CN113267535A - Intelligent gas identification method and device - Google Patents
Intelligent gas identification method and device Download PDFInfo
- Publication number
- CN113267535A CN113267535A CN202110826601.8A CN202110826601A CN113267535A CN 113267535 A CN113267535 A CN 113267535A CN 202110826601 A CN202110826601 A CN 202110826601A CN 113267535 A CN113267535 A CN 113267535A
- Authority
- CN
- China
- Prior art keywords
- gas
- voltage
- gas sensor
- processor
- intelligent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/0004—Gaseous mixtures, e.g. polluted air
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computational Mathematics (AREA)
- Analytical Chemistry (AREA)
- Evolutionary Computation (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Software Systems (AREA)
- Pathology (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Artificial Intelligence (AREA)
- Mathematical Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Operations Research (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Combustion & Propulsion (AREA)
- Food Science & Technology (AREA)
- Medicinal Chemistry (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Electrochemistry (AREA)
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
Abstract
The invention discloses an intelligent gas identification device. The intelligent gas identification device comprises a hardware part and a software part, wherein the hardware part comprises at least one gas sensor, a processor, a voltage modulation unit, a communication unit and a data acquisition unit; the software part comprises a data processing module and a gas identification model building module. The invention also discloses an intelligent gas identification method. The intelligent gas identification method and the intelligent gas identification device have a simplified structure, can only test the dynamic response of a single gas sensor to different types of gases or the same gas with different concentrations under different voltage modulation modes (temperature modes) during testing, and then combine a data processing module to perform signal preprocessing and feature extraction processing, and construct a deeply-learned gas identification model by using a gas identification model construction module so as to achieve the purpose of identifying and classifying the gases; and the method has the advantages of simple operation, high identification precision, small material consumption, strong stability, easy realization and low technical requirement on equipment or workers.
Description
Technical Field
The invention relates to the technical field of gas identification, in particular to an intelligent gas identification method and an intelligent gas identification device.
Background
In addition, in food detection, whether food is deteriorated or not can be judged through the identification of specific gas, the eating safety is guaranteed, the quality states of tea leaves and wine can be judged, the quality levels of the tea leaves and the wine can be effectively distinguished, and therefore the gas detection and identification are necessary.
The existing gas detector generally determines the mass proportion of molecules or atoms in gas through spectral analysis so as to effectively identify the gas, but the gas detector has extremely high requirements on technical personnel in use and extremely high manufacturing cost, the gas identification is realized through spectral analysis, no matter the labor cost or the equipment cost is very high, certain industrial and living scenes cannot bear, so that the industrial environment cannot realize full-range detection and alarm and the safety of workers is difficult to guarantee, and the gas detector also has an electronic nose system based on a sensor array.
Disclosure of Invention
Based on this, the present invention provides an intelligent gas identification method and an apparatus thereof, which have a simplified structure, and during testing, only dynamic responses of a single gas sensor to different types of gases or the same gas with different concentrations in different voltage modulation modes (temperature modes) can be tested, and then a data processing module is combined with signal preprocessing, feature extraction and deep learning technologies, and a gas identification model construction module is used to construct a deep learning gas identification model, so as to achieve the purpose of gas identification and classification.
The purpose of the invention is realized by the following technical scheme:
in a first aspect, an intelligent gas identification device comprises a hardware part and a software part, wherein the hardware part comprises at least one gas sensor, a processor, a voltage modulation unit, a communication unit and a data acquisition unit, the voltage modulation unit is electrically connected with the processor and the gas sensor and modulates the voltage at two ends of the gas sensor according to preset parameters of the processor, and the data acquisition unit is electrically connected with the processor and the gas sensor module respectively and acquires training response signals of the gas sensor under the modulated voltage; the software part comprises a data processing module and a gas recognition model building module, wherein the data processing module is used for receiving training response signals transmitted by the processor through the communication unit and carrying out preprocessing and characteristic extraction on the training response signals to obtain a gas training data set, and the gas recognition model building module is used for feeding the gas training data set obtained by the data processing module into a deep neural network to build a deep learning gas recognition model.
The intelligent gas identification device is simple in structure, based on the influence and control of the gas-sensitive characteristics of the gas sensor by the temperature of the device, the response to different gases/odors is different in different working temperature ranges, the dynamic response to different types of gases or the same gas with different concentrations under different voltage modulation modes (temperature modes) of a single gas sensor can be tested during testing, and a gas identification model for deep learning is constructed by the aid of the gas identification model construction module in combination with signal preprocessing, feature extraction and deep learning technologies of the data processing module, so that the purpose of identifying and classifying the gases is achieved. The intelligent gas identification method and the intelligent gas identification device have the advantages of simple operation, high identification precision, small material consumption, strong stability, easy realization and low technical requirements on equipment or workers.
Further preferably, the voltage modulation unit includes a digital-to-analog conversion subunit and an operational amplifier circuit, the digital-to-analog conversion subunit is electrically connected to the processor and the operational amplifier circuit, and the operational amplifier circuit is electrically connected to the gas sensor.
Further preferably, the data acquisition unit comprises a data acquisition chip and an analog-to-digital conversion subunit, the analog-to-digital conversion subunit is electrically connected with the data acquisition chip and the gas sensor, and the data acquisition chip is electrically connected with the processor.
Further preferably, the hardware part of the intelligent gas identification device further comprises a power supply and a voltage stabilizing unit, and the voltage stabilizing unit is electrically connected with the power supply and the processor.
Further preferably, the voltage stabilizing unit is a low dropout linear regulator.
Further preferably, the hardware part of the intelligent gas identification device further comprises a server, and the server is in communication connection with the processor through the communication unit.
Further preferably, the gas sensor is one of a ceramic wafer type gas sensor, a ceramic tube type gas sensor, or a MEMS sensor.
In a second aspect, an intelligent gas identification method uses the intelligent gas identification device described in any one of the above to perform the following steps:
placing at least one gas sensor in an environment containing a certain gas, and enabling the initial voltage at two ends of the gas sensor to be in a stable state;
enabling the processor to receive a starting modulation signal, triggering the voltage modulation unit to modulate the voltage of the gas sensor according to preset parameters, and simultaneously starting the data acquisition unit to acquire a training response signal of the gas sensor under the modulation voltage;
receiving a training response signal transmitted by a processor through a data processing module, and carrying out preprocessing and feature extraction on the training response signal to obtain a gas training data set;
feeding the gas training data set obtained by the data processing module into a deep neural network through a gas identification model building module to build a deep learning gas identification model;
the gas sensor is placed in an environment containing gas to be identified, the gas sensor is subjected to the same voltage modulation by the voltage modulation unit, meanwhile, response signals to be tested of the gas sensor under the modulated voltage are collected by the data acquisition unit, the data processing module is used for receiving the response signals to be tested transmitted by the processor and carrying out the same preprocessing and characteristic extraction on the response signals to be tested, a data set of the gas to be tested is obtained and sent to the gas identification model, and gas identification is achieved.
Further preferably, the trigger voltage modulation unit modulates the voltage of the gas sensor according to preset parameters, specifically, the trigger voltage modulation unit modulates the voltage of the gas sensor according to the preset parameters and by adopting one or a combination of square waves, step waves, sawtooth waves, triangular waves or sine waves;
the starting data acquisition unit acquires a training response signal of the gas sensor under the modulation voltage and the data acquisition unit acquires a response signal to be tested of the gas sensor under the modulation voltage, wherein the training response signal and the response signal to be tested are acquired repeatedly for 20-100 times under the same modulation voltage, modulation temperature range, sampling frequency and sampling duration;
the deep neural network is one of a deep convolutional neural network CNN, a deep back propagation neural network BPNN and a deep cyclic neural network RNN.
Further preferably, the receiving, preprocessing and feature extraction of the training response signal transmitted by the processor through the data processing module and the receiving, preprocessing and feature extraction of the response signal to be tested transmitted by the processor through the data processing module include performing sensitivity preprocessing, concentration normalization preprocessing and discrete wavelet transform feature extraction processing on the training response signal and the response signal to be tested;
the sensitivity pretreatment is specifically represented by the formula G (cc, u, t) = Rair(t)/ Rgas(cc, u, t); the concentration normalization pretreatment is specifically represented by the formula Y (cc, u, t) = (G (cc, u, t) -G (cc, u, t)min)/ (G(cc,u,t)max- G(cc,u,t)min) Carrying out the following steps; the discrete wavelet transform characteristicsThe extraction treatment is specifically carried out through Daubechies wavelets; where t represents the sampling time, u represents the u-th test, cc represents a certain gas or a certain gas concentration, R represents the training response signal or the response signal to be identified, G represents the sensitivity of the signal, R represents the sensitivity of the signalair(t) represents the training response signal of the air at the sampling time t, Rgas(cc, u, t) represents a training response signal of the gas at the u-th test gas concentration cc at the sampling time t, Y (cc, u, t) represents a concentration normalization pre-processing expression, G (cc, u, t) represents the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t, G (cc, u, t)maxG (cc, u, t) which represents the maximum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time tminRepresents the minimum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t.
Compared with the prior art, the intelligent gas identification method and the intelligent gas identification device have the advantages that the structure is simplified, the gas-sensitive characteristics of the gas sensor are influenced and controlled by the temperature of the device, the response to different gases/smells is different in different working temperature ranges, the dynamic response to different types of gases or the same gas with different concentrations under different voltage modulation modes (temperature modes) of a single gas sensor can be tested during testing, and a gas identification model for deep learning is built by combining the signal preprocessing, feature extraction and deep learning technologies of the data processing module, so that the purpose of identifying and classifying the gases is achieved. The intelligent gas identification method and the intelligent gas identification device have the advantages of simple operation, high identification precision, small material consumption, strong stability, easy realization and low technical requirements on equipment or workers.
For a better understanding and practice, the invention is described in detail below with reference to the accompanying drawings.
Drawings
Fig. 1 is a schematic block diagram of an intelligent gas identification device of the present invention.
Fig. 2 is a schematic diagram of an ESP32 master control chip of the intelligent gas identification device of the present invention.
Fig. 3 is a circuit for implementing the voltage modulation unit of the intelligent gas identification device of the present invention.
Fig. 4 is a schematic connection diagram of a plurality of gas sensors in the intelligent gas identification device of the present invention.
Fig. 5 is a flow chart of the intelligent gas identification method of the present invention.
Fig. 6 is a schematic diagram of the intelligent gas identification method of the present invention using triangular waves for voltage modulation.
Fig. 7 is a graph of the resistance response change between the heating voltage and the sensing layer of the gas sensor when the intelligent gas identification method of the present invention employs square waves for voltage modulation.
Fig. 8 is a CNN structural diagram of a gas recognition model constructed by the intelligent gas recognition method of the present invention.
FIG. 9 is a diagram of the change of cross entropy and gas recognition accuracy when the intelligent gas recognition method trains the gas recognition model. The graph a is a cross entropy change situation graph of the deep learning gas identification model, and the graph b is a gas identification accuracy rate change situation graph of the deep learning gas identification model.
FIG. 10 is a recognition confusion matrix diagram of a deep-learned gas recognition model after training by the intelligent gas recognition method of the present invention.
Detailed Description
The terms of orientation of up, down, left, right, front, back, top, bottom, and the like, referred to or may be referred to in this specification, are defined relative to their configuration, and are relative concepts. Therefore, it may be changed according to different positions and different use states. Therefore, these and other directional terms should not be construed as limiting terms.
The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
The terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
An intelligent gas identification device is shown in fig. 1 and comprises a hardware part and a software part, wherein the hardware part comprises at least one gas sensor, a processor, a voltage modulation unit, a communication unit and a data acquisition unit, the voltage modulation unit is electrically connected with the processor and the gas sensor and modulates the voltage at two ends of the gas sensor according to preset parameters of the processor, and the data acquisition unit is electrically connected with the processor and a gas sensor module respectively and acquires training response signals of the gas sensor under the modulated voltage; the software part comprises a data processing module and a gas recognition model building module, wherein the data processing module is used for receiving training response signals transmitted by the processor through the communication unit and carrying out preprocessing and characteristic extraction on the training response signals to obtain a gas training data set, and the gas recognition model building module is used for feeding the gas training data set obtained by the data processing module into a deep neural network to build a deep learning gas recognition model.
Specifically, the intelligent gas identification device of the present invention preferably contains only a single gas sensor to reduce the complexity of the overall structure.
Preferably, the gas sensor is one of a ceramic wafer type gas sensor, a ceramic tube type gas sensor, or a MEMS sensor. The sensitive material of the gas sensor is tin oxide (SnO)2) Nickel oxide (NiO), tungsten oxide (WO)3) Indium oxide (In)2O3) Zinc oxide (ZnO), copper oxide (CuO), or cobalt oxide (Co)3O4) One or a combination thereof.
The gas sensor can be selected according to actual production conditions.
Preferably, the voltage modulation unit includes a digital-to-analog conversion subunit and an operational amplifier circuit, the digital-to-analog conversion subunit is electrically connected to the processor and the operational amplifier circuit, and the operational amplifier circuit is electrically connected to the gas sensor.
Specifically, in one embodiment, as shown in fig. 2-3, the processor is preferably an ESP32 main control chip, which has multiple analog-to-digital conversion interfaces to facilitate conversion processing of analog signals, and the SPI bus also facilitates control of the driving capability of the digital-to-analog conversion subunit to output a specific voltage and amplify the voltage through the operational amplifier circuit. Further, a digital-to-analog conversion chip of the digital-to-analog conversion subunit is an AD5621BKSZ-REEL7 digital-to-analog conversion chip, an operational amplification chip of the operational amplification circuit is an OPA4313 operational amplification chip, and the communication unit is a WIFI communication unit or a bluetooth communication unit. More specifically, the processor is electrically connected with the SYNC, SCLK and SDIN pins of the digital-to-analog conversion sub-unit through the GPIO15, GPIO18 and GPIO23 pins thereof, respectively, and the VOUT pin of the digital-to-analog conversion sub-unit is electrically connected with the + INA, + INB, + INC and + IND pins of the operational amplification circuit.
The processor, the digital-to-analog conversion subunit and the operational amplification circuit which are connected in this way enable the structure of the intelligent gas identification device of the invention to be more reasonable and compact, and improve the integral integration level.
In other embodiments, the voltage modulation unit is a PWM voltage modulation unit, the processor is a 16F877A single chip, and the communication unit is a serial communication unit.
The PWM voltage modulation unit adjusts the duty ratio, and obtains copy-adjustable voltage after RC filtering so as to thermally modulate the voltage at two ends of the gas sensor.
Although the intelligent gas identification device of the present invention preferably includes a single gas sensor, a plurality of gas sensors may be provided, and specifically, as shown in fig. 4, the connection of the plurality of gas sensors may be realized by a time division multiplexing circuit or a path switching circuit, so as to perform multidimensional sampling, increase the data volume reasonably, and improve the accuracy and efficiency of the intelligent gas identification device of the present invention in gas identification.
Preferably, the data acquisition unit comprises a data acquisition chip and an analog-to-digital conversion subunit, the analog-to-digital conversion subunit is electrically connected with the data acquisition chip and the gas sensor, and the data acquisition chip is electrically connected with the processor.
The analog-to-digital conversion subunit is used for converting the analog signal into a digital signal, so that subsequent signal processing and acquisition and reference are facilitated.
Preferably, the hardware part of the intelligent gas identification device further comprises a power supply and a voltage stabilizing unit, and the voltage stabilizing unit is electrically connected with the power supply and the processor.
The voltage stabilizing unit can stabilize the power supply voltage, and further protect the use safety of the whole intelligent gas identification device.
Preferably, the voltage stabilizing unit is a low dropout linear regulator. Specifically, the voltage stabilizing unit is an AMS1117 low dropout linear regulator.
The low-dropout linear regulator is beneficial to the intelligent gas identification device to obtain pure power voltage.
Preferably, the hardware part of the intelligent gas identification device further comprises a server, and the server is in communication connection with the processor through the communication unit.
In particular, the server has a display screen through which a customer or operator can access data regarding the smart gas identification device of the present invention.
The server can be convenient for operating the processor and the data acquisition unit, and can also be used for updating programs in the processor and the data acquisition unit at regular time.
Preferably, the communication unit is a wireless communication unit or a wired communication unit.
Specifically, the communication unit is one of a WIFI communication unit, a serial communication unit and a Bluetooth communication unit.
The communication unit facilitates communication among the units and the modules in the intelligent gas identification device, and can improve the efficiency and convenience of data transmission.
Preferably, the data processing module and the gas identification model construction module may be implemented in the form of a computer-readable storage medium storing a single program that, when executed by a processor, performs its functions.
It should be appreciated that the computer-readable storage medium is any data storage device that can store data or programs which can thereafter be read by a computer system. Examples of computer-readable storage media include: read-only memory, random access memory, CD-ROM, HDD, DVD, magnetic tape, optical data storage devices, and the like.
The computer readable storage medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.
Program code embodied on a computer readable storage medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, Radio Frequency (RF), etc., or any suitable combination of the foregoing.
In some embodiments, the computer-readable storage medium may also be non-transitory.
The invention also discloses an intelligent gas identification method, as shown in fig. 5, the intelligent gas identification device is utilized to perform the following steps:
s1: at least one gas sensor is placed in an environment containing a gas and has an initial voltage across it at a steady state.
Specifically, in the present embodiment, the number of gas sensors is preferably one.
S2: and the processor is enabled to receive the starting modulation signal, the voltage modulation unit is triggered to perform voltage modulation on the gas sensor according to the preset parameters, and the data acquisition unit is started to acquire a training response signal of the gas sensor under the modulation voltage.
Specifically, the gas sensor may be placed in several different gas environments, the same voltage modulation may be performed on the gas sensor according to the same preset parameters, and training response signals of the gas sensor under different gas environments and at the modulated voltage may be collected, for example, gas environments containing the same kind of gas but different gas concentrations, or gas environments containing different kinds of gas, such as o-xylene, p-xylene, m-xylene, toluene, and benzene.
More specifically, the modulation waveform adopted for voltage modulation of the gas sensor is one or a combination of square wave, step wave, sawtooth wave, triangular wave or sine wave.
As shown in fig. 6, the voltage modulation is performed by using a triangular wave, the processor sends preset parameters (i.e. specific voltage data) in a normal state, and after the collection is started, the sent voltage data is gradually increased until the sampling is finished, and the default data is sent again.
Fig. 7 is a graph showing the resistance response change between the heating voltage and the sensing layer of the gas sensor when the intelligent gas identification method of the present invention employs square waves for voltage modulation.
Further, the training response signal is collected under the same modulation voltage, modulation temperature range, sampling frequency and sampling duration, for example, the modulation voltage of 2.5-3.8V, the modulation temperature range of 150-.
Still further, a gas sensor characteristic change curve can be drawn according to the acquired training response signals. The gas sensor comprises a heater and a sensing layer, the heater can adjust the temperature of the sensing layer through voltage, and a drawn characteristic change curve of the gas sensor represents change data of the resistance of the sensing layer of the gas sensor along with temperature change of thermal modulation (voltage modulation), namely a set of training response signals of the gas sensor under the voltage modulation.
S3: and receiving the training response signal transmitted by the processor through the data processing module, and performing preprocessing and feature extraction on the training response signal to obtain a gas training data set.
S3 is to remove the characteristics (such as temperature resistivity, concentration and noise signal) independent of the gas type step by step mainly by performing the preprocessing and the characteristic extraction by performing the geometric characteristic transformation and the frequency domain information extraction on the response signal of the gas sensor.
Specifically, forThe training response signal is preprocessed and feature extracted, wherein the preprocessing comprises sensitivity preprocessing, concentration normalization preprocessing and discrete wavelet transform feature extraction processing; the sensitivity pretreatment is specifically represented by the formula G (cc, u, t) = Rair(t)/ Rgas(cc, u, t); the concentration normalization pretreatment is specifically performed by the formula Y (cc, u, t) = (G (cc, u, t) -G (cc, u, t) min)/(G (cc, u, t) max-G (cc, u, t) min); the discrete wavelet transform feature extraction processing is specifically performed through Daubechies wavelets to filter out noise signals in sensitivity during concentration normalization, remove detail coefficients, and obtain a gas training data set by using low-frequency approximation coefficients. Where t represents the sampling time, u represents the u-th test, cc represents a certain gas or a certain gas concentration, R represents the training response signal or the response signal to be identified, G represents the sensitivity of the signal, R represents the sensitivity of the signalair(t) represents the training response signal of the air at the sampling time t, Rgas(cc, u, t) represents a training response signal of the gas at the u-th test gas concentration cc at the sampling time t, Y (cc, u, t) represents a concentration normalization pre-processing expression, G (cc, u, t) represents the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t, G (cc, u, t)maxG (cc, u, t) which represents the maximum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time tminRepresents the minimum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t.
S4: and feeding the gas training data set obtained by the data processing module into a deep neural network through a gas recognition model building module to build a deep learning gas recognition model.
Specifically, the deep neural network is one of a deep convolutional neural network CNN, a deep back propagation neural network BPNN and a deep recurrent neural network RNN. Further, the gas identification model is constructed using one of a TensorFlow framework, a PyTorch framework, a Keras framework, an MXNet framework, or a Caffe2 framework.
More specifically, the specific process of constructing the deep learning gas identification model is as follows: and constructing a CNN model in a PC by adopting Python-Tensorflow. The CNN network structure is shown in FIG. 8 as I-5x5-2x 2-200-P, and the input layer I has 100 nodes of 10x10, which is the gas training data of the label corresponding to different gases or the same gas with different concentrations after signal preprocessing and feature extraction. The output layer has P nodes representing P gases to be identified. The convolutional layer of CNN has 6 convolutional kernels of 5x5, the pooling layer is the maximum pooling of 2x2, and the number of nodes of two fully-connected layers is 200. The activation function of the CNN adopts a Relu function, the loss function adopts a method of combining Cross entropy (Cross entropy) and a Softmax output function, and the weight of the network is initialized by normally distributed random numbers. In the deep neural network training process, the parameter gradient is calculated by using a back propagation algorithm, and the gradient value is updated by using a random gradient descent method. The change conditions of the cross entropy and the recognition accuracy of the neural network model along with the increase of the number of training rounds are shown in fig. 9, and finally the deep learning gas recognition model is obtained.
Additionally, the activation function and the output function may also be a Sigmoid function or a Tanh function; the loss function may also be a mean square error loss function or a log-likelihood loss function; the weight initialization method of the network can also be constant initialization, Gaussian distribution initialization, uniform distribution initialization, positive _ unit initialization, msra initialization, xavier initialization or bilinear initialization; the specific use can be selected according to specific situations.
S5: the gas sensor is placed in an environment containing gas to be identified, the gas sensor is subjected to the same voltage modulation by the voltage modulation unit, meanwhile, response signals to be tested of the gas sensor under the modulated voltage are collected by the data acquisition unit, the data processing module is used for receiving the response signals to be tested transmitted by the processor and carrying out the same preprocessing and characteristic extraction on the response signals to be tested, a data set of the gas to be tested is obtained and sent to the gas identification model, and gas identification is achieved.
The gas sensor is placed in an environment containing gas to be identified, wherein the gas to be identified is ortho-xylene, para-xylene, meta-xylene, toluene or benzene, the gas sensor in the environment containing the gas to be identified is subjected to voltage modulation similar to S2 and processing similar to S3, then a gas identification model is utilized to carry out gas identification on a data set of the gas to be identified by using trained parameters (weights and thresholds) in forward propagation, and an identification confusion matrix is shown in FIG. 10.
Compared with the prior art, the intelligent gas identification method and the intelligent gas identification device have the advantages that the structure is simplified, the gas-sensitive characteristics of the gas sensor are influenced and controlled by the temperature of the device, the response to different gases/smells is different in different working temperature ranges, the dynamic response to different types of gases or the same gas with different concentrations under different voltage modulation modes (temperature modes) of a single gas sensor can be tested during testing, and a gas identification model for deep learning is built by combining the signal preprocessing, feature extraction and deep learning technologies of the data processing module, so that the purpose of identifying and classifying the gases is achieved. The intelligent gas identification method and the intelligent gas identification device have the advantages of simple operation, high identification precision, small material consumption, strong stability, easy realization and low technical requirements on equipment or workers.
The invention converts the space expansion mode of the traditional electronic nose array technology into the time expansion mode, effectively utilizes the temperature control effect of the gas sensor, develops the characteristic extraction algorithm of the transient thermal modulation response of the single oxide gas sensor, can effectively extract the internal characteristics of different adsorbed gas molecules, greatly expands the odor recognition capability of the single gas sensor, is expected to promote the deep application of the metal oxide semiconductor gas sensor in the emerging fields of indoor and outdoor air quality monitoring, food safety, respiration monitoring, automobile manufacturing, danger monitoring and early warning and the like, and realizes the intelligent industrial revolution of the information physical fusion system in the artificial olfaction field.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.
Claims (10)
1. An intelligent gas identification device is characterized by comprising a hardware part and a software part, wherein the hardware part comprises at least one gas sensor, a processor, a voltage modulation unit, a communication unit and a data acquisition unit, the voltage modulation unit is electrically connected with the processor and the gas sensor and modulates the voltage at two ends of the gas sensor according to preset parameters of the processor, and the data acquisition unit is respectively electrically connected with the processor and a gas sensor module and acquires training response signals of the gas sensor under the modulated voltage; the software part comprises a data processing module and a gas recognition model building module, wherein the data processing module is used for receiving training response signals transmitted by the processor through the communication unit and carrying out preprocessing and characteristic extraction on the training response signals to obtain a gas training data set, and the gas recognition model building module is used for feeding the gas training data set obtained by the data processing module into a deep neural network to build a deep learning gas recognition model.
2. An intelligent gas identification device according to claim 1, wherein the voltage modulation unit comprises a digital-to-analog conversion subunit and an operational amplification circuit, the digital-to-analog conversion subunit is electrically connected to the processor and the operational amplification circuit, and the operational amplification circuit is electrically connected to the gas sensor.
3. The intelligent gas identification device of claim 2, wherein the data acquisition unit comprises a data acquisition chip and an analog-to-digital conversion subunit, the analog-to-digital conversion subunit is electrically connected with the data acquisition chip and the gas sensor, and the data acquisition chip is electrically connected with the processor.
4. The intelligent gas identification device according to claim 1, wherein the hardware portion of the intelligent gas identification device further comprises a power supply and a voltage regulation unit, the voltage regulation unit electrically connecting the power supply and the processor.
5. The intelligent gas identification device according to claim 4, wherein the voltage stabilization unit is a low dropout linear regulator.
6. The intelligent gas identification device of claim 1 wherein the hardware portion of the intelligent gas identification device further comprises a server communicatively coupled to the processor via the communication unit.
7. The smart gas identification device according to any one of claims 1 to 6, wherein the gas sensor is one of a ceramic wafer type gas sensor, a ceramic tube type gas sensor, or a MEMS sensor.
8. An intelligent gas identification method, characterized by using the intelligent gas identification device according to any one of claims 1 to 7 to perform the following steps:
placing at least one gas sensor in an environment containing a certain gas, and enabling the initial voltage at two ends of the gas sensor to be in a stable state;
enabling the processor to receive a starting modulation signal, triggering the voltage modulation unit to modulate the voltage of the gas sensor according to preset parameters, and simultaneously starting the data acquisition unit to acquire a training response signal of the gas sensor under the modulation voltage;
receiving a training response signal transmitted by a processor through a data processing module, and carrying out preprocessing and feature extraction on the training response signal to obtain a gas training data set;
feeding the gas training data set obtained by the data processing module into a deep neural network through a gas identification model building module to build a deep learning gas identification model;
the gas sensor is placed in an environment containing gas to be identified, the gas sensor is subjected to the same voltage modulation by the voltage modulation unit, meanwhile, response signals to be tested of the gas sensor under the modulated voltage are collected by the data acquisition unit, the data processing module is used for receiving the response signals to be tested transmitted by the processor and carrying out the same preprocessing and characteristic extraction on the response signals to be tested, a data set of the gas to be tested is obtained and sent to the gas identification model, and gas identification is achieved.
9. The intelligent gas identification method according to claim 8,
the trigger voltage modulation unit modulates the voltage of the gas sensor according to preset parameters, specifically, the trigger voltage modulation unit modulates the voltage of the gas sensor according to the preset parameters and by adopting one or a combination of square waves, step waves, sawtooth waves, triangular waves or sine waves;
the starting data acquisition unit acquires a training response signal of the gas sensor under the modulation voltage and the data acquisition unit acquires a response signal to be tested of the gas sensor under the modulation voltage, wherein the training response signal and the response signal to be tested are acquired repeatedly for 20-100 times under the same modulation voltage, modulation temperature range, sampling frequency and sampling duration;
the deep neural network is one of a deep convolutional neural network CNN, a deep back propagation neural network BPNN and a deep cyclic neural network RNN.
10. The intelligent gas identification method according to claim 8,
the method comprises the steps that a training response signal transmitted by a processor is received by a data processing module and is subjected to preprocessing and characteristic extraction, and a response signal to be tested transmitted by the processor is received by the data processing module and is subjected to preprocessing and characteristic extraction, wherein the preprocessing comprises sensitivity preprocessing, concentration normalization preprocessing and discrete wavelet transformation characteristic extraction processing on the training response signal and the response signal to be tested;
the sensitivity pretreatment is specifically represented by the formula G (cc, u, t) = Rair(t)/ Rgas(cc, u, t); the concentration normalization pretreatment is specifically represented by the formula Y (cc, u, t) = (G (cc, u, t) -G (cc, u, t)min)/ (G(cc,u,t)max- G(cc,u,t)min) Carrying out the following steps; the discrete wavelet transform feature extraction processing is specifically carried out through Daubechies wavelets; where t represents the sampling time, u represents the u-th test, cc represents a certain gas or a certain gas concentration, R represents the training response signal or the response signal to be identified, G represents the sensitivity of the signal, R represents the sensitivity of the signalair(t) represents the training response signal of the air at the sampling time t, Rgas(cc, u, t) represents a training response signal of the gas at the u-th test gas concentration cc at the sampling time t, Y (cc, u, t) represents a concentration normalization pre-processing expression, G (cc, u, t) represents the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t, G (cc, u, t)maxG (cc, u, t) which represents the maximum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time tminRepresents the minimum value of the sensitivity of the signal at the u-th test gas concentration cc at the sampling time t.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110826601.8A CN113267535A (en) | 2021-07-21 | 2021-07-21 | Intelligent gas identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110826601.8A CN113267535A (en) | 2021-07-21 | 2021-07-21 | Intelligent gas identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113267535A true CN113267535A (en) | 2021-08-17 |
Family
ID=77236992
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110826601.8A Pending CN113267535A (en) | 2021-07-21 | 2021-07-21 | Intelligent gas identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113267535A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113533654A (en) * | 2021-07-21 | 2021-10-22 | 深圳盛方科技有限公司 | Intelligent gas identification method, system, equipment and computer readable storage medium |
| CN114113491A (en) * | 2021-12-10 | 2022-03-01 | 中国科学技术大学 | Integrated machine olfaction chip calculated by utilizing convolutional neural network |
| CN114460226A (en) * | 2022-01-10 | 2022-05-10 | 上海工程技术大学 | Scene gas memory and recognition method and application thereof |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1611182A (en) * | 2003-10-29 | 2005-05-04 | 上海洁润丝纺织品工业有限公司 | Multifunctional nondestructive lung area ventilating and blood flow meausuring instrument |
| CN202362266U (en) * | 2011-11-22 | 2012-08-01 | 河南汉威电子股份有限公司 | Combustible gas detector |
| CN103492867A (en) * | 2011-01-31 | 2014-01-01 | 斯科特科技股份有限公司 | System and method for automatically adjusting gas sensor settings and parameters |
| CN203772835U (en) * | 2014-04-02 | 2014-08-13 | 太仓华风环保科技有限公司 | Constant-temperature probe |
| CN104820072A (en) * | 2015-04-30 | 2015-08-05 | 重庆大学 | Electronic nose air quality monitoring system and method based on cloud computing |
| CN205679593U (en) * | 2016-06-22 | 2016-11-09 | 临沂市安福电子有限公司 | A kind of imflammable gas detector based on bluetooth |
| CN108398533A (en) * | 2018-05-03 | 2018-08-14 | 北京物资学院 | Electric nasus system and its air source discriminating in storage and localization method |
| CN109709163A (en) * | 2019-01-17 | 2019-05-03 | 中国科学院合肥物质科学研究院 | P-type metal-oxide gas transducer and its preparation and application |
| CN110095508A (en) * | 2019-05-24 | 2019-08-06 | 西安交通大学 | The method and apparatus of gas identification is carried out based on single sensor |
| CN110309867A (en) * | 2019-06-21 | 2019-10-08 | 北京工商大学 | A kind of Mixed gas identification method based on convolutional neural networks |
| CN110869754A (en) * | 2017-05-05 | 2020-03-06 | 墨尔本皇家理工大学 | Multi-gas sensing system |
| CN111126575A (en) * | 2020-01-09 | 2020-05-08 | 同济大学 | Gas sensor array mixed gas detection method and device based on machine learning |
| CN111443165A (en) * | 2020-03-27 | 2020-07-24 | 华中科技大学 | Odor identification method based on gas sensor and deep learning |
| CN111879825A (en) * | 2020-06-24 | 2020-11-03 | 南方科技大学 | Gas-sensitive sensing device, system and gas detection method |
| CN112986332A (en) * | 2021-02-04 | 2021-06-18 | 安徽大学 | VOCs detection method and system based on dynamic temperature modulation |
| CN113030194A (en) * | 2021-02-03 | 2021-06-25 | 中国科学院合肥物质科学研究院 | Gas sensor array, three alcohol types and identification method of concentration of three alcohol types |
-
2021
- 2021-07-21 CN CN202110826601.8A patent/CN113267535A/en active Pending
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1611182A (en) * | 2003-10-29 | 2005-05-04 | 上海洁润丝纺织品工业有限公司 | Multifunctional nondestructive lung area ventilating and blood flow meausuring instrument |
| CN103492867A (en) * | 2011-01-31 | 2014-01-01 | 斯科特科技股份有限公司 | System and method for automatically adjusting gas sensor settings and parameters |
| CN202362266U (en) * | 2011-11-22 | 2012-08-01 | 河南汉威电子股份有限公司 | Combustible gas detector |
| CN203772835U (en) * | 2014-04-02 | 2014-08-13 | 太仓华风环保科技有限公司 | Constant-temperature probe |
| CN104820072A (en) * | 2015-04-30 | 2015-08-05 | 重庆大学 | Electronic nose air quality monitoring system and method based on cloud computing |
| CN205679593U (en) * | 2016-06-22 | 2016-11-09 | 临沂市安福电子有限公司 | A kind of imflammable gas detector based on bluetooth |
| CN110869754A (en) * | 2017-05-05 | 2020-03-06 | 墨尔本皇家理工大学 | Multi-gas sensing system |
| CN108398533A (en) * | 2018-05-03 | 2018-08-14 | 北京物资学院 | Electric nasus system and its air source discriminating in storage and localization method |
| CN109709163A (en) * | 2019-01-17 | 2019-05-03 | 中国科学院合肥物质科学研究院 | P-type metal-oxide gas transducer and its preparation and application |
| CN110095508A (en) * | 2019-05-24 | 2019-08-06 | 西安交通大学 | The method and apparatus of gas identification is carried out based on single sensor |
| CN110309867A (en) * | 2019-06-21 | 2019-10-08 | 北京工商大学 | A kind of Mixed gas identification method based on convolutional neural networks |
| CN111126575A (en) * | 2020-01-09 | 2020-05-08 | 同济大学 | Gas sensor array mixed gas detection method and device based on machine learning |
| CN111443165A (en) * | 2020-03-27 | 2020-07-24 | 华中科技大学 | Odor identification method based on gas sensor and deep learning |
| CN111879825A (en) * | 2020-06-24 | 2020-11-03 | 南方科技大学 | Gas-sensitive sensing device, system and gas detection method |
| CN113030194A (en) * | 2021-02-03 | 2021-06-25 | 中国科学院合肥物质科学研究院 | Gas sensor array, three alcohol types and identification method of concentration of three alcohol types |
| CN112986332A (en) * | 2021-02-04 | 2021-06-18 | 安徽大学 | VOCs detection method and system based on dynamic temperature modulation |
Non-Patent Citations (2)
| Title |
|---|
| 刘弘禹: "氧化物半导体气体传感器的热调制特征提取及VOCs分子识别研究", 《中国优秀博硕士论文全文数据库（博士）》 * |
| 胡庆 等: "《数字电路基础》", 31 January 2009, 电子科学技术出版社 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113533654A (en) * | 2021-07-21 | 2021-10-22 | 深圳盛方科技有限公司 | Intelligent gas identification method, system, equipment and computer readable storage medium |
| CN114113491A (en) * | 2021-12-10 | 2022-03-01 | 中国科学技术大学 | Integrated machine olfaction chip calculated by utilizing convolutional neural network |
| CN114113491B (en) * | 2021-12-10 | 2023-03-07 | 中国科学技术大学 | Integrated machine olfaction chip calculated by utilizing convolutional neural network |
| CN114460226A (en) * | 2022-01-10 | 2022-05-10 | 上海工程技术大学 | Scene gas memory and recognition method and application thereof |
| CN114460226B (en) * | 2022-01-10 | 2024-02-06 | 上海工程技术大学 | Scene gas memorizing and identifying method and application thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113267535A (en) | Intelligent gas identification method and device | |
| Konstantopoulos et al. | Converting a plant to a battery and wireless sensor with scatter radio and ultra-low cost | |
| CN106980264B (en) | The Dynamic Hysteresis modeling method of piezoelectric actuator neural network based | |
| Siregar et al. | Integrated pollution monitoring system for smart city | |
| CN113533654A (en) | Intelligent gas identification method, system, equipment and computer readable storage medium | |
| Mukhopadhyay | Intelligent sensing, instrumentation and measurements | |
| KR20190005820A (en) | Gas sensor controller | |
| Kabilan et al. | Surveillance and steering of irrigation system in cloud using Wireless Sensor Network and Wi-Fi module | |
| CN107817382B (en) | Intelligent electric meter, electric appliance identification method and intelligent apartment system with electric meter | |
| CN117407707A (en) | Odor identification model training method and device | |
| CN112286263A (en) | Crop regulation and control method, device and system and communication equipment | |
| CN113687609A (en) | Intelligent monitoring system and monitoring method for Internet of things applied to abnormal environment | |
| CN112202741A (en) | Gateway device based on small signal analysis and automatic identification communication bus and protocol | |
| CN103822662A (en) | Temperature-humidity illumination intelligent sensor for agriculture and forestry | |
| KR101332750B1 (en) | A pattern-based intelligence sesibility lighting system | |
| CN208636681U (en) | A kind of monitoring system of bio-incubator | |
| CN105759149A (en) | Electrical apparatus type judgment device for student dormitory | |
| CN105092485A (en) | Dangerous article detection method and device | |
| Ryman et al. | Temporal responses of chemically diverse sensor arrays for machine olfaction using artificial intelligence | |
| CN218446434U (en) | Fingerprint image construction device in power load prediction | |
| CN106618552B (en) | A kind of detection model acquisition methods and device | |
| CN107036717B (en) | Self-testing infrared thermopile temperature sensor and self-testing method | |
| US6487817B2 (en) | Electronic device to detect and direct biological microvariations in a living organism | |
| CN109374652A (en) | Humidity perception and detection method and system based on millimeter-wave signal | |
| CN115392338A (en) | Information processing method and device and readable storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210817 |
|
| RJ01 | Rejection of invention patent application after publication |