CN111475936A - Taste perception model-based taste recognition method - Google Patents
Taste perception model-based taste recognition method Download PDFInfo
- Publication number
- CN111475936A CN111475936A CN202010256895.0A CN202010256895A CN111475936A CN 111475936 A CN111475936 A CN 111475936A CN 202010256895 A CN202010256895 A CN 202010256895A CN 111475936 A CN111475936 A CN 111475936A
- Authority
- CN
- China
- Prior art keywords
- module
- taste
- node
- nodes
- connection coefficient
- 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
- 235000019640 taste Nutrition 0.000 title claims abstract description 79
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000014860 sensory perception of taste Effects 0.000 title claims abstract description 49
- 210000001932 glossopharyngeal nerve Anatomy 0.000 claims abstract description 23
- 210000001103 thalamus Anatomy 0.000 claims abstract description 23
- 235000019613 sensory perceptions of taste Nutrition 0.000 claims abstract description 22
- 230000035923 taste sensation Effects 0.000 claims abstract description 22
- 210000001186 vagus nerve Anatomy 0.000 claims abstract description 21
- 210000000256 facial nerve Anatomy 0.000 claims abstract description 18
- 210000001679 solitary nucleus Anatomy 0.000 claims description 15
- 230000000638 stimulation Effects 0.000 claims description 15
- 230000002093 peripheral effect Effects 0.000 claims description 10
- 230000003461 thalamocortical effect Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 8
- 238000013461 design Methods 0.000 claims description 5
- 208000021833 sensation perception Diseases 0.000 claims description 4
- 230000003068 static effect Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims 1
- 241000282414 Homo sapiens Species 0.000 abstract description 5
- 238000012706 support-vector machine Methods 0.000 abstract description 4
- 238000004451 qualitative analysis Methods 0.000 abstract 1
- 108091006146 Channels Proteins 0.000 description 67
- 235000013405 beer Nutrition 0.000 description 14
- 239000000243 solution Substances 0.000 description 11
- 238000004140 cleaning Methods 0.000 description 9
- 210000005036 nerve Anatomy 0.000 description 8
- 239000012088 reference solution Substances 0.000 description 7
- 235000019658 bitter taste Nutrition 0.000 description 6
- 230000005284 excitation Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 235000009508 confectionery Nutrition 0.000 description 5
- 230000002964 excitative effect Effects 0.000 description 5
- 230000002401 inhibitory effect Effects 0.000 description 5
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 4
- WCUXLLCKKVVCTQ-UHFFFAOYSA-M Potassium chloride Chemical compound [Cl-].[K+] WCUXLLCKKVVCTQ-UHFFFAOYSA-M 0.000 description 4
- 230000001339 gustatory effect Effects 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 230000001953 sensory effect Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 101001026214 Homo sapiens Potassium voltage-gated channel subfamily A member 5 Proteins 0.000 description 3
- 235000013361 beverage Nutrition 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 241000544061 Cuculus canorus Species 0.000 description 2
- VEXZGXHMUGYJMC-UHFFFAOYSA-N Hydrochloric acid Chemical compound Cl VEXZGXHMUGYJMC-UHFFFAOYSA-N 0.000 description 2
- 102100037445 Potassium voltage-gated channel subfamily A member 5 Human genes 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 239000012153 distilled water Substances 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 239000001103 potassium chloride Substances 0.000 description 2
- 235000011164 potassium chloride Nutrition 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000003756 stirring Methods 0.000 description 2
- 238000011830 transgenic mouse model Methods 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 235000014101 wine Nutrition 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 206010013911 Dysgeusia Diseases 0.000 description 1
- KWYUFKZDYYNOTN-UHFFFAOYSA-M Potassium hydroxide Chemical compound [OH-].[K+] KWYUFKZDYYNOTN-UHFFFAOYSA-M 0.000 description 1
- 235000013334 alcoholic beverage Nutrition 0.000 description 1
- 235000019606 astringent taste Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004166 bioassay Methods 0.000 description 1
- 235000005770 birds nest Nutrition 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- FEWJPZIEWOKRBE-LWMBPPNESA-N levotartaric acid Chemical compound OC(=O)[C@@H](O)[C@H](O)C(O)=O FEWJPZIEWOKRBE-LWMBPPNESA-N 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008587 neuronal excitability Effects 0.000 description 1
- 239000011664 nicotinic acid Substances 0.000 description 1
- 235000001968 nicotinic acid Nutrition 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012567 pattern recognition method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 235000019643 salty taste Nutrition 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000000542 thalamic effect Effects 0.000 description 1
- 238000007794 visualization technique Methods 0.000 description 1
- 235000005765 wild carrot Nutrition 0.000 description 1
Images
Abstract
The invention discloses a taste recognition method based on a taste perception model, which relates to the technical field of taste recognition methods and comprises the steps of establishing a taste perception model, inputting taste information of different samples acquired by a taste sensor into the taste perception model, extracting the characteristics of the taste perception model under the condition of inputting the taste information by an SD (secure digital) method, and inputting the characteristics into a CS-SVM (support vector machine) model to realize qualitative classification of the different samples, wherein the taste perception model is constructed according to a taste conduction path of a human body and comprises a facial nerve module, a glossopharyngeal nerve module, a vagus nerve module, an orphan nucleus module, a thalamus retroventral medial nucleus module and an island cortex module. The taste sensation identification method can be used for carrying out qualitative analysis on different taste sensation information more accurately.
Description
Technical Field
The invention relates to the technical field of taste recognition methods, in particular to a taste recognition method based on a taste perception model.
Background
The electronic tongue is also called a mechanical gustation system, can simulate a gustation recognition system of a human body, and is an instrument for analyzing, recognizing and detecting complex smell and most of volatile components.
Taste nerves are currently studied mainly by means of patch clamps, cell chips, neuro-recording, electroencephalograms, etc., and the results of electrophysiological and molecular biological assays indicating that Shaker Kv1.5 channel (KCNA5) is the main functional DRK channel expressed in the Anterior Rat tongue are described in "L iu L, Hansen D, Kim I, Gilbertson T. Expression and Characterization of delayed recovery K + Channels", the results of direct bitter Taste tracking of Taste nerves from the Taste nerves of the mouse sensory nerve 1R 36-8 (bitterness of the transgenic mouse WG 2, Sweet 2J. The bitterness of the sensory nerve 2, Sweet # 3. The Biol., Hkak [ KCNA5 ] are the main functional DRK channel expressed in the Anterior Rat tongue, and the results of "Matsumoto I, Ohmoto M, YasuokaaA, YoshiharaY, Abe K. Genetic training of Gustaving of the sensory nerve origin of the bitter Taste nerves of the mouse, Swategory 1, Sweet # 11, Sweet # 7, Sweet # 3. A, Sweet-3. A. As shown in the results of the bitter Taste nerves of the transgenic mouse.
The above research aiming at the taste sense conduction mechanism is mainly a research result obtained from a local loop of a taste sense system or a physiological experiment, but the research on the whole taste sense conduction path and the perception capability thereof is not carried out, and a model really close to a human taste sense recognition system is not available, so that the high bionic performance cannot be realized.
Disclosure of Invention
In order to solve the above problems, the present invention provides a taste sensation recognition method based on a taste sensation perception model.
In order to achieve the purpose, the technical scheme provided by the invention is as follows: a taste recognition method based on a taste perception model, the method mainly comprising the following steps:
s1: constructing a topological structure and dynamic characteristic description model of each module in the taste perception model, wherein the modules comprise a facial nerve module, a glossopharyngeal nerve module, a vagus nerve module, a solitary nucleus module, a thalamus retroperitoneal nucleus module and an island cortex module;
s2: constructing a feedback loop dynamic characteristic description model of an island cortex module, an solitary bundle nucleus module and a thalamus ventral posterior medial nucleus module;
s3: constructing an integral taste perception model through the modules in the steps S1 and S2;
s4: obtaining taste information of different samples through a taste sensor and inputting the taste information into the taste perception model in the step S3;
s5: features of the taste perception model under the condition of inputting taste information of different samples are extracted through an SD (standard visualization method) method, and the features are input into a CS-SVM (customer support vector machine) model, so that qualitative classification of the different samples is realized.
Further, the topological structure of the facial nerve module in step S1 is a single-input single-output structural mode, that is, the facial nerve module receives external stimulation and outputs the stimulation to the solitary bundle nucleus, and the facial nerve module node E1The dynamic characteristic description model of (2) is a differential equation described by equation (1):
Furthermore, the topology of the glossopharyngeal nerve module in step S1 is a single-input and multiple-output structure mode, that is, the glossopharyngeal nerve module receives external stimulation and outputs the stimulation to the solitary nucleus and the vagus nerve module, and the node E of the glossopharyngeal nerve module2The dynamic characteristic description model of (2) is a differential equation as described in equation (2):
Furthermore, the topology of the vagus nerve module in step S1 is a multi-input single-output structure mode, that is, the vagus nerve module receives external stimulation and the output information of the glossopharyngeal nerve module, and outputs the information to the solitary bundle nucleus, and the node E of the vagus nerve module is3The dynamic characteristic description model of (a) is a differential equation described by equation (3):
E3i(t) denotes the ith channel
The potential state of the node;
and
are all differential equation coefficients;
representing an external input of the ith channel;
a gaussian-distributed random number, which is a positive number mean, representing the peripheral noise of the taste system;
is the connection coefficient;
is distributed
The output of the node, i.e. the glossopharyngeal nerve module,
is in a static state
A function; i takes the values 1 to n.
Further, the topological structure of the solitary bundle nucleus module in step S1 is a multi-input multi-output structure mode, that is, the solitary bundle nucleus receives the output of the facial nerve module, the glossopharyngeal nerve module, the vagus nerve module and the feedback information of the islet cortex module, and outputs the taste information to the thalamus retroventral medial nucleus module and the islet cortex module, and the solitary bundle nucleus module is provided with four nodes, J being respectively1、J2、K1And K2The dynamics description model is the differential equation of equations (4), (5), (6) and (7), respectively:
a and b are differential equation coefficients, and n is the number of channels of the model; j. the design is a square1i(t）、J2i(t)、K1i(t)、K2i(t) are respectively the ith passage J1、J2、K1、K2The potential state of the node; q (J)1i(t))，Q(J2i(t))， Q(K1i(t))， Q(K2i(t))， Q(E1i(t))，Q(E2i(t))， Q(E3i(t)) is the ith channel J1， J2， K1， K2， E1， E2， E3An output of the node;
Q(J1j(t)) is the jth channel J1An output of the node; q (K)1j(t)) is the jth channel K1An output of the node; d1(t) the potential state of the first feedback link sent by the IC; wjjThe connection coefficients of all the nodes of the channel J1 are obtained; wj1j2，Wj2j1Is the same channel J1And J2Connection coefficient of nodes, all channels J1And J2The connection coefficient values of the nodes are consistent;
Wj1k1，Wk1j1is the same channel J1And K1Connection coefficient of nodes, all channels J1And K1The connection coefficient values of the nodes are consistent; wj1k2，Wk2j1Is the same channel J1And K2Connection coefficient of nodes, all channels J1And K2The connection coefficient values of the nodes are consistent; wj1e1Is the same channel J1And E1Connection coefficient of nodes, all channels J1And E1The connection coefficient values of the nodes are consistent;
Wj1e2is the same channel J1And E2Connection coefficient of nodes, all channels J1And E2The connection coefficient values of the nodes are consistent; wj1e3Is the same channel J1And E3The connection coefficient of the nodes is consistent with the connection coefficient of the nodes of all channels J1 and E3; wj2k1，Wk1j2Is the same channel J2And K1Connection coefficient of nodes, all channels J2And K1The connection coefficient values of the nodes are consistent; wkkFor all channels K1The connection coefficient between the nodes; wk1k2，Wk2k1Is the same channel K1And K2Connection coefficient of nodes, all channels K1And K2The connection coefficient values of the nodes are consistent; kd1Is J1Node and feedback link D1The connection coefficient between; the values of i and j are all 1 to n, and i is not equal to j.
Further, the topology of the thalamocortical medial nucleus module in step S1 is a multi-input single-output structure, that is, the thalamocortical nucleus module receives the outputs of the solitary nucleus module and the islet cortex module and outputs the taste information to the islet cortex module, and the thalamocortical nucleus module is provided with four nodes, J being J3、J4、K3、K4The dynamics description model is the differential equation of equations (8), (9), (10) and (11), respectively:
a and b are differential equation coefficients, and n is the number of channels of the model; j3(t), J4(t), K3(t), K4(t) are potential states of J3, J4, K3, K4 nodes; q (J3(t)), Q (J4(t)), Q (K3(t)), Q (K4(t)) are the outputs of J3, J4, K3, K4 nodes; q (J1J (t)) is the output of the J1 th channel, wherein J is 1 to n; d2(t) is the potential state of the second feedback link sent by the IC; wj3J1 is the connection coefficient of all channel J1 nodes and J3 nodes; wj3J4, Wj4J3 is the connection coefficient of J3 and J4 nodes; wj3K3, Wk3J3 is the connection coefficient of J3 and K3 nodes; wj3K4, Wk4J3 is the connection coefficient of J3 and K4 nodes; wj4K3, Wk3J4 is the connection coefficient of the J4 and K3 nodes; wk3K4 and Wk4K3 are connection coefficients of K3 and K4 nodes; kd2 is a connection coefficient between a J3 node and a feedback link D2;
is the center noise.
Further, the islet cortex module in step S1 is represented by a node M, and its topology is a multi-input multi-output structure mode, that is, the islet cortex module receives the outputs of the solitary bundle nucleus module and the thalamus ventral posterior medial nucleus module and sends feedback information to the solitary bundle nucleus module and the thalamus ventral posterior medial nucleus module, and the dynamic characteristic description model of the thalamus ventral posterior medial nucleus module is a differential equation of formula (12):
Further, in the step S2, a feedback loop dynamics description model is constructed by using the islet cortex module-solitary bundle nucleus module and thalamus retroventral medial nucleus module, wherein dynamics of two pieces of feedback information sent by the IC are described by differential equations (13) and (14), respectively:
The invention has the beneficial effects that:
the gustatory recognition method provided by the invention describes the processing process of human gustatory information from the aspect of dynamic characteristics, is closer to the real gustatory sense of human beings, and improves the bionics degree of a gustatory recognition system.
The taste recognition method of the invention can accurately recognize the taste information of wines or other liquid beverages and make qualitative classification, and can be widely applied to the production and identification of wines and other beverages.
In addition to the objects, features and advantages described above, other objects, features and advantages of the present invention are also provided. The present invention will be described in further detail below with reference to the drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention.
FIG. 1 is a topological structure of the facial nerve module of a taste recognition method based on a taste perception model of the present invention;
FIG. 2 is a topological structure of the glossopharyngeal neural module of a taste recognition method based on a taste perception model according to an embodiment of the present invention;
FIG. 3 is a topological structure of a vagus nerve module of a taste sensation recognition method based on a taste perception model according to an embodiment of the present invention;
FIG. 4 is a topological structure of a solitary beam kernel module of a taste recognition method based on a taste perception model according to an embodiment of the present invention;
FIG. 5 is a topological structure of the thalamic retroventral medial nucleus module of a taste recognition method based on a taste perception model according to an embodiment of the present invention;
FIG. 6 is a topological structure of an island cortex module of a taste recognition method based on a taste perception model according to an embodiment of the present invention;
FIG. 7 is a topological structure of the overall taste perception model of a taste recognition method based on the taste perception model according to an embodiment of the present invention;
FIG. 8 is a flow chart of a method of taste recognition based on a taste perception model according to an embodiment of the present invention;
FIG. 9 is a histogram of neuronal excitability features of a taste perception model under different brands of beer taste information input of a taste perception model-based taste recognition method of an embodiment of the present invention;
FIG. 10 is a parameter search process of CS-SVM in a taste recognition method based on a taste perception model according to the present invention;
FIG. 11 is the classification result of CS-SVM in a taste recognition method based on a taste perception model according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In the examples and the drawings, in order to make the description more simple and clear, the facial nerve module is represented by CN VII, the glossopharyngeal nerve module by CN IX, the vagus nerve module by CN X, the nucleus solithromus module by NST, the posterior medial thalamus nucleus module by vpmcp, and the islet cortex module by IC.
In the present invention, the process of taste sensation information recognition and qualitative classification of beer of different brands by the taste sensation recognition method of the present invention will be described in detail by taking beer of different brands as an example, but in practical application, the taste sensation information of other alcoholic beverages and liquid beverages may be classified qualitatively and accurately.
Examples
The embodiment of the invention provides a taste sensation identification method based on a taste sensation perception model, which mainly comprises the following steps:
s1: constructing a topological structure and dynamic characteristic description model of each module in the taste perception model, wherein the modules comprise a facial nerve module, a glossopharyngeal nerve module, a vagus nerve module, a solitary nucleus module, a thalamus retroperitoneal nucleus module and an island cortex module;
referring to fig. 1, in any embodiment of the present invention, the topological structure of the facial nerve module is a single-input single-output structural mode, that is, the facial nerve module receives external stimulation and outputs the stimulation to the solitary bundle nucleus, and the node E of the facial nerve module1The dynamic characteristic description model of (2) is a differential equation described by equation (1):
Referring to fig. 2, in any embodiment of the present invention, the topology of the glossopharyngeal nerve module is a single-input multiple-output structure mode, that is, the glossopharyngeal nerve module receives external stimulation and outputs the stimulation to the solitary nucleus and the vagus nerve module, and the node E of the glossopharyngeal nerve module2The dynamic characteristic description model of (2) is a differential equation as described in equation (2):
Referring to fig. 3, in any embodiment of the present invention, the topology of the vagus nerve module is a multi-input single-output structure mode, that is, the vagus nerve module receives external stimulation and the output information of the glossopharyngeal nerve module and outputs the information to the solitary bundle nucleus, and the node E of the vagus nerve module3The dynamic characteristic description model of (a) is a differential equation described by equation (3):
Referring to fig. 4, in any embodiment of the present invention, the topological structure of the solitary bundle nucleus module is a multi-input multi-output structure mode, that is, the solitary bundle nucleus receives the output of the facial nerve module, the glossopharyngeal nerve module, the vagus nerve module and the feedback information of the islet cortex module, and outputs the taste information to the thalamus retroventral medial nucleus module and the islet cortex module, and the solitary bundle nucleus module is provided with four nodes, J being respectively1、J2、K1And K2The dynamics description model is the differential equation of equations (4), (5), (6) and (7), respectively:
a and b are differential equation coefficients, and n is the number of channels of the model; j. the design is a square1i(t）、J2i(t)、K1i(t)、K2i(t) are respectively the ith passage J1、J2、K1、K2The potential state of the node; q (J)1i(t))，Q(J2i(t))， Q(K1i(t))， Q(K2i(t))， Q(E1i(t))，Q(E2i(t))， Q(E3i(t)) is the ith channel J1， J2， K1， K2， E1， E2， E3An output of the node;
Q(J1j(t)) is the jth channel J1An output of the node; q (K)1j(t)) is the jth channel K1An output of the node; d1(t) the potential state of the first feedback link sent by the IC; wjjThe connection coefficients of all the nodes of the channel J1 are obtained; wj1j2，Wj2j1Is the same channel J1And J2Connection coefficient of nodes, all channels J1And J2The connection coefficient values of the nodes are consistent;
Wj1k1，Wk1j1is the same channel J1And K1Connection coefficient of nodes, all channels J1And K1The connection coefficient values of the nodes are consistent; wj1k2，Wk2j1Is the same channel J1And K2Connection coefficient of nodes, all channels J1And K2The connection coefficient values of the nodes are consistent; wj1e1Is the same channel J1And E1Connection coefficient of nodes, all channels J1And E1The connection coefficient values of the nodes are consistent;
Wj1e2is the same channel J1And E2Connection coefficient of nodes, all channels J1And E2The connection coefficient values of the nodes are consistent; wj1e3Is the same channel J1And E3The connection coefficient of the nodes is consistent with the connection coefficient of the nodes of all channels J1 and E3; wj2k1，Wk1j2Is the same channel J2And K1Connection coefficient of nodes, all channels J2And K1The connection coefficient values of the nodes are consistent; wkkFor all channels K1The connection coefficient between the nodes; wk1k2，Wk2k1Is the same channel K1And K2Connection coefficient of nodes, all channels K1And K2The connection coefficient values of the nodes are consistent; kd1Is J1Node and feedback link D1The connection coefficient between; the values of i and j are all 1 to n, and i is not equal to j.
Referring to fig. 5, in any embodiment of the present invention, the topology of the thalamocortical nucleus module is a multi-input single-output structure mode, that is, the thalamocortical nucleus module receives the outputs of the solitary nucleus module and the islet cortex module and outputs the taste information to the islet cortex module, and the thalamocortical nucleus module is provided with four nodes, J being J3、J4、K3、K4The dynamics description model is the differential equation of equations (8), (9), (10) and (11), respectively:
a and b are differential equation coefficients, and n is the number of channels of the model; j3(t), J4(t), K3(t), K4(t) are potential states of J3, J4, K3, K4 nodes; q (J3(t)), Q (J4(t)), Q (K3(t)), Q (K4(t)) are the outputs of J3, J4, K3, K4 nodes; q (J1J (t)) is the output of the J1 th channel, wherein J is 1 to n; d2(t) is the potential state of the second feedback link sent by the IC; wj3J1 is the connection coefficient of all channel J1 nodes and J3 nodes; wj3J4, Wj4J3 is the connection coefficient of J3 and J4 nodes; wj3K3, Wk3J3 is the connection coefficient of J3 and K3 nodes; the number of the Wj3k4,wk4J3 is the connection coefficient of the J3 and K4 nodes; wj4K3, Wk3J4 is the connection coefficient of the J4 and K3 nodes; wk3K4 and Wk4K3 are connection coefficients of K3 and K4 nodes; kd2 is J3 node and feedback link D2The connection coefficient between;
is the center noise.
Referring to fig. 6, in any embodiment of the present invention, the islet cortex module is represented by a node M, and the topology thereof is a multi-input multi-output structural mode, that is, the islet cortex module receives the outputs of the solitary tract nucleus module and the thalamus ventral posterior medial nucleus module and sends feedback information to the solitary tract nucleus module and the thalamus ventral posterior medial nucleus module, and the dynamic characteristic description model of the thalamus ventral posterior medial nucleus module is a differential equation of formula (12):
S2: constructing a feedback loop dynamic characteristic description model of an island cortex module, an solitary bundle nucleus module and a thalamus ventral posterior medial nucleus module;
in any embodiment of the present invention, in step S2, a feedback loop dynamics description model is constructed for the islet cortex module-solitary bundle nucleus module and the thalamus retroventral medial nucleus module, wherein dynamics of two pieces of feedback information sent by the IC are described by differential equations (13) and (14), respectively:
S3: constructing an integral taste perception model through the modules in the steps S1 and S2;
referring to fig. 7, in any embodiment of the present invention, the overall taste perception model is established by: distributed E of taste model1、E2、E3The node receives external stimulation and peripheral noise and outputs the processed stimulation information to the distributed J1Node, E1、E2、E3There are no excitatory/inhibitory connections between nodes; distributed J1The nodes are laterally connected with all the same nodes except the nodes, and the received stimulation information and the feedback information from the IC are processed; in n channels J1The output mean of the node represents the output of NST, and the taste information is transmitted to J3Node, M nodePoint wherein J1The node transmits its excitatory output to J2、K1、K2Node, J2The node transmits its excitatory output to J1、K1Node, K1Passing its inhibitory output to J1、J2、K2Node, K2The node passes its inhibitory output to J1、K1A node; j. the design is a square3The node representation VPMpc receives the output of the center noise, NST and feedback information from the IC, K3The output of the node represents the output of the VPMpc, and taste information is transmitted to the M node, i.e. IC, J3The node transmits its excitatory output to J4、K3、K4Node, J4The node transmits its excitatory output to J3、K3Node, K3Passing its inhibitory output to J3、J4、K4Node, K4The node passes its inhibitory output to J3、K3A node; m node represents IC, accepts J from n channels1Output mean value and K of node3Output of node and through feedback link D1、D2And distributed J1Node, J3The nodes are connected; the taste perception model describes the forward path from CN VII, CN IX, CN X to NST, vpmcp to IC and the feedback loop from IC to NST, vpmcp.
S4: obtaining taste information of different samples through a taste sensor and inputting the taste information into the taste perception model in the step S3;
referring to fig. 8, illustratively, the taste sensor may be an SA-402B electronic tongue whose sensor array consists of 2 reference electrodes and 5 basic taste sensors; the basic taste sensor can realize the sensory information detection of 5 basic tastes including sour, fresh, salty, bitter and astringent tastes of a sample to be detected.
The different samples can be 5 kinds of beer with different brands, and the specific parameters are shown in the table 1:
TABLE 1 parameter information for different brands of beer
The specific process for acquiring taste information of different brands of beer is as follows:
(I) Experimental Equipment and Material
(1) Placing a beer sample, a reference solution and a positive and negative electrode cleaning solution;
the reference solution is a solution containing 0.3 mmol/L tartaric acid and 30 mmol/L potassium chloride;
the preparation process of the anode cleaning solution comprises the steps of adding 300M L95% ethanol into about 500ml of distilled water, fully stirring, adding 100ml of 1M hydrochloric acid solution, transferring the solution into a 1000M L volumetric flask for constant volume to obtain the anode cleaning solution;
the preparation process of the cathode cleaning solution comprises the steps of adding 7.46g of potassium chloride and 500M of L95% ethanol into about 500ml of distilled water, stirring uniformly, adding 10M of L1M of potassium hydroxide solution, transferring to a volumetric flask with 1000M of L, and carrying out constant volume to obtain the cathode cleaning solution.
(2) Before the test starts, the positive electrode sensor array is placed into positive electrode cleaning solution, the negative electrode sensor array is placed into negative electrode cleaning solution and cleaned for 90s, after the test is finished, the positive electrode sensor array and the negative electrode sensor array are respectively placed into two containers containing reference solution and cleaned for 120s, the reference solution is replaced and cleaned continuously for 120s, and then the reference solution is replaced to enable the balance of the sensors to return to zero for 30s so as to ensure that output signals are stable;
(3) after the response output of the sensor reaches balance, starting to acquire taste information of the beer, wherein the detection time of each type of beer is 30s, after the measurement is finished, quickly cleaning the beer in a reference solution for 2 times, detecting the aftertaste value of the basic taste information in the replaced reference solution, finishing the measurement once, repeating the step (2), and cleaning and calibrating the sensor;
(4) 3 parallel samples are prepared for the beer of different brands, and each group of samples are repeatedly detected for 6 times by setting system parameters, namely 18 groups of taste information data are intelligently obtained for the beer of each brand, and 90 groups of taste data are obtained after the experiment is finished.
The experimental conditions of the electronic tongue are that the room temperature is 20 +/-0.5 ℃, the relative humidity is 65 +/-2% RH, and the voltage value of the 30 th s of the response curve of the sensor is taken as a characteristic value for data analysis.
(II) setting parameter values of differential equations of different modules of taste perception model
Setting parameter values according to a computer numerical simulation result of the taste perception model:
TABLE 2 values of the parameters of the differential equations of the different modules of the taste perception model
And S5, extracting the characteristics of the taste perception model under the condition of inputting the taste information of different samples by an SD method, and inputting the characteristics into a CS-SVM model to realize qualitative classification of the different samples. The taste sensor can be an electronic tongue, and can also be any sensor capable of acquiring taste information.
Referring to FIGS. 8-11, exemplary taste perception models of 5 different brands of beer are inputted with taste data, and 8 channels NST layer J of the taste perception models are obtained by SD method1The neuron excitation of the node output sequence is calculated by the SD method as follows:
inputting each set of taste data into taste model, and inputting taste perception model J1At the node of
Of a channel
mean division of ms segments into
Respectively calculating the standard deviation of each segment and recording as the excitability
And averaging the channel excitation
，
The calculation method is shown in formula (15):
all channels J1The excitation of the nodes constitutes a vector
And using the data as a group of feature vectors to finally obtain 90 groups of feature data, and fig. 7 shows a histogram of neuron excitation features.
The method randomly divides 90 groups of characteristic data into 60 groups of training set data and 30 groups of test set data, selects a Radial Basis Function (RBF) as a kernel function, and optimizes a penalty factor in a support vector machine model by using a cuckoo algorithm
And kernel parameters
. In this connection, it is possible to use,
and
in the range of [0.01, 100%]The iteration number N is 200, the initial bird nest number is 20, and the probability Pa that a new egg is found by a host bird is set to be 0.25. In the parameter searching process, the cross validation classification error rate is adopted as a fitness function, and the parameter with the minimum classification error rate
And
return toAnd a model is established.
The neuron excitation characteristic data set of the taste model is processed by the method, the searching process of the optimal fitness function is shown in figure 10, the searching curve diagram of the fitness function shows that the optimal fitness of the neuron excitation characteristic value set searched by the cuckoo can reach 0.032, the classification result of the CS-SVM is shown in table 3,
TABLE 3 parameter optimization results of SVM under three characteristics
From Table 3, in the optimum parameters
And
and the accuracy rate of the CS-SVM training set is 100%, the accuracy rate of the test set is 96.67%, and the numerical result shows that the pattern recognition method provided by the invention can realize qualitative classification of beer samples of different brands.
The invention relates to a working principle of a taste sensation identification method based on a taste sensation perception model, which comprises the following steps: by using the taste sensor, which can be an electronic tongue, the taste information of different samples can be obtained as external stimulation
Inputting the external noise into the taste perception model constructed by the invention; and extracting the characteristics of the taste perception model under the input stimulation, inputting the characteristics into the CS-SVM model, and verifying the effectiveness of the mode identification method according to the qualitative classification result of the CS-SVM on the taste information of different samples.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (8)
1. A taste sensation identification method based on a taste sensation perception model is characterized by mainly comprising the following steps:
s1: establishing a topological structure and dynamic characteristic description model of each module in a taste perception model, wherein the modules comprise a facial nerve module, a glossopharyngeal nerve module, a vagus nerve module, a solitary nucleus module, a thalamus retroperitoneal nucleus module and an island cortex module;
s2: constructing a feedback loop dynamic characteristic description model of an island cortex module, an solitary bundle nucleus module and a thalamus ventral posterior medial nucleus module;
s3: constructing an integral taste perception model through the modules in the steps S1 and S2;
s4: obtaining taste information of different samples through a taste sensor and inputting the taste information into the taste perception model in the step S3;
s5: the characteristics of the taste perception model under the condition of inputting taste information of different samples are extracted through an SD method, and the characteristics are input into a CS-SVM model, so that qualitative classification of the different samples is realized.
2. The taste perception model-based taste sensation identification method of claim 1, wherein the topological structure of the facial nerve module in step S1 is a single-input single-output structure mode, i.e., the facial nerve module receives external stimulus and outputs to the solitary bundle nucleus, and the facial nerve module node E1The dynamic characteristic description model of (2) is a differential equation of formula (1):
3. The taste sensation recognition method according to claim 1, wherein the topology of the glossopharyngeal nerve module in step S1 is a single-input multiple-output structure, that is, the glossopharyngeal nerve module receives external stimulus and outputs the stimulus to the solitary nucleus and the vagus nerve module, and the node E of the glossopharyngeal nerve module2The dynamic characteristic description model of (2) is a differential equation of formula (2):
4. The taste sensation recognition method based on taste sensation model of claim 1, wherein the topology of the vagus nerve module in step S1 is a multi-input single-output structure mode, that is, the vagus nerve module receives external stimulation and the output information of the glossopharyngeal nerve module and outputs the information to the solitary nucleus, and the node E of the vagus nerve module3The dynamic characteristic description model of (2) is a differential equation of formula (3):
E3i(t) denotes the ith channel E3The potential state of the node;
and
are all differential equation coefficients;
representing an external input of the ith channel;
a gaussian-distributed random number, which is a positive number mean, representing the peripheral noise of the taste system;
is the connection coefficient;
is distributed
The output of the node, i.e. the glossopharyngeal nerve module,
is a static Sigmoid function; i takes the values 1 to n.
5. The taste sensation recognition method according to claim 1, wherein the topological structure of the solitary nucleus module in step S1 is a multi-input multi-output structure mode in which the solitary nucleus receives the output of the facial nerve module, the glossopharyngeal nerve module, the vagus nerve module and the feedback information of the islet cortex module, and outputs the taste sensation information to the thalamus retroventral nucleus module and the islet cortex module, and the solitary nucleus module is provided with four nodes, J being J1、J2、K1And K2The dynamics description model is the differential equation of equations (4), (5), (6) and (7), respectively:
（4）
（5）
（6）
a and b are differential equation coefficients, and n is the number of channels of the model; j. the design is a square1i(t）、J2i(t)、K1i(t)、K2i(t) are respectively the ith passage J1、J2、K1、K2The potential state of the node; q (J)1i(t))，Q(J2i(t))， Q(K1i(t))， Q(K2i(t))， Q(E1i(t))，Q(E2i(t))， Q(E3i(t)) is the ith channel J1， J2， K1， K2， E1， E2， E3An output of the node;
Q(J1j(t)) is the jth channel J1An output of the node; q (K)1j(t)) is the jth channel K1An output of the node; d1(t) the potential state of the first feedback link sent by the IC; wjjThe connection coefficients of all the nodes of the channel J1 are obtained; wj1j2，Wj2j1Is the same channel J1And J2Connection coefficient of nodes, all channels J1And J2The connection coefficient values of the nodes are consistent;
Wj1k1，Wk1j1is the same channel J1And K1Connection coefficient of nodes, all channels J1And K1The connection coefficient values of the nodes are consistent; wj1k2，Wk2j1Is the same channel J1And K2Connection coefficient of nodes, all channels J1And K2The connection coefficient values of the nodes are consistent; wj1e1Is the same channel J1And E1Connection coefficient of nodes, all channels J1And E1The connection coefficient values of the nodes are consistent;
Wj1e2is the same channel J1And E2Connection coefficient of nodes, all channels J1And E2The connection coefficient values of the nodes are consistent; wj1e3Is the same channel J1And E3The connection coefficient of the nodes is consistent with the connection coefficient of the nodes of all channels J1 and E3; wj2k1，Wk1j2Is the same channel J2And K1Connection coefficient of nodes, all channels J2And K1The connection coefficient values of the nodes are consistent; wkkFor all channels K1The connection coefficient between the nodes; wk1k2，Wk2k1Is the same channel K1And K2Connection coefficient of nodes, all channels K1And K2The connection coefficient values of the nodes are consistent; kd1Is J1Node and feedback link D1The connection coefficient between; the values of i and j are all 1 to n, and i is not equal to j.
6. The taste sensation recognition method according to claim 1, wherein the topology of the thalamocortical posterior medial nucleus module in step S1 is a multi-input single-output structure mode, that is, the thalamocortical posterior medial nucleus module receives the outputs of the solitary tract nucleus module and the insular cortex module and outputs the taste sensation information to the insular cortex module, and the thalamocortical posterior medial nucleus module is provided with four nodes, J being J3、J4、K3、K4The dynamics description model is the differential equation of equations (8), (9), (10) and (11), respectively:
a and b are differential equation coefficients, and n is the number of channels of the model; j. the design is a square3(t)， J4(t)， K3(t)， K4(t) is J3， J4， K3，K4The potential state of the node; q (J)3(t))， Q(J4(t))， Q(K3(t))， Q(K4(t)) is J3， J4， K3， K4An output of the node; q (J)1j(t)) is the jth channel J1The output of the node, wherein j takes the value of 1 to n; d2(t) is the potential state of the second feedback link sent by the IC; wj3j1For all channels J1Node and J3The connection coefficient of the node; wj3j4，Wj4j3Is J3And J4The connection coefficient of the node; wj3k3，Wk3j3Is J3And K3The connection coefficient of the node; wj3k4，Wk4j3Is J3And K4The connection coefficient of the node; wj4k3，Wk3j4Is J4And K3The connection coefficient of the node; wk3k4，Wk4k3Is K3And K4The connection coefficient of the node; kd2Is J3Node and feedback link D2The connection coefficient between;
is the center noise.
7. The taste recognition method based on taste perception model of claim 1, wherein the islet cortex module in step S1 is represented by node M, and its topology is a multi-input multi-output structure mode, i.e. the islet cortex module receives the outputs of the solitary bundle nucleus module and the thalamus ventral-posterior-medial nucleus module and sends feedback information to the solitary bundle nucleus module and the thalamus ventral-posterior-medial nucleus module, and the dynamical characteristics of the thalamus ventral-medial nucleus module describes a differential equation of formula (12):
8. The taste recognition method based on taste perception model according to claim 1, wherein said step S2 is a construction of feedback loop dynamics description model of islet cortex module-solitary bundle nucleus module and thalamus retroventral medial nucleus module, wherein dynamics of two feedback messages from IC are described by differential equations (13) and (14), respectively:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010256895.0A CN111475936A (en) | 2020-04-03 | 2020-04-03 | Taste perception model-based taste recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010256895.0A CN111475936A (en) | 2020-04-03 | 2020-04-03 | Taste perception model-based taste recognition method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111475936A true CN111475936A (en) | 2020-07-31 |
Family
ID=71750457
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010256895.0A Pending CN111475936A (en) | 2020-04-03 | 2020-04-03 | Taste perception model-based taste recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111475936A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111950721A (en) * | 2020-10-09 | 2020-11-17 | 东北电力大学 | Flavor identification method based on smell-taste joint perception model |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002150258A (en) * | 2000-11-06 | 2002-05-24 | Sony Corp | Circuit using neuron model and information processing method |
| CN101416055A (en) * | 2005-06-22 | 2009-04-22 | 塞诺米克斯公司 | Identification of human t2r receptors that are activated by bitter molecules in coffee (chlorogenic lactones) and related assays for identifying human bitter taste modulators |
| US20110213439A1 (en) * | 2010-02-26 | 2011-09-01 | The Rockefeller University | Neuromodulation Having Non-Linear Dynamics |
| CN104698044A (en) * | 2013-12-10 | 2015-06-10 | 上海闵临机电科技有限公司 | Electronic tongue for food detection |
| CN107657214A (en) * | 2017-09-04 | 2018-02-02 | 重庆大学 | A kind of local discriminant keeps the electronic tongues taste identification method of projection |
| CN108542385A (en) * | 2018-04-02 | 2018-09-18 | 东北电力大学 | A method of carrying out sense organ flavor substance classification using smell brain wave |
| KR20180103297A (en) * | 2017-03-09 | 2018-09-19 | 연세대학교 산학협력단 | Device and system for test of gustatory function |
| CN108981800A (en) * | 2018-06-25 | 2018-12-11 | 东北电力大学 | It is a kind of to smell-gustatism effect visualization method using neurodynamics system model progress machine |
-
2020
- 2020-04-03 CN CN202010256895.0A patent/CN111475936A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002150258A (en) * | 2000-11-06 | 2002-05-24 | Sony Corp | Circuit using neuron model and information processing method |
| CN101416055A (en) * | 2005-06-22 | 2009-04-22 | 塞诺米克斯公司 | Identification of human t2r receptors that are activated by bitter molecules in coffee (chlorogenic lactones) and related assays for identifying human bitter taste modulators |
| US20110213439A1 (en) * | 2010-02-26 | 2011-09-01 | The Rockefeller University | Neuromodulation Having Non-Linear Dynamics |
| CN104698044A (en) * | 2013-12-10 | 2015-06-10 | 上海闵临机电科技有限公司 | Electronic tongue for food detection |
| KR20180103297A (en) * | 2017-03-09 | 2018-09-19 | 연세대학교 산학협력단 | Device and system for test of gustatory function |
| CN107657214A (en) * | 2017-09-04 | 2018-02-02 | 重庆大学 | A kind of local discriminant keeps the electronic tongues taste identification method of projection |
| CN108542385A (en) * | 2018-04-02 | 2018-09-18 | 东北电力大学 | A method of carrying out sense organ flavor substance classification using smell brain wave |
| CN108981800A (en) * | 2018-06-25 | 2018-12-11 | 东北电力大学 | It is a kind of to smell-gustatism effect visualization method using neurodynamics system model progress machine |
Non-Patent Citations (4)
| Title |
|---|
| FU, J (FU, JUN) ; LI, G (LI, GUANG) ; QIN, Y (QIN, YUQI) ; FREEMAN, WJ (FREEMAN, WALTER J.): "A pattern recognition method for electronic noses based on an olfactory neural network", SENSORS AND ACTUATORS B-CHEMICAL, vol. 125, no. 2, 8 August 2007 (2007-08-08), pages 489 - 497, XP022296106, DOI: 10.1016/j.snb.2007.02.058 * |
| OHLA, K (OHLA, KATHRIN)等: "Recognizing Taste: Coding Patterns Along the Neural Axis in Mammals", CHEMICAL SENSES, vol. 44, no. 5, 20 February 2019 (2019-02-20), pages 237 - 247 * |
| ZHANG, J (ZHANG, JIN) ; TIAN, TT (TIAN, TIANTIAN); WANG, SC (WANG, SHENGCHUN)等: "Research on an olfactory neural system model and its applications based on deep learning", NEURAL COMPUTING & APPLICATIONS, vol. 32, no. 10, 25 September 2019 (2019-09-25), pages 5713 - 5724, XP037110834, DOI: 10.1007/s00521-019-04498-x * |
| 杨如乃，胡志忠，卢健: "生物嗅觉神经系统模型的模拟与分析", 生物医学工程研究, vol. 25, no. 3, 30 September 2006 (2006-09-30), pages 131 - 136 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111950721A (en) * | 2020-10-09 | 2020-11-17 | 东北电力大学 | Flavor identification method based on smell-taste joint perception model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dębska et al. | Application of artificial neural network in food classification | |
| Tang et al. | A pruning neural network model in credit classification analysis | |
| CN109190665A (en) | A kind of general image classification method and device based on semi-supervised generation confrontation network | |
| Fu et al. | A pattern recognition method for electronic noses based on an olfactory neural network | |
| Wang et al. | An optimized deep convolutional neural network for dendrobium classification based on electronic nose | |
| Ju et al. | Effects of synaptic connectivity on liquid state machine performance | |
| Qin et al. | Using a one-dimensional convolutional neural network with a conditional generative adversarial network to classify plant electrical signals | |
| CN111863244B (en) | Functional connection mental disease classification method and system based on sparse pooling graph convolution | |
| Nayebi et al. | Identifying learning rules from neural network observables | |
| CN111340132B (en) | Machine olfaction mode identification method based on DA-SVM | |
| CN113008742A (en) | Method and system for detecting deposition amount of fog drops | |
| CN112417166A (en) | Knowledge graph triple confidence evaluation method | |
| Yang et al. | Synergetic application of E-tongue and E-eye based on deep learning to discrimination of Pu-erh tea storage time | |
| CN108981800B (en) | Method for visualizing machine smell-taste perception effect by using neuro-dynamics system model | |
| Fu et al. | Discrimination of geographical indication of Chinese green teas using an electronic nose combined with quantum neural networks: A portable strategy | |
| Wang et al. | Classification and concentration prediction of VOCs with high accuracy based on an electronic nose using an ELM-ELM integrated algorithm | |
| CN111475936A (en) | Taste perception model-based taste recognition method | |
| Raman et al. | Processing of chemical sensor arrays with a biologically inspired model of olfactory coding | |
| Liu et al. | Using spike-based bio-inspired olfactory model for data processing in electronic noses | |
| Zheng et al. | Computational model of taste pathways: a biomimetic algorithm for electronic tongue based on nerve conduction mechanism | |
| Bhatheja et al. | Detection of cotton plant disease for fast monitoring using enhanced deep learning technique | |
| Singh | Better application of Bayesian deep learning to diagnose disease | |
| De Stefano et al. | Evolutionary computation to implement an IoT-based system for water pollution detection | |
| Haveri et al. | Review on plant disease detection using deep learning | |
| CN114626408A (en) | Electroencephalogram signal classification method and device, electronic equipment, medium and product |
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 |