2513 0.2269 tartrate 0.1112 0.1333 citric acids 0.5930 0.7921 lactic acid 0.4612 0.3961 |
Then, the network of back-propagation artificial neural network being connected train, is that the unit matrix of order is the network training target with the order of known French brandy sample, sets up network and connects, promptly produce " 1 " all the other position outputs " 0 " in the output vector relevant position; This competitiveness reverse transmittance nerve network is made of front end reverse transmittance nerve network and rear end competitive function layer, reverse transmittance nerve network is that input layer contains 45 inputs, output layer contains n (the French brandy standard model number that n provides for authoritative institution, n is 12 in the present embodiment) output three layers of (single hidden layer) error back propagation network, hidden layer contains 10 neurons, adopt Logsig () function as excitation function, adopt adaptive learning speed to train in conjunction with the momentum mode, the momentum term constant is 0.95, with the total error quadratic sum as convergence criterion.Because input matrix has noise in hands-on, the use, make difficult accurately produce " 1 " and " 0 " of output vector, counterpropagation network is exported after the conversion of competition layer function, the output of the easiest generation " 1 " is by competition layer competition-inhibiting effect, only an input becomes the victor at last, to relevant each connection weight of triumph neuron towards the direction adjustment that more helps competing, just expression is to the input pattern classification for the neuron of this triumph, and the back-propagation artificial neural network structural drawing of French brandy sample identification is seen Fig. 2.
By on the basis of the raw data training that obtains of step, the data set that adopts many groups to contain noise connects network trains, to reach the ability of handling real data; The training data group that contains noise is on the raw data basis, according to the noise level of determination data, utilizes noise to produce function and produces multi-group data at random.
At last, French brandy sample to be identified is identified, Hennessy VSOP to be identified and the Remy Martin XO of taking from same batch are measured same metal element content, organic acid content, volatile components content (seeing Table 2) respectively according to above-mentioned steps;
Table 2.
The plain component poem of the pavilion Buddhist nun of metal unit Remy Martin XO mg/L VSOP | Volatile components Hennessy Remy Martin mg/L VSOP XO | Organic acid Hennessy Remy Martin component VSOP XO g/L |
Cadmium 0.042 0.113 cobalt 0.069 0.069 chromium 0.249 0.121 iron 1.405 0.379 manganese 0.069 0.019 nickel 0.153 0.093 bronze medal 1.391 0.754 magnesium 1.109 0.734 calcium 2.892 3.197 tin 11.171 6.298 lithiums 0.132 0.193 sodium 3.412 2.177 potassium 0.013 0.004 antimony 0.112 0.221 arsenic 0.009 0.003 barium 0.102 0.048 plumbous 0.048 0.033 selenium, 0.019 0.008 zinc, 0.992 0.191 mercury 0.008 0.006 | Furfural 15.22 45.87 benzyl carbinols 1.81 7.54 4-hydroxyls-3,5-diformazan chloro-benzaldehyde 5.76 31.32 n-hexyl alcohols 5.03 4.77 ethyl hexanoates 2.53 3.18 ethyl caprilates 12.83 19.43 sad 63.29 110.78 ethyl caprates, 0.59 23.39 capric acid, 72.56 0.14 tetradecylic acids, 2.77 2.46 hexadecylic acids, 26.69 44.12 diethyl succinates, 0.40 0.76 ethyl myristate, 0.08 0.13 ethyl palmitate, 0.23 0.24 ethyl linoleate, 0.77 0.78 ethyl laurate, 0.05 8.32 docosane, 0.65 0.91 hexacosane, 0.87 0.93 dodecoic acid, 21.34 0.14 saualanes 0.78 0.82 | Malic acid 0.0798 0.0771 succinic acids 0.2523 0.2298 tartrate 0.1098 0.1305 citric acids 0.5965 0.7943 lactic acid 0.4587 0.3989 |
When identifying, import the vector of metal element content, organic acid content and the volatile components content of Hennessy VSOP and Remy Martin XO respectively, after the conversion of counterpropagation network and competitive function layer, the most significant neuron is won and is as a result of showed, unique boolean of output vector relevant position generation " 1 " is output as the title of corresponding wine, and other vector element is all " 0 " output, and is as shown in table 3, check the title of the wine of training objective correspondence, qualification result is accurate.
Table 3
Hennessy VSOP → ans=1000000000
0??0
Remy Martin XO → ans=0000010000
0??0
Claims (2)
1, a kind of France originates in the brandy authentication method, it is characterized in that with the known multiple organic acid of French brandy sample, multiple volatile components and multiple metal element content be the variable space, is the input matrix that the sample point space constitutes competitive back-propagation artificial neural network with known French brandy sample; This competitiveness reverse transmittance nerve network is made of front end reverse transmittance nerve network and rear end competitive function layer, with the order of known French brandy sample is that the unit matrix of order is the network training target, with known French brandy sample input matrix and simulate noisy vocal input matrix and network is connected weights train, set up network and connect; The neural network that the vector input that the organic acid of French brandy to be identified, volatile components and metal element content are constituted is set up, after the conversion of counterpropagation network and competitive function layer, the most significant neuron is won and is as a result of showed, and the title of checking the French brandy of training objective correspondence obtains final qualification result.
2, authentication method as claimed in claim 1 is characterized in that described multiple organic acid is malic acid, succinic acid, citric acid, tartrate, lactic acid; Described multiple volatile components is 2 furan carboxyaldehyde, n-hexyl alcohol, ethyl hexanoate, phenylethyl alcohol, diethyl succinate, sad, ethyl caprilate, capric acid, ethyl caprate, dodecoic acid, ethyl laurate, 4-hydroxyl-3,5-dimethoxy-benzaldehyde, tetradecanoic acid, ethyl myristate, hexadecanoic acid, ethyl palmitate, ethyl linoleate, docosane, hexacosane, squalane; Described multiple metallic element is arsenic, antimony, barium, selenium, lead, copper, nickel, cadmium, cobalt, zinc, potassium, sodium, lithium, tin, calcium, manganese, magnesium, iron, chromium, mercury.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410024379 CN1595164A (en) | 2004-06-22 | 2004-06-22 | Method for verifying original brandy made in French |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410024379 CN1595164A (en) | 2004-06-22 | 2004-06-22 | Method for verifying original brandy made in French |
Publications (1)
Publication Number | Publication Date |
---|---|
CN1595164A true CN1595164A (en) | 2005-03-16 |
Family
ID=34663468
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200410024379 Pending CN1595164A (en) | 2004-06-22 | 2004-06-22 | Method for verifying original brandy made in French |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1595164A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103645144A (en) * | 2013-12-23 | 2014-03-19 | 华中科技大学 | Method for quantitatively analyzing components in white spirit |
CN110009053A (en) * | 2019-04-12 | 2019-07-12 | 浙江树人学院(浙江树人大学) | A kind of comprehensive classification judgment method of the yellow rice wine based on BP deep neural network |
-
2004
- 2004-06-22 CN CN 200410024379 patent/CN1595164A/en active Pending
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103645144A (en) * | 2013-12-23 | 2014-03-19 | 华中科技大学 | Method for quantitatively analyzing components in white spirit |
CN103645144B (en) * | 2013-12-23 | 2017-01-18 | 华中科技大学 | Method for quantitatively analyzing components in white spirit |
CN110009053A (en) * | 2019-04-12 | 2019-07-12 | 浙江树人学院(浙江树人大学) | A kind of comprehensive classification judgment method of the yellow rice wine based on BP deep neural network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Xie et al. | Familiarity increases the number of remembered Pokémon in visual short-term memory | |
Gibson et al. | The development impact of a best practice seasonal worker policy | |
Campbell | The american voter | |
Camerer | Individual decision making | |
Blinder | The quiet revolution: Central banking goes modern | |
Frieden et al. | The currency game: exchange rate politics in Latin America | |
Schecter et al. | Game theory in action: an introduction to classical and evolutionary models | |
Person | A complete guide to technical trading tactics: how to profit using pivot points, candlesticks & other indicators | |
Brobakk | A climate for change? Norwegian farmers’ attitudes to climate change and climate policy | |
Alemanni et al. | Do investors rely on robots? Evidence from an experimental study | |
Sayrs | Trade and Conflict Revisited: Do Politics Matter? | |
Carlsen | Opinion polls and political business cycles: Theory and evidence for the United States | |
CN1595164A (en) | Method for verifying original brandy made in French | |
Anderson et al. | Risk aversion and tacit collusion in a Bertrand duopoly experiment | |
Lang et al. | Intergenerational transmission of gambling: Links between young adult and perceived grandparent gambling attitudes and behavior | |
Shiller et al. | What Have They Been Thinking? Home Buyer Behavior in Hot and Cold Markets: A Ten-Year Retrospect | |
Tideman | Gross national happiness: towards Buddhist economics | |
Langlois et al. | Rational investing: the subtleties of asset management | |
CN111582521A (en) | Deep multi-element transparentization trial method and system | |
Muzerengi et al. | The Sankofa Methodology: A Pan-African Approach to Poverty Alleviation | |
Iaquinto | The Impact of Intended Strategies on Firm Performance: The Case of Japanese Manufactures | |
TWI733456B (en) | Matching system of game nature and buyer’s suitability | |
Oleskog Tryggvason | Under the influence? Understanding media’s coverage of opinion polls and their effects on citizens and politicians | |
MacWilliam | Bonapartism in the South Pacific: The Bainimarama Government in Fiji | |
Geller | Money Anxiety |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C12 | Rejection of a patent application after its publication | ||
RJ01 | Rejection of invention patent application after publication |