CN108932483B - Smell identification method and device based on machine olfaction - Google Patents
Smell identification method and device based on machine olfaction Download PDFInfo
- Publication number
- CN108932483B CN108932483B CN201810628801.0A CN201810628801A CN108932483B CN 108932483 B CN108932483 B CN 108932483B CN 201810628801 A CN201810628801 A CN 201810628801A CN 108932483 B CN108932483 B CN 108932483B
- Authority
- CN
- China
- Prior art keywords
- response signal
- mixed
- target
- unknown
- smell
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N7/00—Computing arrangements based on specific mathematical models
- G06N7/08—Computing arrangements based on specific mathematical models using chaos models or non-linear system models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mathematical Physics (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Genetics & Genomics (AREA)
- Molecular Biology (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Nonlinear Science (AREA)
- Algebra (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
- Investigating Or Analyzing Materials By The Use Of Electric Means (AREA)
Abstract
The invention discloses a smell identification method based on machine olfaction, which comprises the following steps: acquiring a mixed response signal of the target mixed smell; judging whether an unknown response signal exists in the mixed response signal; if so, filtering unknown response signals from the mixed response signals to obtain target response signals, comparing the target response signals with template response signals in a preset database, and determining the odor type and content corresponding to the target response signals. Before processing the mixed response signal of the mixed odor, the method firstly filters the unknown response signal in the mixed response signal to avoid the influence of the unknown odor on the identification result, thereby improving the identification efficiency, accuracy and accuracy of quantitative analysis of the odor. Correspondingly, the invention also discloses a smell recognition device based on machine olfaction, equipment and a computer readable storage medium, which have the technical effects.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a smell identification method, a device, equipment and a computer readable storage medium based on machine olfaction.
Background
At present, the application of machine olfaction technology is more and more extensive, and a large number of application cases exist in the aspects of industrial gas monitoring, chemical product quality monitoring, indoor and outdoor environment monitoring, medical diagnosis, traditional Chinese medicine material identification and the like.
In these application fields, most of the monitored odors are mixed odors, and the odor information in the electronic nose database is limited, so that unknown gases are inevitably present. In addition, the existing odor identification technology does not consider the influence of unknown odor on the identification result when processing the response signal of the sensor, thereby causing larger error of the identification result and reducing the identification efficiency, accuracy and accuracy of quantitative odor analysis.
Therefore, how to improve the efficiency and accuracy of odor identification and the accuracy of odor quantitative analysis is a problem to be solved by those skilled in the art.
Disclosure of Invention
The invention aims to provide a smell identification method, a device, equipment and a computer readable storage medium based on machine olfaction, so as to improve the efficiency and accuracy of smell identification and the precision of quantitative analysis of smell.
In order to achieve the above purpose, the embodiment of the present invention provides the following technical solutions:
a smell recognition method based on machine olfaction, comprising:
acquiring a mixed response signal of the target mixed smell;
judging whether an unknown response signal exists in the mixed response signal;
if so, filtering the unknown response signal from the mixed response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal.
Wherein, the step of filtering the unknown response signal from the mixed response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal comprises:
judging whether the mixed response signal is a linear response signal or not;
if yes, filtering the unknown response signal by using an optimization function in a linear mixed model to obtain the target response signal, and enabling the target to be the targetComparing the response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a multiplier method;
h denotes the filter mixing matrix, z (t) denotes the unknown response signal, t denotes the time variable, and x (t-P) denotes the noise signal.
Wherein, when there is no unknown response signal in the mixed response signal and the mixed response signal is a linear response signal, the method further comprises:
filtering the unknown response signal by using an optimization function in the linear mixed model to obtain the target response signal;
and comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a ridge regression method.
Wherein, when the mixed response signal is a nonlinear response signal, the method further comprises:
filtering the unknown response signal by using an optimization function in a nonlinear hybrid model to obtain the target response signal;
comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal through a chaotic genetic algorithm;
wherein the optimization function in the nonlinear hybrid model is: min | | y (t) -g (t) -z (t) |, y (t) represents a mixed response signal, g (t) represents a nonlinear response signal, and z (t) represents an unknown response signal.
The nonlinear mixed model is generated by using a multivariate self-adaptive regression spline method, and the specific generation steps are as follows:
acquiring experimental data of the mixed smell, and correcting the environmental factors of the experimental data;
dividing the corrected experimental data to obtain a plurality of intervals, and generating a corresponding basis function based on each interval;
and selecting the basis function with the highest accuracy by adopting an averaging error method, and determining the model corresponding to the basis function with the highest accuracy as the nonlinear mixed model.
Wherein the determining of the odor type and content corresponding to the target response signal through the chaotic genetic algorithm comprises:
generating a population of the chaotic genetic algorithm based on the target response signal;
carrying out individual evaluation on individuals in the population, and carrying out selection operation, cross operation and variation operation on the population subjected to the individual evaluation so as to iteratively update the population;
judging whether the current evolution algebra is larger than a preset maximum evolution algebra or not;
if yes, outputting an optimal solution, and determining the odor type and content corresponding to the target response signal according to the optimal solution.
A machine olfaction-based scent recognition device, comprising:
the acquisition module is used for acquiring a mixed response signal of the target mixed smell;
the judging module is used for judging whether an unknown response signal exists in the mixed response signal;
and the identification module is used for filtering the unknown response signal from the mixed response signal to obtain a target response signal when the unknown response signal exists in the mixed response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal.
Wherein the identification module comprises:
the judging unit is used for judging whether the mixed response signal is a linear response signal or not;
the recognition unit is used for filtering the unknown response signal by using an optimization function in a linear mixing model when the mixed response signal is a linear response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a multiplier method;
A machine olfaction-based scent recognition device comprising:
a memory for storing a computer program;
a processor for implementing the steps of any one of the above-mentioned smell recognition methods based on machine olfaction when executing the computer program.
A computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of any of the above-described machine scent-based scent recognition methods.
According to the scheme, the smell identification method based on machine olfaction provided by the embodiment of the invention comprises the following steps: acquiring a mixed response signal of the target mixed smell; judging whether an unknown response signal exists in the mixed response signal; if so, filtering the unknown response signal from the mixed response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal.
Therefore, before processing the mixed response signal of the mixed odor, the method firstly filters the unknown response signal in the mixed response signal to avoid the influence of the unknown odor on the identification result; meanwhile, the target response signal obtained after the unknown response signal is filtered is compared with the template response signal in the preset database, so that the odor type and content corresponding to the target response signal are determined, the recognition efficiency and accuracy are improved, and the accuracy of quantitative odor analysis is also improved.
Accordingly, the odor recognition device, the odor recognition equipment and the computer-readable storage medium based on machine olfaction provided by the embodiment of the invention also have the technical effects.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a smell recognition method based on machine olfaction disclosed in the embodiment of the invention;
fig. 2 is a schematic diagram of response modes of three similar smells disclosed in the embodiment of the present invention:
FIG. 3 is a flow chart of another smell recognition method based on machine olfaction disclosed in the embodiment of the invention;
FIG. 4 is a schematic diagram of a smell recognition device based on machine olfaction according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a smell recognition device based on machine olfaction according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention discloses a smell identification method, a device, equipment and a computer readable storage medium based on machine olfaction, which are used for improving the efficiency and the accuracy of smell identification and the precision of quantitative analysis of smell.
Referring to fig. 1, an embodiment of the present invention provides a smell identification method based on machine olfaction, including:
s101, acquiring a mixed response signal of a target mixed smell;
s102, judging whether an unknown response signal exists in the mixed response signal; if yes, executing S103; if not, executing S104;
s103, filtering unknown response signals from the mixed response signals to obtain target response signals, comparing the target response signals with template response signals in a preset database, and determining the odor type and content corresponding to the target response signals;
specifically, since the unknown response signal exists in the mixed response signal, in order to eliminate the influence of the unknown response signal on the odor analysis process, the unknown response signal needs to be filtered first. Because different response signals with similar smells have approximately the same response mode collected by the same sensor, the unknown response signals can be expressed by adopting a specific function.
For example: fig. 2 shows that the same sensor has the same odor response pattern, so that when there is an unknown odor (a single odor generated by a single gas component or a composite odor generated by multiple gas components), the values of different unknown odors generated by different sensors are different, but the odor response patterns of each sensor are necessarily the same. Based on this, a specific objective function can be optimized to generate a recognition model that takes into account unknown odors.
And S104, comparing the mixed response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the mixed response signal.
Specifically, when the unknown response signal does not exist in the mixed response signal, the mixed response signal may be directly compared with the template response signal in the preset database, so as to determine the odor type and content corresponding to the mixed response signal.
Therefore, the embodiment provides a smell identification method based on machine olfaction, which firstly filters an unknown response signal in a mixed response signal before processing the mixed response signal of the mixed smell to avoid the influence of the unknown smell on an identification result; meanwhile, the target response signal obtained after the unknown response signal is filtered is compared with the template response signal in the preset database, so that the odor type and content corresponding to the target response signal are determined, the recognition efficiency and accuracy are improved, and the accuracy of quantitative odor analysis is also improved.
The embodiment of the invention discloses a smell identification method based on machine olfaction, and compared with the previous embodiment, the embodiment further explains and optimizes the technical scheme.
Referring to fig. 3, another smell identification method based on machine olfaction provided by the embodiment of the present invention includes:
s301, acquiring a mixed response signal of the target mixed smell;
s303, judging whether an unknown response signal exists in the mixed response signal; if yes, executing S303; if not, executing S307;
s303, judging whether the mixed response signal is a linear response signal; if yes, go to S304; if not, executing S305;
s304, filtering unknown response signals by using an optimization function in the linear mixed model to obtain target response signals, comparing the target response signals with template response signals in a preset database, and determining the odor type and content corresponding to the target response signals by a multiplier method;
Moreover, the multiplier method is used for constrained nonlinear programming, belongs to a penalty function method, adds corresponding penalty to the Lagrange function of the constraint problem, can ensure that an iteration point can converge to the minimum point of the constraint problem, and more importantly can ensure the stability of numerical calculation.
Need to make sure thatIt is noted that the optimization function in the linear mixture model can be obtained based on the following procedure. When the steady-state response and the transient response of the machine olfactory sensor array are considered simultaneously under linear mixing conditions, the response of a single sensor to a single scent can be described by a finite impulse response filter. For example: if there are m sensors, n source signals (source odors), the following convolution mixture mathematical model can be established:
wherein y (t), x (t), n (t) are m × 1 sensor response column vectors composed of m sensor responses, n × 1 signal source column vectors composed of n signal sources, and m × 1 noise vectors composed of observation noises of m sensors, respectively, s represents the length of the impulse response vector, and H represents the filter mixing matrix and is time-independent. It should be noted that in both machine olfactive studies and applications, x (t) is assumed to be essentially unchanged during one measurement (about 30 seconds to 1 minute).
Since the signal source content (or intensity) cannot be negative, the following objective function can be constructed:
as can be seen from the above formula, the method is a least square method with constraint conditions, and a signal source vector x (t) can be solved by a Lagrange multiplier method. However, Lagrange's multiplier method faces ill-conditioned solution problems due to the possible co-linearity (complete or high correlation) between sensors. To overcome the limitations of the Lagrange multiplier method, a ridge regression algorithm can be used for solving. Ridge regression is formulated as follows:
wherein X is the source odor vector, H is the transfer matrix, H*A pseudo-inverse matrix of H, Y a sensor response matrix, and γ (0. ltoreq. γ. ltoreq.1) as a regularization parameter, optionallyObtained by cross validation. However, if an unknown odor is present, it is due to the sensor array response matrix A or the filter response matrix HpThe method is described aiming at n source odors of specific application scenes, and does not contain response information of unknown gases, so that a relatively large error is inevitably existed by utilizing the optimization objective solution.
In order to solve the problems, optimization can be carried out according to different response signals with similar smells and the principle that response modes acquired by the same sensor are approximately the same, and then the advantages in the linear mixed model are obtained
If the distinction degree is not available, discarding; the z (t) component of the other 9 sensors is solved using one of the following equations.
Wherein, formula (1) is: x is the number ofi(t)≥0，xi(t) represents a noise signal; the formula (2) is: z (t) is 0;
equation (3) includes four solving modes, where equation (3-1) is: z (t) AeatThe formula (3-2) is: z (t) ═ 1-BebtThe formula (3-3) is: z (t) CectThe formula (3-4) is: z (t) kt2+lt+m。
When the z (t) component of all sensors is equal to 0, it means that there is no unknown smell. Therefore, the functional model can take into account the presence of unknown odors.
Wherein, when the response signal increases linearly and gradually reaches the steady state, Ae is adoptedatSolving for z (t); when the response signal decreases linearly and gradually reaches a steady state, 1-Be is adoptedbtSolving for z (t); when the response signal is a nonlinear response signal and suddenly changes to a pre-stable state, Ce is adoptedctSolving for z (t); when the response signal is a nonlinear response signal and suddenly changes to a stable state, adopting kt2+ lt + m solves for z (t).
S305, filtering unknown response signals by using an optimization function in the nonlinear hybrid model to obtain target response signals, and executing S306;
s306, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal through a chaotic genetic algorithm.
When the mixed response signal is a nonlinear response signal, an unknown response signal can be filtered by using an optimization function in a nonlinear mixed model to obtain a target response signal; wherein, the optimization function in the nonlinear mixed model is as follows: min | | y (t) -g (t) -z (t) |, y (t) represents a mixed response signal, g (t) represents a nonlinear response signal, and z (t) represents an unknown response signal.
S307, comparing the mixed response signal with the template response signal in the preset database, and determining the odor type and content corresponding to the mixed response signal.
In this embodiment, the nonlinear hybrid model is generated by using a multivariate adaptive regression spline method, and the specific generation steps are as follows:
acquiring experimental data of the mixed smell, and correcting the environmental factors of the experimental data;
dividing the corrected experimental data to obtain a plurality of intervals, and generating a corresponding basis function based on each interval;
and selecting the basis function with the highest accuracy by adopting an averaging error method, and determining the model corresponding to the basis function with the highest accuracy as a nonlinear mixed model. The smallest result of the averaging error method corresponds to the basis function with the highest accuracy.
The experimental data of the mixed smell can also be used for generating a preset smell database for comparing with the mixed response signal to be identified.
The correcting the environmental factors of the experimental data comprises the following steps: and (3) compensating the temperature, the humidity and the atmospheric pressure of the experimental data by adopting a neural network method.
For example: under specific application, a three-layer BP neural network is adopted for temperature and humidity compensation, the input of the neural network is sensor response, ambient temperature, ambient humidity and atmospheric pressure, and the output is corrected sensor response. Therefore, a preset smell database (the data stored in the database cannot cover the whole temperature, humidity and atmospheric pressure range) can be perfected, and the data coverage range in the database can be wider through correction.
The basis function is generally a linear function or a quadratic function, and parameters in the function can be corrected through a neural network so as to improve the accuracy of the basis function.
In this embodiment, the determining the odor type and content corresponding to the target response signal by the chaotic genetic algorithm includes:
generating a population of the chaotic genetic algorithm based on the target response signal;
carrying out individual evaluation on individuals in the population, and carrying out selection operation, cross operation and variation operation on the population subjected to the individual evaluation so as to iteratively update the population;
judging whether the current evolution algebra is larger than a preset maximum evolution algebra or not;
if so, outputting an optimal solution, and determining the odor type and content corresponding to the target response signal according to the optimal solution.
Specifically, the result obtained by the chaotic genetic algorithm can be used for estimating the concentration and the type of the mixed smell, wherein the accuracy of the estimation result of the concentration can reach 0.01mg/m3Thereby improving the accuracy of the quantitative analysis of the odor.
The method comprises the steps of judging whether a mixed response signal is a linear response signal or a nonlinear response signal when the mixed response signal of mixed smell is processed, filtering an unknown response signal by using an optimization function in a linear mixing model when the mixed response signal is the linear response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the type and content of the smell corresponding to the target response signal by a multiplier method; when the mixed response signal is a nonlinear response signal, an unknown response signal is filtered by using an optimization function in a nonlinear mixed model to obtain a target response signal, the target response signal is compared with a template response signal in a preset database, and the odor type and content corresponding to the target response signal are determined by a chaotic genetic algorithm, so that the recognition efficiency and accuracy are improved, and the accuracy of quantitative odor analysis is also improved.
It should be noted that, when there is no unknown response signal in the mixed response signal and the mixed response signal is a linear response signal, the method further includes:
filtering the unknown response signal by using an optimization function in the linear mixed model to obtain the target response signal;
and comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a ridge regression method.
Specifically, when the unknown response signal does not exist in the mixed response signal and the mixed response signal is a linear response signal, the optimization function and the ridge regression method in the linear mixing model can be used to determine the odor type and the content corresponding to the target response signal, so that the solving process is simpler and more convenient, and the recognition result is more accurate. Since the existing odor identification method can identify the type and content of the odor, the invention is not described in detail for identifying the type and content of the odor. The odor identification method disclosed by the invention is more operable in processing steps, and the processing model is more optimized, so that the obtained identification result is more accurate.
Based on any of the above embodiments, it should be noted that, the determining whether an unknown response signal exists in the mixed response signal, that is, determining whether an odor that does not exist in the preset database exists in the obtained mixed response signal, and if so, indicating that the unknown response signal exists in the mixed response signal; if the unknown response signal does not exist in the mixed response signal, that is, all signals in the mixed response signal can be found in the preset database.
In the following, a machine smell-based smell recognition device according to an embodiment of the present invention is described, and a machine smell-based smell recognition device described below and a machine smell-based smell recognition method described above may be referred to each other.
Referring to fig. 4, an odor recognition device based on machine olfaction according to an embodiment of the present invention includes:
an obtaining module 401, configured to obtain a mixed response signal of a target mixed smell;
a determining module 402, configured to determine whether an unknown response signal exists in the mixed response signal;
an identifying module 403, configured to, when an unknown response signal exists in the mixed response signal, filter the unknown response signal from the mixed response signal, obtain a target response signal, compare the target response signal with a template response signal in a preset database, and determine a type and a content of an odor corresponding to the target response signal.
Wherein the identification module comprises:
the judging unit is used for judging whether the mixed response signal is a linear response signal or not;
the recognition unit is used for filtering the unknown response signal by using an optimization function in a linear mixing model when the mixed response signal is a linear response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a multiplier method;
Wherein, still include:
the linear processing module is used for filtering the unknown response signal by utilizing an optimization function in the linear mixing model to obtain the target response signal when the unknown response signal does not exist in the mixed response signal and the mixed response signal is a linear response signal; and comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a ridge regression method.
Wherein, still include:
the nonlinear processing module is used for filtering the unknown response signal by utilizing an optimization function in a nonlinear hybrid model to obtain the target response signal when the hybrid response signal is a nonlinear response signal; comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal through a chaotic genetic algorithm;
wherein the optimization function in the nonlinear hybrid model is: min | | y (t) -g (t) -z (t) |, y (t) represents a mixed response signal, g (t) represents a nonlinear response signal, and z (t) represents an unknown response signal.
Wherein the nonlinear processing module further comprises:
the generating unit is used for generating a nonlinear mixed model by using a multivariate adaptive regression spline method;
the generation unit includes:
the acquisition subunit is used for acquiring experimental data of the mixed smell and correcting the environmental factors of the experimental data;
the dividing subunit is used for dividing the corrected experimental data to obtain a plurality of intervals and generating corresponding basis functions based on each interval;
and the determining subunit is used for selecting the basis function with the highest accuracy by adopting an averaging error method, and determining the model corresponding to the basis function with the highest accuracy as the nonlinear mixed model.
Wherein the nonlinear processing module is specifically configured to:
generating a population of the chaotic genetic algorithm based on the target response signal;
carrying out individual evaluation on individuals in the population, and carrying out selection operation, cross operation and variation operation on the population subjected to the individual evaluation so as to iteratively update the population;
judging whether the current evolution algebra is larger than a preset maximum evolution algebra or not;
if yes, outputting an optimal solution, and determining the odor type and content corresponding to the target response signal according to the optimal solution.
In the following, a smell recognition device based on machine olfaction according to an embodiment of the present invention is introduced, and the smell recognition device based on machine olfaction described below and the smell recognition method and apparatus based on machine olfaction described above may be referred to each other.
Referring to fig. 5, an embodiment of the present invention provides a smell recognition device based on machine olfaction, including:
a memory 501 for storing a computer program;
a processor 502 for implementing the steps of the machine scent-based scent recognition method according to any of the embodiments described above when executing the computer program.
In the following, a computer-readable storage medium according to an embodiment of the present invention is described, and a computer-readable storage medium described below and a method, an apparatus, and a device for identifying smells based on machine olfaction described above may be referred to each other.
A computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a machine scent-based scent recognition method according to any of the embodiments described above.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (8)
1. A smell recognition method based on machine olfaction is characterized by comprising the following steps:
acquiring a mixed response signal of the target mixed smell;
judging whether an unknown response signal exists in the mixed response signal;
if so, filtering the unknown response signal from the mixed response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal;
wherein, the step of filtering the unknown response signal from the mixed response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal comprises:
judging whether the mixed response signal is a linear response signal or not;
if so, filtering the unknown response signal by using an optimization function in a linear mixed model to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by using a multiplier method;
wherein the optimization function in the linear hybrid model is:
y (t) represents a mixed response signal, P represents an impulse response signal, S represents the length of said impulse response signal, H represents a filter mixing matrix, z (t) represents an unknown response signal, t represents a time variable, and x (t-P) represents a noise signal.
2. The machine-olfaction-based scent recognition method of claim 1, wherein when the mixed response signal is a linear response signal without an unknown response signal being present in the mixed response signal, further comprising:
filtering the unknown response signal by using an optimization function in the linear mixed model to obtain the target response signal;
and comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a ridge regression method.
3. The machine-olfaction-based scent recognition method of claim 1, wherein when the mixed response signal is a non-linear response signal, further comprising:
filtering the unknown response signal by using an optimization function in a nonlinear hybrid model to obtain the target response signal;
comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal through a chaotic genetic algorithm;
wherein the optimization function in the nonlinear hybrid model is: min | | y (t) -g (t) -z (t) |, y (t) represents a mixed response signal, g (t) represents a nonlinear response signal, and z (t) represents an unknown response signal.
4. The machine olfaction-based odor recognition method according to claim 3, wherein the non-linear mixture model is generated by using a multivariate adaptive regression spline method, and the specific generation steps are as follows:
acquiring experimental data of the mixed smell, and correcting the environmental factors of the experimental data;
dividing the corrected experimental data to obtain a plurality of intervals, and generating a corresponding basis function based on each interval;
and selecting the basis function with the highest accuracy by adopting an averaging error method, and determining the model corresponding to the basis function with the highest accuracy as the nonlinear mixed model.
5. The machine olfaction-based scent recognition method according to claim 3, wherein the determining the scent type and content corresponding to the target response signal through a chaotic genetic algorithm comprises:
generating a population of the chaotic genetic algorithm based on the target response signal;
carrying out individual evaluation on individuals in the population, and carrying out selection operation, cross operation and variation operation on the population subjected to the individual evaluation so as to iteratively update the population;
judging whether the current evolution algebra is larger than a preset maximum evolution algebra or not;
if yes, outputting an optimal solution, and determining the odor type and content corresponding to the target response signal according to the optimal solution.
6. A smell recognition device based on machine olfaction is characterized by comprising:
the acquisition module is used for acquiring a mixed response signal of the target mixed smell;
the judging module is used for judging whether an unknown response signal exists in the mixed response signal;
the recognition module is used for filtering the unknown response signal from the mixed response signal to obtain a target response signal when the unknown response signal exists in the mixed response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal;
wherein the identification module comprises:
the judging unit is used for judging whether the mixed response signal is a linear response signal or not;
the recognition unit is used for filtering the unknown response signal by using an optimization function in a linear mixing model when the mixed response signal is a linear response signal to obtain a target response signal, comparing the target response signal with a template response signal in a preset database, and determining the odor type and content corresponding to the target response signal by a multiplier method;
wherein the optimization function in the linear hybrid model is:
y (t) represents a mixed response signal, P represents an impulse response signal, S represents the length of said impulse response signal, H represents a filter mixing matrix, z (t) represents an unknown response signal, t represents a time variable, and x (t-P) represents a noise signal.
7. A smell recognition device based on machine olfaction, comprising:
a memory for storing a computer program;
a processor for implementing the steps of the machine scent-based scent recognition method according to any one of claims 1 to 5 when executing the computer program.
8. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the machine scent-based scent recognition method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810628801.0A CN108932483B (en) | 2018-06-19 | 2018-06-19 | Smell identification method and device based on machine olfaction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810628801.0A CN108932483B (en) | 2018-06-19 | 2018-06-19 | Smell identification method and device based on machine olfaction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108932483A CN108932483A (en) | 2018-12-04 |
| CN108932483B true CN108932483B (en) | 2022-03-25 |
Family
ID=64446013
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810628801.0A Active CN108932483B (en) | 2018-06-19 | 2018-06-19 | Smell identification method and device based on machine olfaction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108932483B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110057975B (en) * | 2019-04-22 | 2023-07-25 | 广州德芯半导体科技有限公司 | Sniffing descriptor acquisition method, system and related device |
| CN111024776A (en) * | 2019-12-19 | 2020-04-17 | 安徽康佳同创电器有限公司 | Electronic nose, smell identification method and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1482453A (en) * | 2003-07-11 | 2004-03-17 | 华东理工大学 | Machine olfaction odor distinguishing method based on modularized composite neural net |
| US20090048689A1 (en) * | 2007-08-14 | 2009-02-19 | David Andrew Pelton | Actuation response oscillation detection monitor |
| CN101470121A (en) * | 2007-12-29 | 2009-07-01 | 广东工业大学 | Built-in bionic smell recognition method and device |
| CN101788517A (en) * | 2010-02-23 | 2010-07-28 | 广东工业大学 | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction |
| CN104460948A (en) * | 2013-09-15 | 2015-03-25 | 南京大五教育科技有限公司 | Olfactory sensing technology |
| CN108123724A (en) * | 2017-12-22 | 2018-06-05 | 西安烽火电子科技有限责任公司 | A kind of communication system based on shortwave narrowband waveform |
-
2018
- 2018-06-19 CN CN201810628801.0A patent/CN108932483B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1482453A (en) * | 2003-07-11 | 2004-03-17 | 华东理工大学 | Machine olfaction odor distinguishing method based on modularized composite neural net |
| US20090048689A1 (en) * | 2007-08-14 | 2009-02-19 | David Andrew Pelton | Actuation response oscillation detection monitor |
| CN101470121A (en) * | 2007-12-29 | 2009-07-01 | 广东工业大学 | Built-in bionic smell recognition method and device |
| CN101788517A (en) * | 2010-02-23 | 2010-07-28 | 广东工业大学 | Pungent traditional Chinese medicine odor fingerprint map construction system and method based on bionic olfaction |
| CN104460948A (en) * | 2013-09-15 | 2015-03-25 | 南京大五教育科技有限公司 | Olfactory sensing technology |
| CN108123724A (en) * | 2017-12-22 | 2018-06-05 | 西安烽火电子科技有限责任公司 | A kind of communication system based on shortwave narrowband waveform |
Non-Patent Citations (4)
| Title |
|---|
| Classification of Fruit Species Based on Effective Sensor Array Selection;Biliang Xia et al;《2016 IEEE International Conference on Consumer Electronics-China (ICCE-China)》;20170213;全文 * |
| Electronic Noses and Tongues: Applications for the Food and Pharmaceutical Industries;Elizabeth A. Baldwin et al;《Sensors》;20110502;第11卷(第5期);全文 * |
| 基于机器嗅觉的柑橘品种无损检测与识别;彭柯 等;《江西农业大学学报》;20171020;第39卷(第5期);全文 * |
| 机器嗅觉在气味类别与浓度同时估计中的应用;高大启;《2004年中国香料香精学术研讨会》;20040331;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108932483A (en) | 2018-12-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ni et al. | Adaptive learning in tracking control based on the dual critic network design | |
| CN108932483B (en) | Smell identification method and device based on machine olfaction | |
| JP2010157492A (en) | Battery model identification method | |
| CN107194026B (en) | Bayesian network-based absorption tower desulfurization process modeling method | |
| Deng et al. | Feedback-linearization-based neural adaptive control for unknown nonaffine nonlinear discrete-time systems | |
| Formentin et al. | Enhancing statistical performance of data-driven controller tuning via L2-regularization | |
| CN111433689B (en) | Generation of control systems for target systems | |
| Marconato et al. | Improved initialization for nonlinear state-space modeling | |
| Seppelt et al. | “It was an artefact not the result”: A note on systems dynamic model development tools | |
| JP2005147793A (en) | Odor measuring apparatus | |
| Zhang et al. | A correlation-test-based validation procedure for identified neural networks | |
| CN113759708A (en) | System optimization control method and device and electronic equipment | |
| Cadini et al. | A Monte Carlo method for the model-based estimation of nuclear reactor dynamics | |
| Du et al. | A novel locally regularized automatic construction method for RBF neural models | |
| Costa et al. | Optimal recursive fuzzy model identification approach based on particle swarm optimization | |
| Han et al. | Identification of CSTR using extreme learning machine based Hammerstein-Wiener model | |
| Bessa et al. | Recursive system identification using outlier-robust local models | |
| Chen | Reconstructing genetic regulatory networks using two-step algorithms with the differential equation models of neural networks | |
| CN117059198B (en) | Emission list feedback updating method, system and equipment based on response surface model | |
| CN117556214A (en) | Self-adaptive peculiar smell treatment method, system, equipment and medium | |
| Zhang et al. | Sparse Identification of Fractional Chaotic Systems based on the time-domain data | |
| CN110622075B9 (en) | System identification device and system identification method | |
| CN116266214A (en) | Equipment monitoring and early warning method based on STL time sequence decomposition | |
| Zhang et al. | Robust Parameter Estimation for a Class of Nonlinear System With EM Algorithm | |
| Jian-hong | Recursive Algorithms of Closed Loop Identification with a Tailor Made Parameterization |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |