CN101788517A - Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction - Google Patents
Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction Download PDFInfo
- Publication number
- CN101788517A CN101788517A CN201010115650A CN201010115650A CN101788517A CN 101788517 A CN101788517 A CN 101788517A CN 201010115650 A CN201010115650 A CN 201010115650A CN 201010115650 A CN201010115650 A CN 201010115650A CN 101788517 A CN101788517 A CN 101788517A
- Authority
- CN
- China
- Prior art keywords
- crude drug
- chinese crude
- smell
- sample
- flavor chinese
- 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
Images
Landscapes
- Medicines Containing Plant Substances (AREA)
Abstract
The invention discloses pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction. The pungent traditional Chinese medicine odor fingerprint map construction system based on bionic olfaction comprises a sensor array (2), a signal modulation module (3), a digital analog conversion circuit (4) and a computer information processing system (5), wherein the sensor array (2) is arranged on the position which can detect the odor of pungent traditional Chinese medicines, the input end of the signal modulation module (3) is connected with the output end of the sensor array (2), the output end of the signal modulation module (3) is connected with the input end of the digital analog conversion circuit (4), and the output end of the digital analog conversion circuit (4) is connected with the computer information processing system (5). The pungent traditional Chinese medicine odor fingerprint map construction system based on bionic olfaction has the advantages of simple structure, low cost and objective and reliable determination result. The pungent traditional Chinese medicine odor fingerprint map construction method based on bionic olfaction is simple and convenient.
Description
Technical field
The method for designing that the present invention relates to a kind of conducting polymer gas sensor array is come anthropomorphic dummy's olfactory cell function, survey the smell information of hot flavor Chinese crude drug, and cutting edge technology Design Mode recognizers such as utilization multivariate statistical analysis method and artificial intelligence analyze and then make up the suffering flavor Chinese crude drug smell fingerprint map construction system and the method based on bionic olfactory of its smell finger-print to the smell information of being obtained, and belong to based on the suffering flavor Chinese crude drug smell fingerprint map construction system of bionic olfactory and the renovation technique of method.
Background technology
The traditional discriminating of Chinese crude drug exactly with soon, empirical method such as hand touches, nose is heard, mouth tasted, water tries, fire tries, the shape by the observation Chinese crude drug, size, smell, quality, change color etc. are with the true and false quality of judgement Chinese crude drug.Yet, to be subjected to factor affecting such as physiology, experience, mood, environment inevitably based on the experience differential method of human body, its subjectivity is strong, poor repeatability, accuracy is low, is difficult to the formation standard.For in line with international standards, guarantee the Chinese crude drug quality, 20 end of the centurys, the Chinese government has proposed this call of the modernization of Chinese medicine, National Drug Administration proposes with building on the system testing of traditional Chinese medicine ingredients, the quality standard that the standard finger-print on the analysis result is estimated Chinese crude drug, Chinese patent drug for describing and control traditional Chinese medicine quality quantitatively, all sidedly.
The research of traditional Chinese medicine fingerprint is a research focus in recent years.At present, traditional Chinese medicine fingerprint structure side commonly used mainly contains chromatogram and spectral method, and chromatography method commonly used is divided into thin-layered chromatography (TLC) finger-print, vapor-phase chromatography (GC) finger-print, high performance liquid chromatography (HPLC) finger-print, electrophoresis (CE) finger-print, high-performance capillary electrophoresis (HPCE) finger-print again; The method that spectral method is commonly used is divided into infra-red sepectrometry (IR) finger-print, tandem mass spectrometry (MS/MS) finger-print, nuclear magnetic resonance spectroscopy (NMR) finger-print, X-ray diffraction method (XRD) finger-print.
Vapor-phase chromatography is by the gas information of gathering Chinese crude drug it to be analyzed, but it is that the principal ingredient in the smell is analyzed, discern the occupancy volume of its main chemical compositions, thereby draw Chinese crude drug true and false quality, its starting point is the occupancy volume of each chemical constitution of Chinese crude drug.Its mensuration need prepare and handle sample, select suitable extraction solvent and suitable chromatographic separation condition, and wants dozens of minutes to a hour analysis time usually.Because volatile matter concentration is generally all lower, can not meet the demands sometimes with the vapor-phase chromatography detection, and usually can't be qualitative as if no standard specimen to unknown material.Promptly enable qualitatively, also can only obtain the discriminating of certain or certain several composition, rather than the smell information of sense of smell integral body.And many gases, the great majority of its composition are in outside the sensing range of instruments such as gas chromatograph, and the relation between the chemical property of scent molecule and the sense of smell effect still is a "black box", only can not express the quality of measured matter all sidedly with several chemical constitutions of measuring.Adopt spectral method to make up collection of illustrative plates, its instrument involves great expense, and data mining work remains development, is difficult at present apply.
Summary of the invention
The objective of the invention is to consider the problems referred to above and provide a kind of simple in structure, the suffering flavor Chinese crude drug smell fingerprint map construction system that cost is low based on bionic olfactory.
Another object of the present invention is to provide a kind of simply and easily based on the method for the suffering flavor Chinese crude drug smell fingerprint map construction of bionic olfactory.
Technical scheme of the present invention is: the suffering flavor Chinese crude drug smell fingerprint map construction system that the present invention is based on bionic olfactory, include sensor array, signal condition module, D/A converting circuit, computer information processing system, wherein sensor array places on the position that can detect hot flavor Chinese crude drug smell, the input end of signal condition module is connected with the output terminal of sensor array, the output terminal of signal condition module is connected with the input end of D/A converting circuit, and the output terminal of D/A converting circuit is connected with computer information processing system.
Above-mentioned hot flavor Chinese crude drug places the sample bottle in the automatic sampler, automatic sampler vibrates sample bottle and heats, generation has repeatable suffering flavor Chinese crude drug head space gas, automatically be delivered to then in the sensor array, sensor array places on the position that can detect hot flavor Chinese crude drug smell in the automatic sampler.
The sensor array is an array type metal oxide type sensor.
Above-mentioned signal condition module comprises signal amplification circuit and filtering circuit; The output terminal of aforementioned calculation machine information disposal system also is connected with output display device.
The present invention is based on the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system of bionic olfactory, it comprises the steps:
1) with the smell information of sensor array detection/collection known hot flavor Chinese crude drug as much as possible, obtains response modes to various smells;
2) the smell information that detection is obtained is carried out the signal pre-treatment, to reach drift compensation, Information Compression and noise filtering;
3) the smell information through pre-treatment is carried out aftertreatment, effective recognition feature of selection/extraction hot flavor Chinese crude drug smell information to be measured, reduce the measurement space dimension and extract the information relevant, thereby make up the smell fingerprints database of known hot flavor Chinese crude drug with pattern-recognition;
4) Chinese crude drug smell information to be identified is detected and forms according to the method described above its smell finger-print.
5) the mutual comparative analysis of smell finger-print known in acquired Chinese crude drug smell finger-print to be identified and the database is provided the identification evaluation result.
Above-mentioned steps 5) recognition methods of identification evaluation result is adopted based on the ANN mode discriminator of artificial neural network technology blending inheritance Evolution Theory algorithm from the smell recognizer of establishment, the ANN mode discriminator is that the artificial neural network information processing is to be realized by the interaction between the neuron, the storage of knowledge and information shows as between network element interconnection distributed physics and gets in touch, discern the dynamic evolutionary process that depends on the neuron connection weight in the e-learning, use genetic algorithm to the evolve optimization of formula of the weighting parameter of artificial neural network and network structure, training, in the hope of finding the most suitable hot flavor Chinese crude drug odor characteristics data, be weights and the network structure that fingerprint map analyzing uses, make it become one and can work in the recognizer that hot flavor Chinese crude drug is differentiated classification.
Above-mentioned steps 2) process of signal pre-treatment is as follows:
A) sensor array adopts metal oxide type sensor array, after metal oxide type sensor array is listed in and detects hot flavor Chinese crude drug smell, its conductivity changes, being that high frequency noise takes place to change relatively and have resistance value, is that curtage changes with signal conditioning circuit with this conversion of signals;
B) with microprocessor it is sampled, the data after sampling are carried out digital filtering, to eliminate high frequency noise with wave filter; Wave filter can make response curve reach the level and smooth of the best by the filtering window size is set, and bigger distortion not occur by the level and smooth response curve of polynomial approximation.
C) filtered data are again with the degraded of baseline correction method, baseline electrical resistance value R
0Calculate the mean value R of maximum resistance in the sample dispose procedure by counting point with the emptying post-sampling before the sample release
MaxCalculated by sampled point in the absolute value of maximum resistance and the sample dispose procedure, therefore, the response of sensor is defined as (R
Max-R
0)/R
0=Δ R/R
0
Above-mentioned steps 3) process of smell information aftertreatment: computer system will obtain the feature mode of the known hot flavor Chinese crude drug smell information of reflection by study, it is the characteristic matrix, further form corresponding Chinese crude drug smell finger-print, Chinese crude drug smell finger-print is stored in computer system with the form of data matrix.
Above-mentioned steps 1) choosing method of hot flavor Chinese crude drug is as follows: according to factor classification samplings such as the cultivar origin of suffering flavor traditional Chinese medicinal material samples, the places of production, and pay attention to the representativeness and the homogeneity of sample, in the medicinal material of choosing genunie medicinal materials are arranged, the medicinal material of the same race that the different places of production are also arranged, so not only can study various genunie medicinal materials, also can study simultaneously different places of production medicinal material of the same race, sample guarantees its authenticity, complete sampling raw readings is arranged, and for number and the amount that guarantees sample has enough representativenesses, every kind of sample is all collected 10 batches; The experiment pre-treating method of sample is as follows: this method only needs simple sample pre-treatments can obtain stable sample headspace gas; Before hand sampling, put the hot flavor of the solid traditional Chinese medicinal material samples after grinding is weighed into sample bottle, adopt sealing film that bottleneck is sealed again, make its generation have the sample headspace gas of repeatability, hand sampling gets final product in sensor array then; Under the auto injection condition: the sample after grinding weighed is put the automatic sampler that has sample bottle into, automatic sampler will vibrate voluntarily and heat sample bottle, generation has the suffering flavor Chinese crude drug head space gas of repeatability, is delivered in the sensor array automatically then.
Above-mentioned steps 4) construction method of smell finger-print is:
Hot flavor Chinese crude drug smell finger-print is a visual image that characterizes corresponding Chinese crude drug smell data characteristics, and each collection of illustrative plates image is represented a kind of smell pattern of medicinal material, is to store in the mode of characteristic matrix on computers;
Suppose that certain class Chinese crude drug is being expressed as by sensor array sampling n time and through the sample space after the pre-treatment
X=(x
1,x
2,...,x
n)
T (1)
X wherein
iIt is a dimensional vector.
Can solve the classification problem of the relatively poor sample of linear separability preferably by means of kernel function;
This method adopts the radially kernel function of fundamental mode of Gauss
k(x,y)=exp(-||x-y||
2/(2σ
2)) (2)
The use of kernel function will make the sample data space implicitly be expressed as feature space, by means of the kernel function Nonlinear Mapping primitive character space s be transformed to a high-dimensional feature space F, sample data can be expressed as in F
Y=(y
1,y
2,...,y
m)
T (3)
Wherein, m is the dimension of sample in high-dimensional feature space F;
Use the discriminating criterion of multivariate statistics classification again in high-dimensional feature space F, this method is mapped as a few comprehensive characteristics by principal component analysis (PCA) with high dimensional feature once more in the F of space, forms the proper vector of hot flavor Chinese crude drug smell pattern
Z=(z
1,z
2,...z
l) (4)
Wherein, l is the principal component number, and l<m.The number of principal component is determined by contribution rate of accumulative total, for example can surpass 85% with contribution rate of accumulative total is that standard is selected the principal component number, promptly when contribution rate of accumulative total has reached 85%, then principal component can be defined as l, its expression, a selected l principal component has kept most information that original m ties up variable;
Z is the odor characteristics vector of corresponding hot flavor Chinese crude drug sample, knowledge base is analyzed and information fusion by differentiating with the suffering flavor Chinese crude drug expert who is set up, constantly Z is carried out feedback modifiers, finally form its smell finger-print, and be stored in the computing machine in the mode of suffering flavor Chinese crude drug odor characteristics database.
The suffering flavor Chinese crude drug smell fingerprint map construction that the present invention is based on bionic olfactory is by comprehensive smell information of volatility that detects hot flavor Chinese crude drug and then the finger-print of determining its smell with discrimination method, based on distinguish the flavor of Chinese crude drug smell fingerprint map construction and discrimination method and vapor-phase chromatography of the suffering of bionic olfactory in essence difference is arranged.Compare with vapor-phase chromatography, the operation of bionic olfactory method is fast and convenient, and sample only needs simple pre-treatment, does not also need any organic solvent to extract, and therefore being one helps environmental protection, does not influence " green " analytical technology of operator ' s health.The recognition reaction that this method has artificial intelligence to unknown sample is compared with methods such as people's sense of smell or gas chromatographies, and measurement result is more objective, reliable.Bionic olfactory method utilization conducting polymer gas sensor array comes anthropomorphic dummy's olfactory cell function, adopt the nerve of artificial neural networks isotype recognition technology simulation human brain to handle maincenter, the bionic olfactory method is an artificial intelligence approach, and it has the characteristic of " bionical ".The bionic olfactory method is used the smell of sensor array " news " Chinese crude drug, draws the smell finger-print of Chinese crude drug at last, and its starting point is the comprehensive smell information of the volatility of Chinese crude drug." news " of sensor array is the comprehensive smell information that obtains Chinese crude drug, and its function is equivalent to people's nose, and experienced traditional Chinese physician adopts " smelling " method to pass judgment on Chinese crude drug as same experience.The present invention is that a kind of design is ingenious, function admirable, the convenient and practical suffering flavor Chinese crude drug smell fingerprint map construction method based on bionic olfactory.
Description of drawings
Fig. 1 is bionic olfactory system and human body olfactory system contrast figure;
Fig. 2 is the suffering flavor Chinese crude drug odor detection schematic diagram that the present invention is based on bionic olfactory;
Fig. 3 is the suffering flavor Chinese crude drug smell fingerprint map construction process flow diagram that the present invention is based on bionic olfactory.
Embodiment
Embodiment:
The present invention is based on bionic olfactory suffering flavor Chinese crude drug smell fingerprint map construction system schematic diagram as shown in Figure 2, include sensor array 2, signal condition module 3, D/A converting circuit 4, computer information processing system 5, wherein sensor array 2 places on the position that can detect hot flavor Chinese crude drug smell, the input end of signal condition module 3 is connected with the output terminal of sensor array 2, the output terminal of signal condition module 3 is connected with the input end of D/A converting circuit 4, and the output terminal of D/A converting circuit 4 is connected with computer information processing system 5.
In the present embodiment, above-mentioned hot flavor Chinese crude drug places the sample bottle in the automatic sampler 1,1 pair of sample bottle of automatic sampler vibrates and heats, generation has repeatable suffering flavor Chinese crude drug head space gas, automatically be delivered to then in the sensor array 2, sensor array 2 places on the position that can detect hot flavor Chinese crude drug smell in the automatic sampler 1.
In the present embodiment, the sensor array 2 is an array type metal oxide type sensor.Above-mentioned signal condition module 3 comprises signal amplification circuit and filtering circuit; The output terminal of aforementioned calculation machine information disposal system 5 also is connected with output display device 6.
The present invention is based on the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system of bionic olfactory, it comprises the steps:
1) with the smell information of sensor array 2 detection/collections known hot flavor Chinese crude drug as much as possible, obtains response modes to various smells;
2) the smell information that detection is obtained is carried out the signal pre-treatment, to reach drift compensation, Information Compression and noise filtering;
3) the smell information through pre-treatment is carried out aftertreatment, effective recognition feature of selection/extraction hot flavor Chinese crude drug smell information to be measured, reduce the measurement space dimension and extract the information relevant, thereby make up the smell fingerprints database of known hot flavor Chinese crude drug with pattern-recognition;
4) Chinese crude drug smell information to be identified is detected and forms according to the method described above its smell finger-print.
5) the mutual comparative analysis of smell finger-print known in acquired Chinese crude drug smell finger-print to be identified and the database is provided the identification evaluation result.
Above-mentioned steps 5) recognition methods of identification evaluation result is adopted based on the ANN mode discriminator of artificial neural network technology blending inheritance Evolution Theory algorithm from the smell recognizer of establishment, the ANN mode discriminator is that the artificial neural network information processing is to be realized by the interaction between the neuron, the storage of knowledge and information shows as between network element interconnection distributed physics and gets in touch, discern the dynamic evolutionary process that depends on the neuron connection weight in the e-learning, use genetic algorithm to the evolve optimization of formula of the weighting parameter of artificial neural network and network structure, training, in the hope of finding the most suitable hot flavor Chinese crude drug odor characteristics data, be weights and the network structure that fingerprint map analyzing uses, make it become one and can work in the recognizer that hot flavor Chinese crude drug is differentiated classification.
Above-mentioned steps 2) process of signal pre-treatment is as follows:
A) sensor array 2 adopts metal oxide type sensor array, after metal oxide type sensor array is listed in and detects hot flavor Chinese crude drug smell, its conductivity changes, being that high frequency noise takes place to change relatively and have resistance value, is that curtage changes with signal conditioning circuit with this conversion of signals;
B) with microprocessor it is sampled, the data after sampling are carried out digital filtering, to eliminate high frequency noise with wave filter; Wave filter can make response curve reach the level and smooth of the best by the filtering window size is set, and bigger distortion not occur by the level and smooth response curve of polynomial approximation.
C) filtered data are again with the degraded of baseline correction method, baseline electrical resistance value R
0Calculate the mean value R of maximum resistance in the sample dispose procedure by counting point with the emptying post-sampling before the sample release
MaxCalculated by sampled point in the absolute value of maximum resistance and the sample dispose procedure, therefore, the response of sensor is defined as (R
Max-R
0)/R
0=Δ R/R
0
Above-mentioned steps 3) process of smell information aftertreatment: computer system will obtain the feature mode (characteristic matrix) of the known hot flavor Chinese crude drug smell information of reflection by study, further form corresponding Chinese crude drug smell finger-print (Chinese crude drug smell finger-print is stored in computer system with the form of data matrix).
Above-mentioned steps 1) choosing method of hot flavor Chinese crude drug is as follows: according to factor classification samplings such as the cultivar origin of suffering flavor traditional Chinese medicinal material samples, the places of production, and pay attention to the representativeness and the homogeneity of sample, in the medicinal material of choosing genunie medicinal materials are arranged, the medicinal material of the same race that the different places of production are also arranged, so not only can study various genunie medicinal materials, also can study simultaneously different places of production medicinal material of the same race, sample guarantees its authenticity, complete sampling raw readings is arranged, and for number and the amount that guarantees sample has enough representativenesses, every kind of sample is all collected 10 batches; The experiment pre-treating method of sample is as follows: this method only needs simple sample pre-treatments can obtain stable sample headspace gas; Before hand sampling, will be after grinding be weighed the hot flavor of the solid traditional Chinese medicinal material samples of (sample abrasive grains size is fixed as 20 orders in this method, quality is 15 grams) to put capacity into be that (cumulative volume is no more than half bottle to the 250ml sample bottle, the sample bottle specification is determined, be respectively 70mm, 140mm, 30mm as external diameter, the high and bore of bottle), adopt sealing film with bottleneck sealing 30 minutes again, make its generation have the sample headspace gas of repeatability, hand sampling gets final product in sensor array then; Under the auto injection condition: the sample after grinding weighed is put the automatic sampler that has sample bottle into, automatic sampler will vibrate voluntarily and heat sample bottle, generation has the suffering flavor Chinese crude drug head space gas of repeatability, is delivered in the sensor array automatically then.
Above-mentioned steps 4) construction method of smell finger-print is:
Hot flavor Chinese crude drug smell finger-print is a visual image that characterizes corresponding Chinese crude drug smell data characteristics, and each collection of illustrative plates image is represented a kind of smell pattern of medicinal material, is to store in the mode of characteristic matrix on computers;
Suppose that certain class Chinese crude drug is being expressed as by sensor array sampling n time and through the sample space after the pre-treatment
X=(x
1,x
2,...,x
n)
T (1)
X wherein
iIt is a dimensional vector.
Can solve the classification problem of the relatively poor sample of linear separability preferably by means of kernel function;
This method adopts the radially kernel function of fundamental mode of Gauss
k(x,y)=exp(-||x-y||
2(2σ
2)) (2)
The use of kernel function will make the sample data space implicitly be expressed as feature space, by means of the kernel function Nonlinear Mapping primitive character space s be transformed to a high-dimensional feature space F, sample data can be expressed as in F
Y=(y
1,y
2,...,y
m)
T (3)
Wherein, m is the dimension of sample in high-dimensional feature space F;
Use the discriminating criterion of multivariate statistics classification again in high-dimensional feature space F, this method is mapped as a few comprehensive characteristics by principal component analysis (PCA) with high dimensional feature once more in the F of space, forms the proper vector of hot flavor Chinese crude drug smell pattern
Z=(z
1,z
2,...z
l) (4)
Wherein, l is the principal component number, and l<m.The number of principal component is determined by contribution rate of accumulative total, for example can surpass 85% with contribution rate of accumulative total is that standard is selected the principal component number, promptly when contribution rate of accumulative total has reached 85%, then principal component can be defined as l, its expression, a selected l principal component has kept most information that original m ties up variable;
Z is the odor characteristics vector of corresponding hot flavor Chinese crude drug sample, knowledge base is analyzed and information fusion by differentiating with the suffering flavor Chinese crude drug expert who is set up, constantly Z is carried out feedback modifiers, finally form its smell finger-print, and be stored in the computing machine in the mode of suffering flavor Chinese crude drug odor characteristics database.
Claims (10)
1. one kind based on the suffering of bionic olfactory flavor Chinese crude drug smell fingerprint map construction system, it is characterized in that including sensor array (2), signal condition module (3), D/A converting circuit (4), computer information processing system (5), wherein sensor array (2) places on the position that can detect hot flavor Chinese crude drug smell, the input end of signal condition module (3) is connected with the output terminal of sensor array (2), the output terminal of signal condition module (3) is connected with the input end of D/A converting circuit (4), and the output terminal of D/A converting circuit (4) is connected with computer information processing system (5).
2. the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 1, it is characterized in that above-mentioned hot flavor Chinese crude drug places the sample bottle in the automatic sampler (1), automatic sampler (1) vibrates sample bottle and heats, generation has repeatable suffering flavor Chinese crude drug head space gas, automatically be delivered to then in the sensor array (2), sensor array (2) places on the position that can detect hot flavor Chinese crude drug smell in the automatic sampler (1).
3. the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 1 is characterized in that the sensor array (2) is an array type metal oxide type sensor.
4. the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 1 is characterized in that above-mentioned signal condition module (3) comprises signal amplification circuit and filtering circuit; The output terminal of aforementioned calculation machine information disposal system (5) also is connected with output display device (6).
5. the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 1 is characterized in that comprising the steps:
1) with the smell information of sensor array (2) detection/collection known hot flavor Chinese crude drug as much as possible, obtains response modes to various smells;
2) the smell information that detection is obtained is carried out the signal pre-treatment, to reach drift compensation, Information Compression and noise filtering;
3) the smell information through pre-treatment is carried out aftertreatment, effective recognition feature of selection/extraction hot flavor Chinese crude drug smell information to be measured, reduce the measurement space dimension and extract the information relevant, thereby make up the smell fingerprints database of known hot flavor Chinese crude drug with pattern-recognition;
4) Chinese crude drug smell information to be identified is detected and forms according to the method described above its smell finger-print.
5) the mutual comparative analysis of smell finger-print known in acquired Chinese crude drug smell finger-print to be identified and the database is provided the identification evaluation result.
6. the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 5, it is characterized in that above-mentioned steps 5) discern the recognition methods employing of evaluation result based on the ANN mode discriminator of artificial neural network technology blending inheritance Evolution Theory algorithm from the smell recognizer of establishment, the ANN mode discriminator is that the artificial neural network information processing is to be realized by the interaction between the neuron, the storage of knowledge and information shows as between network element interconnection distributed physics and gets in touch, discern the dynamic evolutionary process that depends on the neuron connection weight in the e-learning, use genetic algorithm to the evolve optimization of formula of the weighting parameter of artificial neural network and network structure, training, in the hope of finding the most suitable hot flavor Chinese crude drug odor characteristics data, be weights and the network structure that fingerprint map analyzing uses, make it become one and can work in the recognizer that hot flavor Chinese crude drug is differentiated classification.
7. the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 5 is characterized in that above-mentioned steps 2) process of signal pre-treatment is as follows:
A) sensor array (2) adopts metal oxide type sensor array, after metal oxide type sensor array is listed in and detects hot flavor Chinese crude drug smell, its conductivity changes, being that high frequency noise takes place to change relatively and have resistance value, is that curtage changes with signal conditioning circuit with this conversion of signals;
B) with microprocessor it is sampled, the data after sampling are carried out digital filtering, to eliminate high frequency noise with wave filter; Wave filter can make response curve reach the level and smooth of the best by the filtering window size is set, and bigger distortion not occur by the level and smooth response curve of polynomial approximation.
C) filtered data are counted point with the emptying post-sampling before baseline electrical resistance value R0 is discharged by sample and are calculated the mean value R of maximum resistance in the sample dispose procedure again with the degraded of baseline correction method
MaxCalculated by sampled point in the absolute value of maximum resistance and the sample dispose procedure, therefore, the response of sensor is defined as (R
Max-R
0)/R
0=Δ R/R
0
8. the member method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 5, it is characterized in that above-mentioned steps 3) process of smell information aftertreatment: computer system will obtain the feature mode of the known hot flavor Chinese crude drug smell information of reflection by study, it is the characteristic matrix, further form corresponding Chinese crude drug smell finger-print, Chinese crude drug smell finger-print is stored in computer system with the form of data matrix.
9. the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 5, it is characterized in that above-mentioned steps 1) choosing method of hot flavor Chinese crude drug is as follows: according to the cultivar origin of suffering flavor traditional Chinese medicinal material samples, factors such as place of production classification sampling, and pay attention to the representativeness and the homogeneity of sample, in the medicinal material of choosing genunie medicinal materials are arranged, the medicinal material of the same race that the different places of production are also arranged, so not only can study various genunie medicinal materials, also can study simultaneously different places of production medicinal material of the same race, sample guarantees its authenticity, complete sampling raw readings is arranged, and for number and the amount that guarantees sample has enough representativenesses, every kind of sample is all collected 10 batches; The experiment pre-treating method of sample is as follows: this method only needs simple sample pre-treatments can obtain stable sample headspace gas; Before hand sampling, put the hot flavor of the solid traditional Chinese medicinal material samples after grinding is weighed into sample bottle, adopt sealing film that bottleneck is sealed again, make its generation have the sample headspace gas of repeatability, hand sampling gets final product in sensor array then; Under the auto injection condition: the sample after grinding weighed is put the automatic sampler that has sample bottle into, automatic sampler will vibrate voluntarily and heat sample bottle, generation has the suffering flavor Chinese crude drug head space gas of repeatability, is delivered in the sensor array automatically then.
10. the construction method of the suffering flavor Chinese crude drug smell fingerprint map construction system based on bionic olfactory according to claim 1 is characterized in that above-mentioned steps 4) construction method of smell finger-print is:
Hot flavor Chinese crude drug smell finger-print is a visual image that characterizes corresponding Chinese crude drug smell data characteristics, and each collection of illustrative plates image is represented a kind of smell pattern of medicinal material, is to store in the mode of characteristic matrix on computers;
Suppose that certain class Chinese crude drug is being expressed as by sensor array sampling n time and through the sample space after the pre-treatment
X=(x
1,x
2,...,x
n)
T (1)
X wherein
iIt is a dimensional vector.
Can solve the classification problem of the relatively poor sample of linear separability preferably by means of kernel function;
This method adopts the radially kernel function of fundamental mode of Gauss
k(x,y)=exp(-||x-y||
2/(2σ
2)) (2)
The use of kernel function will make the sample data space implicitly be expressed as feature space, by means of the kernel function Nonlinear Mapping primitive character space s be transformed to a high-dimensional feature space F, sample data can be expressed as in F
Y=(y
1,y
2,...,y
m)
T (3)
Wherein, m is the dimension of sample in high-dimensional feature space F;
Use the discriminating criterion of multivariate statistics classification again in high-dimensional feature space F, this method is mapped as a few comprehensive characteristics by principal component analysis (PCA) with high dimensional feature once more in the F of space, forms the proper vector of hot flavor Chinese crude drug smell pattern
Z=(z
1,z
2,...z
l) (4)
Wherein, l is the principal component number, and l<m.The number of principal component is determined by contribution rate of accumulative total;
Z is the odor characteristics vector of corresponding hot flavor Chinese crude drug sample, knowledge base is analyzed and information fusion by differentiating with the suffering flavor Chinese crude drug expert who is set up, constantly z is carried out feedback modifiers, finally form its smell finger-print, and be stored in the computing machine in the mode of suffering flavor Chinese crude drug odor characteristics database.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010115650A CN101788517A (en) | 2010-02-23 | 2010-02-23 | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201010115650A CN101788517A (en) | 2010-02-23 | 2010-02-23 | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101788517A true CN101788517A (en) | 2010-07-28 |
Family
ID=42531811
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201010115650A Pending CN101788517A (en) | 2010-02-23 | 2010-02-23 | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101788517A (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102435713A (en) * | 2010-09-21 | 2012-05-02 | 成都中医药大学 | Automatic detection system for quality of traditional Chinese medicine |
CN102680650A (en) * | 2012-05-14 | 2012-09-19 | 上海鼎为软件技术有限公司 | Odor identification terminal, odor diffusion terminal and information communication system |
CN103063815A (en) * | 2012-12-31 | 2013-04-24 | 广东工业大学 | Rapid identification method for origin places of pungent traditional Chinese medicines |
CN103645284A (en) * | 2013-12-13 | 2014-03-19 | 林丽君 | Quick smell fingerprint detection method based on improved RANSAC (random sample consensus) theory |
CN104501925A (en) * | 2014-04-21 | 2015-04-08 | 四川大学 | Simulation lowpass filter circuit of medicine bottling line dynamic weighing system |
CN104569313A (en) * | 2015-01-15 | 2015-04-29 | 上海应用技术学院 | Method for rapidly analyzing fragrance of bayberry juice based on electronic nose smell fingerprint information |
CN106662517A (en) * | 2014-08-29 | 2017-05-10 | 株式会社而摩比特 | Odour detection system, odour identification device, and odour identification method |
CN106918626A (en) * | 2017-04-14 | 2017-07-04 | 中国石油化工股份有限公司 | Dangerous atmosphere comprehensive state fingerprint identification method |
CN106940340A (en) * | 2017-04-14 | 2017-07-11 | 中国石油化工股份有限公司 | Dangerous atmosphere comprehensive state fingerprint recognition system |
CN107607604A (en) * | 2017-07-17 | 2018-01-19 | 宁波大学 | A kind of gas-detecting device and method based on PCA |
CN107991460A (en) * | 2017-11-16 | 2018-05-04 | 吉林大学 | A kind of soil nutrient device for fast detecting based on bionic olfactory |
CN108226053A (en) * | 2017-12-22 | 2018-06-29 | 南京中医药大学 | Unstructured feature differentiation method based on multi-element intelligent fusion |
CN108333300A (en) * | 2018-01-05 | 2018-07-27 | 中国食品发酵工业研究院有限公司 | Method for differentiating the maca place of production |
CN108932483A (en) * | 2018-06-19 | 2018-12-04 | 广东工业大学 | A kind of odor identification method and device based on machine olfaction |
CN110163101A (en) * | 2019-04-17 | 2019-08-23 | 湖南省中医药研究院 | The difference of Chinese medicine seed and grade quick discrimination method |
CN110864792A (en) * | 2019-11-27 | 2020-03-06 | 武汉科技大学 | Traditional Chinese medicine weighing system control method and traditional Chinese medicine weighing system |
CN111954812A (en) * | 2017-12-08 | 2020-11-17 | 耶达研究及发展有限公司 | Utilization of electronic nose-based odorant analysis |
CN115684316A (en) * | 2022-11-08 | 2023-02-03 | 广州市中耕信息技术有限公司 | Odor identification method of bionic electronic nose based on MOS gas sensor array |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101261280A (en) * | 2007-12-29 | 2008-09-10 | 广东工业大学 | Traditional Chinese herb odor discriminate method based on bionic olfaction and its device |
CN101470121A (en) * | 2007-12-29 | 2009-07-01 | 广东工业大学 | Built-in bionic smell recognition method and device |
-
2010
- 2010-02-23 CN CN201010115650A patent/CN101788517A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101261280A (en) * | 2007-12-29 | 2008-09-10 | 广东工业大学 | Traditional Chinese herb odor discriminate method based on bionic olfaction and its device |
CN101470121A (en) * | 2007-12-29 | 2009-07-01 | 广东工业大学 | Built-in bionic smell recognition method and device |
Non-Patent Citations (4)
Title |
---|
《***工程》 20050228 吴今培 基于核函数的主成分分析及应用 第23卷, 第2期 * |
《食品研究与开发》 20061231 骆德汉等 仿生物嗅觉传感技术在卷烟品牌识别中的应用 第27卷, 第1期 * |
吴今培: "基于核函数的主成分分析及应用", 《***工程》, vol. 23, no. 2, 28 February 2005 (2005-02-28) * |
骆德汉等: "仿生物嗅觉传感技术在卷烟品牌识别中的应用", 《食品研究与开发》, vol. 27, no. 1, 31 December 2006 (2006-12-31) * |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102435713A (en) * | 2010-09-21 | 2012-05-02 | 成都中医药大学 | Automatic detection system for quality of traditional Chinese medicine |
CN102680650A (en) * | 2012-05-14 | 2012-09-19 | 上海鼎为软件技术有限公司 | Odor identification terminal, odor diffusion terminal and information communication system |
CN103063815A (en) * | 2012-12-31 | 2013-04-24 | 广东工业大学 | Rapid identification method for origin places of pungent traditional Chinese medicines |
CN103645284B (en) * | 2013-12-13 | 2015-09-30 | 阿默思(天津)科技发展有限公司 | A kind of quick smell fingerprint detection method based on improving RANSAC theory |
CN103645284A (en) * | 2013-12-13 | 2014-03-19 | 林丽君 | Quick smell fingerprint detection method based on improved RANSAC (random sample consensus) theory |
CN104501925A (en) * | 2014-04-21 | 2015-04-08 | 四川大学 | Simulation lowpass filter circuit of medicine bottling line dynamic weighing system |
CN111398529A (en) * | 2014-08-29 | 2020-07-10 | 株式会社而摩比特 | Olfactory system, smell recognition device and smell recognition method |
CN106662517A (en) * | 2014-08-29 | 2017-05-10 | 株式会社而摩比特 | Odour detection system, odour identification device, and odour identification method |
US11946904B2 (en) | 2014-08-29 | 2024-04-02 | Kabushikigaisha Aroma Bit, Inc. | Sample information acquisition unit, information processing unit, odor identification device |
CN104569313A (en) * | 2015-01-15 | 2015-04-29 | 上海应用技术学院 | Method for rapidly analyzing fragrance of bayberry juice based on electronic nose smell fingerprint information |
CN106918626A (en) * | 2017-04-14 | 2017-07-04 | 中国石油化工股份有限公司 | Dangerous atmosphere comprehensive state fingerprint identification method |
CN106940340A (en) * | 2017-04-14 | 2017-07-11 | 中国石油化工股份有限公司 | Dangerous atmosphere comprehensive state fingerprint recognition system |
CN106940340B (en) * | 2017-04-14 | 2023-03-28 | 中国石油化工股份有限公司 | Dangerous atmosphere comprehensive state fingerprint identification system |
CN107607604A (en) * | 2017-07-17 | 2018-01-19 | 宁波大学 | A kind of gas-detecting device and method based on PCA |
CN107991460A (en) * | 2017-11-16 | 2018-05-04 | 吉林大学 | A kind of soil nutrient device for fast detecting based on bionic olfactory |
CN111954812A (en) * | 2017-12-08 | 2020-11-17 | 耶达研究及发展有限公司 | Utilization of electronic nose-based odorant analysis |
US11674938B2 (en) | 2017-12-08 | 2023-06-13 | Yeda Research And Development Co. Ltd. | Utilization of electronic nose-based analysis of odorants |
CN108226053A (en) * | 2017-12-22 | 2018-06-29 | 南京中医药大学 | Unstructured feature differentiation method based on multi-element intelligent fusion |
CN108333300A (en) * | 2018-01-05 | 2018-07-27 | 中国食品发酵工业研究院有限公司 | Method for differentiating the maca place of production |
CN108932483A (en) * | 2018-06-19 | 2018-12-04 | 广东工业大学 | A kind of odor identification method and device based on machine olfaction |
CN108932483B (en) * | 2018-06-19 | 2022-03-25 | 广东工业大学 | Smell identification method and device based on machine olfaction |
CN110163101A (en) * | 2019-04-17 | 2019-08-23 | 湖南省中医药研究院 | The difference of Chinese medicine seed and grade quick discrimination method |
CN110163101B (en) * | 2019-04-17 | 2022-09-23 | 湖南省中医药研究院 | Method for rapidly distinguishing seeds of traditional Chinese medicinal materials and rapidly judging grades of seeds |
CN110864792A (en) * | 2019-11-27 | 2020-03-06 | 武汉科技大学 | Traditional Chinese medicine weighing system control method and traditional Chinese medicine weighing system |
CN115684316A (en) * | 2022-11-08 | 2023-02-03 | 广州市中耕信息技术有限公司 | Odor identification method of bionic electronic nose based on MOS gas sensor array |
CN115684316B (en) * | 2022-11-08 | 2023-12-22 | 中耕生态环境科技(广州)有限公司 | Smell recognition method of bionic electronic nose based on MOS gas sensor array |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101788517A (en) | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction | |
Banerjee et al. | Black tea classification employing feature fusion of E-Nose and E-Tongue responses | |
CN103134850B (en) | A kind of tea leaf quality method for quick based on characteristic perfume | |
Li et al. | A novel headspace integrated E-nose and its application in discrimination of Chinese medical herbs | |
CN101470121A (en) | Built-in bionic smell recognition method and device | |
CN101261280A (en) | Traditional Chinese herb odor discriminate method based on bionic olfaction and its device | |
CN104792826A (en) | System and method for detecting milk freshness based on electronic nose | |
US11886504B2 (en) | Systems and methods for generating smell data and a database thereof | |
CN103389323B (en) | Method for evaluating ages of precious medicinal materials quickly and losslessly | |
Xu et al. | Integrated sensor array optimization with statistical evaluation | |
Vest Nielsen et al. | Full second-order chromatographic/spectrometric data matrices for automated sample identification and component analysis by non-data-reducing image analysis | |
CN107132311A (en) | A kind of fast gas recognizer extracted based on frequency domain character | |
Zarra et al. | Instrumental odour monitoring system classification performance optimization by analysis of different pattern-recognition and feature extraction techniques | |
JP2016180599A (en) | Data analysis device | |
Peirano et al. | Supervised semi-automated data analysis software for gas chromatography/differential mobility spectrometry (GC/DMS) metabolomics applications | |
Wang et al. | The recognition of different odor using convolutional neural networks extracted from time and temperature features | |
CN202794093U (en) | Device for fast detecting quality of baked food based on bionic olfaction | |
Yang et al. | A matched-profile method for simple and robust vapor recognition in electronic nose (E-nose) system | |
Yin et al. | A Gaussian-based kernel Fisher discriminant analysis for electronic nose data and applications in spirit and vinegar classification | |
CN115078327A (en) | Rapid detection method for dangerous chemicals based on neural network architecture search | |
CN102636588A (en) | Method for discriminating white spirits by using quartz crystal oscillator electronic nose | |
CN207036597U (en) | A kind of red grape grape variety of making wine and place of production fast discriminating device | |
Ahmadou et al. | Optimization of an electronic nose for rapid quantitative recognition | |
CN111965217A (en) | Portable recovered oil detection device and detection analysis method thereof | |
Men et al. | Data fusion of electronic nose and electronic tongue for discrimination of Chinese liquors |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20100728 |