CN103512920A - Intelligent electronic nose system based device and method for analyzing quality of lemon tea beverage - Google Patents

Abstract

The invention relates to a lemon tea quality analysis device and method based on an intelligent electronic nose system. The method solves the existing technical problems of subjective differences and harm to human body in detecting volatile gas of beverage samples by human body's own organs. The device includes a gas collection unit, a gas collection unit, a processing unit and a control unit. The gas collection unit includes a gas chamber, a sample chamber, a first communication pipe and a second communication pipe, which together form a circular circulation path. The gas collection unit is connected to the gas chamber, the processing unit is connected to the gas collection unit, and the air intake mechanism, gas outlet solenoid valve, and collection unit are respectively connected to the control unit. The advantages of the present invention are: the electronic nose system is constructed, so that the detection results are more comprehensive and objective, and the harm to human health caused by the sample gas is avoided; when the gas is collected, the gas circulates to make the gas mixing more uniform, so that the detection data is more accurate.

Description

A quality analysis device and method for lemon tea beverage based on intelligent electronic nose system

technical field

The invention relates to a technology for detecting beverage quality, in particular to a lemon tea beverage quality analysis device based on an intelligent electronic nose system, and a lemon tea beverage quality analysis method based on an intelligent electronic nose system.

Background technique

Beverages are people's daily consumer goods. The aroma and taste of beverages have a great influence on consumers in the process of purchasing and drinking. If the beverage taste is not suitable for consumers or is unstable, it will actually directly affect the market sales of the product and relate to the benefit of the manufacturer. Therefore, manufacturers will conduct sensory evaluation and identification of beverages during the development and production of beverages.

For a long time, people have judged the texture of beverages through their own senses, and this judgment is often highly subjective, and the results of evaluation and analysis often vary with age, experience, etc., and there are considerable individual differences. Even the same person can get different results due to physical conditions and emotional changes. Moreover, olfactory identification is a process of inhalation of volatile substances. Long-term experiments will cause harm to human health, and some bad smells will make the judges particularly sensitive and cause errors in the results; The process of forming an evaluation team is relatively cumbersome, and the evaluation results are often not repeatable. Therefore, the demand for new analytical techniques is becoming increasingly urgent.

The Chinese invention application with the publication number CN101769889A discloses an electronic nose system for quality detection of agricultural products. Its structure includes a gas enrichment module that mainly completes the collection of low-concentration odors, and a box that mainly converts odor signals into electrical signals. Gas circuit module and sensor array, a sensor conditioning circuit domain data preprocessing module that mainly performs filtering, analog-to-digital conversion, and feature extraction on the output signal of the sensor array, an embedded system that identifies and judges a pair of signals, and has data storage , a display and result output module; the gas enrichment module is composed of an adsorption tube filled with adsorbent, a heating wire and a temperature control device. This invention can also collect gas for identification, but there are still shortcomings in this invention: first, the function is relatively single, and it cannot identify samples other than agricultural products; but the sensor has randomness in the sample collection method, which affects the test results; A method for the system to process the data collected by the sensor to obtain accurate results is proposed.

Contents of the invention

The present invention mainly solves the existing technical problems of detecting volatile gases of beverage samples by the human body's own organs, which have subjective differences and is harmful to the human body, and provides a lemon tea beverage quality analysis device and method based on an intelligent electronic nose system .

The above-mentioned technical problems of the present invention are mainly solved by the following technical solutions: a lemon tea beverage quality analysis device based on an intelligent electronic nose system, comprising a gas collection unit, a gas collection unit, a processing unit and a control unit, the set The gas unit includes a gas chamber, a sample chamber, a first communication pipe and a second communication pipe, both the gas chamber and the sample chamber have an inlet and an outlet, the first communication pipe is connected between the gas chamber outlet and the sample chamber inlet, the The second communication pipe is connected between the inlet of the gas chamber and the outlet of the sample chamber, so that the gas chamber and the sample chamber form a circular circulation passage. The air port and the air outlet are provided with an air outlet electromagnetic valve, the collection unit is connected to the air chamber, and the gas is collected in the air chamber, the processing unit is connected with the gas collection unit, and the air intake mechanism, the air outlet electromagnetic valve, and the collection unit are respectively connected to the control unit connected, they are controlled by the control unit to work. The gas collection unit in the present invention preprocesses the collected gas, circulates the gas to make the gas mix more uniform, and dilutes the gas according to the detection data, so that the gas detection data collected after processing is more accurate. The sample is set in the sample chamber, the carrier gas is passed through the gas chamber, passes through the first connecting pipe and the second connecting pipe, so that the carrier gas circulates between the gas chamber and the sample chamber, and drives the gas emitted by the sample to circulate together. Set the gas outlet to discharge gas for cleaning the gas collection unit. The gas sampling unit samples the sample gas, outputs the signal to the processing unit, and the processing unit analyzes and processes the signal to analyze the quality of the sample. The control unit controls the work of the air intake mechanism, the air outlet solenoid valve, the collection unit and other actuators, so that the steps of gas collection and gas extraction are completed.

As a preferred solution, the sample chamber includes a pull-out seat body, a groove for placing samples is arranged on the seat body, a cavity is left on the upper part of the seat body, the outlet and inlet of the sample chamber are arranged on the upper part of the sample chamber, and the sample chamber The outlet of the chamber is located above the tank body. In this scheme, the sample chamber adopts a drawer type, and the base body can be pulled out, and the sample is placed on the base body, so that it is convenient to put the sample into the sample chamber. There is a cavity on the upper part of the base body for gas circulation. The gas enters through the inlet, flows in the cavity, and takes away the gas emitted by the sample, and the gas enters and exits again.

As a preferred solution, a first air pump is also provided on the second communication pipe, and the first air pump is connected to the control unit. The first gas pump is used to drive gas.

As a preferred solution, the air intake mechanism includes an air inlet, a first electromagnetic valve is arranged on the air inlet, and the first electromagnetic valve is connected with the control unit. The first electromagnetic valve controls the on-off of the air inlet, that is, controls whether the carrier gas is passed in or not. The control unit controls the action of the first electromagnetic valve.

As a preferred solution, the gas collection unit includes a gas collection suction pipe and a sensor array composed of several sensors, each sensor is arranged in an independent chamber, one end of the gas collection suction pipe is connected to the gas chamber, and the gas collection suction pipe The port is provided with a fifth solenoid valve, and the gas collection suction pipe is provided with a second air pump. The other end of the gas collection suction pipe is respectively connected to the chambers of each sensor, and each sensor is respectively connected to the processing unit. A cleaning pipeline is also connected, and a sixth electromagnetic valve and a third air pump are arranged on the cleaning pipeline.

The gas collection unit is connected to the gas chamber and collects gas from the gas chamber. The control unit can control the gas collection unit to start or stop gas collection by controlling the fifth solenoid valve. The collected gas enters the chamber of each sensor through the suction pipe, and the sensor detects the gas and sends the detection data to the processing unit. The cleaning pipeline is used to feed clean air, and a third air pump pumps air into each sensor chamber to clean the sensors. There are 8 sensors, which are sulfide gas sensor, hydrogen sensor, ammonia sensor, nitrogen oxide sensor, hydrocarbon gas sensor, ethanol sensor, benzene sensor and alkanes sensor.

As a preferred solution, the sample chamber is provided with a stirring air outlet mechanism, the stirring air outlet mechanism includes a rotating shaft, the rotating shaft is hollow, the rotating shaft communicates with the inlet of the sample chamber, a stirring tube is connected to the rotating shaft, and a rotating shaft is arranged in the middle of the stirring tube The stirring tube is installed on the rotating shaft through the rotating shaft seat to form a T-shaped structure. The stirring tube is a hollow sealed tube, and the stirring tube is connected with the rotating shaft. The other end of the tube is provided with several second air holes on the side opposite to the first air holes. The stirring and gas outlet mechanism is not in the sample solution, and can be rotated under the condition of ventilation to stir the solution sample to make the sample evenly mixed, and the carrier gas is discharged from the stirring gas outlet mechanism, which can be evenly mixed with the volatile gas of the sample, so that the detection more precise. The first air hole and the second air hole at the two ends of the stirring tube face opposite directions respectively, so that the stirring tube can be automatically driven to rotate around the rotating shaft after the gas is introduced.

A kind of lemon tea beverage quality analysis method based on intelligent electronic nose system, comprises the following steps:

Step 1: Set the temperature of the experimental environment at 15-25°C and the humidity at 45%-55%, clean the sensor array of the gas extraction unit, pass the clean air into the chamber of each sensor, and run for 8-12 minutes, so that each sensor in the initial state;

Step 2: Pre-treat the gas. Take 20ml of the lemon tea sample to be tested and pour it into the sample chamber. First, clean the gas collection unit. The inert gas circulates in the gas collection unit with the carrier gas for 20-30 minutes;

Step 3: Collect the sample gas with the gas sampling pipe, discharge the gas into the chamber of each sensor, and control the sensors to detect the gas in the chamber by the control unit. The detection time is 40-60s, and each sensor will detect the information sent to the processing unit;

Step 4: The processing unit processes the information to obtain the response curve of each sensor, and samples 30 points on each response curve, and uses the data obtained by sampling each curve as the input data Input (t), and calculates it by using the nonlinear stochastic resonance model Signal-to-noise ratio SNR, the nonlinear stochastic resonance model algorithm is as follows:

The stochastic resonance system contains three factors: the bistable system, the input signal and the external noise source, and the characteristics of the system are described by an overdamped Brownian motion particle driven by a periodic force in the bistable potential well,

V(x) is a nonlinear symmetric potential function, ξ(t) is Gaussian white noise, its autocorrelation function is: E[ξ(t)ξ(0)]=2Dδ(t), a is the input signal strength, f ₀ is the modulation signal frequency, D is the noise intensity, a and b are real parameters,

$\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; ax</span>}}^{\mathrm{<span\; class="notranslate">\; 4</span>}}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 4</span>}}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="bx"\; filenames="HDA00003606742000011.png"\; state="\{\{state\}\}">bx</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

So the above formula can be changed to:

The signal-to-noise ratio is obtained as:

$\mathrm{<span\; class="notranslate">\; SNR</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; [</span>\underset{\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}<span\; class="notranslate">\; \&Right\; Arrow;</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\mathrm{<span\; class="notranslate">\; lim</span>}}\underset{\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}}{<span\; class="notranslate">\; \&Integral;</span>}}\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d\&omega;</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; )</span>$

S(ω) is the signal spectral density, S _N (Ω) is the noise intensity in the signal frequency range;

Take the peak value of the signal-to-noise ratio curve as the signal-to-noise ratio characteristic value;

Step 5: Bring the input variables into a nonlinear state-space model

In the formula:

σ is the input variable, that is, the signal-to-noise ratio eigenvalue, ε is the intermediate transfer parameter, τ is the initial phase, is the output variable, κ is the real parameter, η is the real parameter, Γ is the real correction parameter,

Then define the residual variables:

为系统实际输出、为系统理论输出，

is the actual output of the system, For the system theory output,

Redefine the classification standard model:

$\mathrm{<span\; class="notranslate">\; \&Delta;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; L</span>}}\underset{\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; )</span>$

则可以判断被测样品是该阈值所属类型，得到该被测样品品质信息，如果

则需要重新进行类型判断。In the formula, L is the data length, and Δ will be compared with each threshold Thr in the preset threshold library. If there is

Then it can be judged that the tested sample belongs to the type of the threshold, and the quality information of the tested sample can be obtained, if

You need to redo the type judgment.

As a preferred solution, each threshold value of the threshold library is obtained in advance, and the process is: obtain each type of sample in advance, then use the analysis device to detect each type of sample, input the detection data into the stochastic resonance model for analysis, and obtain the signal-to-noise Then measure each type of sample multiple times, take the average value of the signal-to-noise ratio eigenvalue obtained multiple times for this type of sample as the threshold Thr for judging this type of sample, and the threshold values of various samples together constitute the threshold value library.

As a preferred solution, the sampling data in step 4 are subjected to normalization processing before calculating the noise ratio. The processing process is as follows: the data sampled by each sensor response curve is regarded as a set of detection data, and each group of detection data The sampling value in the data is substituted into the formula y=log ₁₀ (x) for calculation, and x is the sampling value before normalization processing.

As a preferred solution, abnormal data processing is performed before normalizing each group of detection data. The steps are: the sampling value Input(t) in each group of detection data, which is denoted as W here, conforms to the normal distribution: W～N(μ,σ ² ), μ is the mean value of the sampled value W in each group of data, σ is the standard deviation of the sampled value W in each group of data, after derivation:

P(|W-μ|＞3σ)≤2-2Φ(3)＝0.003

Substitute the mean μ, standard deviation σ, and each sampled value W of each group of data into the formula |W-μ|>3σ, and remove the sampled value W that satisfies the formula |W-μ|>3σ as abnormal data.

Therefore, the advantages of the present invention are: 1. An electronic nose system is built, and the sample gas is detected by the system, so that the detection results are more comprehensive and objective, and at the same time, the sample gas is prevented from causing harm to human health; An electronic nose composed of various types of sensors, each sensor is set in an independent chamber to detect samples, avoiding mutual interference caused by multiple sensors co-located in one box, improving detection accuracy, fast, and good repeatability; 3. Acquisition When the gas is used, the gas circulates and flows, which makes the gas mix more uniform and makes the detection data more accurate.

Description of drawings

Accompanying drawing 1 is a kind of structural representation of gas collection unit among the present invention;

Accompanying drawing 2 is another kind of structural representation of gas collection unit among the present invention;

Accompanying drawing 3 is a kind of structural representation of gas extraction unit among the present invention;

Accompanying drawing 4 is a kind of framework diagram that control unit of the present invention is connected with sensor, air pump;

Accompanying drawing 5 is a kind of structure schematic diagram of agitating air-out mechanism in the present invention.

1-Air collection unit 2-Air chamber 3-Sample chamber 4-First connecting pipe 5-Second connecting pipe 6-Air outlet 7-Outlet solenoid valve 8-Air inlet 9-Filtered air inlet 10-Inert Gas inlet 11-gas suction pipe 12-first solenoid valve 16-fifth solenoid valve 17-sixth solenoid valve 18-first air pump 19-second air pump 20-seat body 21-tank body 22-cavity 23 -Sensor 24-Chamber 25-Third Air Pump 26-Processing Unit 27-Control Unit 28-Gas Collection Unit 29-Rotating Shaft 30-Rotating Shaft Seat 31-Stirring Tube 32-First Air Hole 33-Second Air Hole

Detailed ways

The technical solutions of the present invention will be further specifically described below through the embodiments and in conjunction with the accompanying drawings.

Example:

In this embodiment, a lemon tea beverage quality analysis device based on an intelligent electronic nose system, as shown in Figure 1 and Figure 2, includes a gas collection unit 1, a gas collection unit 28, a processing unit 26 for processing detection data, and a control execution operation The control unit 27.

The gas collection unit 1 includes four parts: an air chamber 2, a sample chamber 3, a first communication pipe 4 and a second communication pipe 4. The air chamber and the sample chamber all have an inlet and an outlet, and the first communication pipe is connected between the outlet of the air chamber and the second communication pipe 4. Between the inlets of the sample chambers, the second communication pipe is connected between the outlets of the sample chambers and the inlets of the gas chambers, which constitutes a loop-shaped circulation channel structure. Samples are placed in the sample chamber. In this embodiment, the lemon tea beverage sample is poured into the sample chamber. The inlet of the sample chamber is arranged at the bottom. A one-way valve is set on the pipeline, and the outlet of the sample chamber is set on the top. An air outlet 6 is also opened on the first communication pipeline, an air outlet electromagnetic valve 7 for controlling opening and closing is arranged on the air outlet, and a first air pump 18 for driving gas flow is also arranged on the first communication pipeline. The gas chamber is provided with an air intake mechanism for feeding carrier gas. In this embodiment, taking air carrier gas as an example, the air inlet mechanism includes an air inlet 8, which is connected to the air chamber. , the air inlet is provided with a first solenoid valve 12 to control opening and closing.

In the sample chamber 3, a stirring and air-exiting mechanism is arranged, which is submerged in the sample. As shown in Figure 6, the stirring and air-exiting mechanism includes a rotating shaft 29, which is hollow, and the rotating shaft communicates with the inlet of the sample chamber, and a stirring tube is connected to the rotating shaft. 31. There is a rotating shaft seat in the middle of the stirring tube. The stirring tube is installed on the rotating shaft through the rotating shaft seat to form a T-shaped structure. The stirring tube is a hollow sealed tube, and the stirring tube is connected with the rotating shaft. On the side of the end of the stirring tube Several first air holes 32 are arranged on the top, and several second air holes 33 are arranged on the side opposite to the first air holes at the other end of the stirring tube. When the carrier gas is introduced, the carrier gas enters the rotating shaft through the inlet of the sample chamber, enters the stirring tube through the rotating shaft, and then is discharged from the first air hole and the second air hole at opposite ends, so that the stirring tube rotates around the rotating shaft.

The gas collection unit 28 is connected with the gas collection unit 1, and the gas collection unit includes a gas collection suction pipe 11 and a sensor array composed of 8 sensors 23, one end of the gas collection suction pipe is connected to the gas chamber 2, and the gas collection suction pipe The end is provided with the fifth solenoid valve for controlling opening and closing and the second air pump 19 for driving suction. The eight sensors here are sulfide gas sensor, hydrogen sensor, ammonia sensor, nitrogen oxide sensor, hydrocarbon component gas sensor, ethanol sensor, benzene sensor and alkanes sensor, and each sensor is set in an independent chamber. In the chamber 24, the other end of the gas sampling suction pipe is respectively connected to the independent chambers of the sensors.

A cleaning mechanism is also provided on the gas extraction unit for cleaning the sensor. The cleaning mechanism includes a cleaning pipeline, which is connected to the independent chamber of each sensor, and a sixth solenoid valve 17 for controlling opening and closing and a third air pump 25 for driving gas are arranged on the cleaning pipeline.

The processing unit 26 processes the data detected by each sensor. As shown in FIG. 4 , each sensor is connected to the processing unit.

As shown in Fig. 5, each sensor of the gas extraction unit is also controlled and connected to the control unit, and the control unit controls the operation of the sensors. In addition, the first electromagnetic valve, the fifth electromagnetic valve, the sixth electromagnetic valve, the first air pump, the second air pump, the third air pump and the air outlet electromagnetic valve are all controlled and connected to the control unit, and the control unit controls their work to complete the mining process. gas process.

As shown in Figure 2, another structure of the gas collection unit is also provided. Here, the gas collection unit adopts a drawer structure. The sample chamber includes a drawable seat 20, and a groove for placing samples is arranged on the seat. 21. The operator can pull out the seat, put in the sample and then push it into the seat. A cavity 22 for other communication is reserved on the top of the seat body and the sample chamber, the outlet and the inlet of the sample chamber are all arranged on the top, and the outlet of the sample chamber is located above the groove body.

The analysis method of the lemon tea beverage quality analysis device based on the intelligent electronic nose system of the present embodiment is as follows: comprise the following steps,

step one:

Set the experimental environment temperature to 20°C and humidity to 50%, to clean the sensor array of the gas extraction unit, that is, to open the sixth battery valve when the fifth solenoid valve is closed, and pass clean air into each sensor through the third air pump In the chamber, run for 10 minutes, clean each sensor so that each sensor is in the initial state.

Step two:

Put 20ml of lemon tea samples in the sampling chamber, and then clean the gas collection unit. Under the condition of only using a kind of carrier gas such as air, open the first electromagnetic valve and the air outlet electromagnetic valve of the air inlet, so that The gas collection unit forms an exhaust pipeline, and the air is introduced until the gas in the original gas collection unit is exhausted and filled with carrier gas, then the gas outlet battery valve is closed, so that the gas collection unit forms a circular circulation path, which is driven by the first air pump The volatile gas generated by the sample circulates in the gas collection unit for 20 minutes along with the carrier gas.

Step three:

The gas sampling unit starts to collect gas. At this time, the fifth electromagnetic valve of the gas sampling pipe is opened, and the sample gas is sucked into the gas sampling pipe and passed into the independent chamber of each sensor. The control unit controls the work of each sensor, and the gas in the chamber is Detection, the detection time is 50s, and each sensor sends the detected data to the processing unit.

Step 4: Each sensor detects the response value, and takes time and response intensity as the coordinate axis to obtain the response curve of each sensor. Eight sensors obtain 8 response curves, and then sample 30 points on each response curve at equal intervals to obtain 240 The data of each point is used as a set of input data Input(t).

The abnormal data processing is performed on the sampled data, and the data sampled by each sensor response curve is regarded as a set of detection data, and the sampling value Input(t) in each set of detection data, here denoted as W, conforms to the normal distribution: W～N (μ,σ ² ), μ is the mean value of the sampled value W in each group of data, σ is the standard deviation of the sampled value W in each group of data, after derivation:

P(|W-μ|＞3σ)≤2-2Φ(3)＝0.003

Substitute the mean μ, standard deviation σ, and each sampled value W of each group of data into the formula |W-μ|>3σ, and remove the sampled value W that satisfies the formula |W-μ|>3σ as abnormal data.

The sampled data is firstly normalized, and the sampled values in each group of detected data are substituted into the formula y=log ₁₀ (x) for calculation, where x is the sampled value before the normalized process.

Step 5: Substituting the sampling data W into the nonlinear stochastic resonance model to calculate the signal-to-noise ratio SNR, the algorithm of the nonlinear stochastic resonance model is:

The stochastic resonance system contains three factors: the bistable system, the input signal and the external noise source, and the characteristics of the system are described by an overdamped Brownian motion particle driven by a periodic force in the bistable potential well,

V(x) is a nonlinear symmetric potential function, ξ(t) is Gaussian white noise, its autocorrelation function is: E[ξ(t)ξ(0)]=2Dδ(t), a is the input signal strength, f ₀ is the modulation signal frequency, D is the noise intensity, a and b are real parameters,

$\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 8</span>}}{\mathrm{<span\; class="notranslate">\; ax</span>}}^{\mathrm{<span\; class="notranslate">\; 4</span>}}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 4</span>}}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="bx"\; filenames="HDA00003606742000011.png"\; state="\{\{state\}\}">bx</figure-callout></span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}$

So the above formula can be changed to:

The signal-to-noise ratio is obtained as:

$\mathrm{<span\; class="notranslate">\; SNR</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; [</span>\underset{\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}<span\; class="notranslate">\; \&Right\; Arrow;</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\mathrm{<span\; class="notranslate">\; lim</span>}}\underset{\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}}{\overset{\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; \&Delta;\&omega;</span>}}{<span\; class="notranslate">\; \&Integral;</span>}}\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&omega;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d\&omega;</span>}<span\; class="notranslate">\; ]</span><span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; S</span>}}_{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&Omega;</span>}<span\; class="notranslate">\; )</span>$

S(ω) is the signal spectral density, S _N (Ω) is the noise intensity in the signal frequency range;

The input data Input (t) is substituted into the formula to calculate the signal-to-noise ratio SNR. The signal-to-noise ratio SNR is a curve, and the peak value of the signal-to-noise ratio curve is taken as the signal-to-noise ratio characteristic value.

Step five:

Bringing the signal-to-noise ratio eigenvalues, i.e., input variables, into a nonlinear state-space model

In the formula:

为输出变量、κ为实参数、η为实参数、Γ为实矫正参数，σ is the input variable, that is, the signal-to-noise ratio eigenvalue, ε is the intermediate transfer parameter, τ is the initial phase,

is the output variable, κ is the real parameter, η is the real parameter, Γ is the real correction parameter,

Then define the residual variables:

为系统实际输出、为系统理论输出，

is the actual output of the system, For the system theory output,

Redefine the classification standard model:

$\mathrm{<span\; class="notranslate">\; \&Delta;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; L</span>}}\underset{\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&psi;</span>}<span\; class="notranslate">\; )</span>$

则需要重新进行类型判断。In the formula, L is the data length, and Δ will be compared with each threshold Thr in the preset threshold library. If there is Then it can be judged that the tested sample belongs to the type of the threshold, and the quality information of the tested sample can be obtained, if

You need to redo the type judgment.

Wherein, each threshold value Thr of the threshold value library mentioned above is obtained in advance, and the process is: obtain each type of sample in advance, such as the lemon tea sample from the first day to the eighth day, and then use an analysis device to detect each type of sample, and the detected The data is input into the stochastic resonance model for analysis to obtain the characteristic value of the signal-to-noise ratio, and multiple measurements are performed on each type of sample. The average value of the signal-to-noise ratio eigenvalues of this type of sample obtained multiple times is used as the threshold Thr for judging this type of sample, and then the threshold value of each type of sample is obtained, and the threshold values of each type of sample together constitute the threshold value library.

The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which the present invention belongs can make various modifications or supplements to the described specific embodiments or adopt similar methods to replace them, but they will not deviate from the spirit of the present invention or go beyond the definition of the appended claims range.

Although terms such as gas collection unit, gas chamber, sample chamber, first communicating pipe, and second communicating pipe are frequently used herein, the possibility of using other terms is not excluded. These terms are used only for the purpose of describing and explaining the essence of the present invention more conveniently; interpreting them as any kind of additional limitation is against the spirit of the present invention.

Claims (10)

1. the lemon tea beverage attributional analysis device based on smart electronics nasus system, it is characterized in that: comprise gas collection unit (1), gas production unit (28), processing unit (26) and control module (27), described gas collection unit comprises air chamber (2), sample chamber (3), the first communicating pipe (4) and the second communicating pipe (5), air chamber and sample chamber all have import and outlet, be connected to described the first communicating pipe between air chamber outlet and sample chamber entrance, be connected to described the second communicating pipe between air chamber entrance and sample chamber outlet, make air chamber and sample chamber form the circulation path of a hollow, in described gas compartment, be provided with admission gear, on the second communicating pipe, be provided with gas outlet (6), on gas outlet, be provided with the solenoid valve of giving vent to anger (7), described gas production unit is connected to air chamber, by gathering gas in air chamber, processing unit is connected with gas production unit, described admission gear, the solenoid valve of giving vent to anger, collecting unit is connected with control module respectively, by control module, control them and carry out work.

2. a kind of lemon tea beverage attributional analysis device based on smart electronics nasus system according to claim 1, it is characterized in that described sample chamber (3) comprises drawing and pulling type pedestal (20), on pedestal, be provided with the cell body (21) of placing sample, on the top of pedestal, leave cavity (22), sample chamber outlet and entrance are arranged on the outlet of ，Qie sample chamber, top, sample chamber and are positioned at cell body top position.

3. a kind of lemon tea beverage attributional analysis device based on smart electronics nasus system according to claim 1 and 2, is characterized in that being also provided with the first air pump (18) on the second communicating pipe, and described the first air pump is connected on control module (27).

4. a kind of lemon tea beverage attributional analysis device based on smart electronics nasus system according to claim 3, it is characterized in that described admission gear comprises on air inlet (8), air inlet is provided with the first solenoid valve (12), and the first solenoid valve (12) is connected with control module (27).

5. a kind of lemon tea beverage attributional analysis device based on smart electronics nasus system according to claim 4, it is characterized in that described gas production unit (28) comprises gas production suction pipe (11) and the sensor array consisting of some sensors (23), each sensor setting is one independently in chamber (24), described gas production suction pipe one end is connected on air chamber, on gas production suction pipe port, be provided with the 5th solenoid valve, on gas production suction pipe, be provided with the second air pump (19), the gas production suction pipe other end is connected respectively on the chamber of each sensor, each sensor is connected on processing unit (26), on the chamber of each sensor, be also connected with detergent line, in detergent line, be provided with the 6th solenoid valve (17) and the 3rd air pump (25), described sensor (23) has 8, be respectively sulfide gas sensor, hydrogen gas sensor, ammonia gas sensor, NOx sensor, charcoal hydrogen component gas sensor, ethanol sensor, benzene class sensor and alkanes sensor.

6. a kind of lemon tea beverage attributional analysis device based on smart electronics nasus system according to claim 1, it is characterized in that being provided with stirring air-out mechanism in described sample chamber (3), stir air-out mechanism and comprise rotating shaft (29), rotating shaft is hollow, rotating shaft is communicated with sample chamber import, in rotating shaft, be connected with stirring pipe (31), the centre position of stirring pipe is provided with shaft seat, stirring pipe is arranged in rotating shaft by shaft seat, form a T-shaped structure, stirring pipe is hollow seal pipe, stirring pipe is communicated with rotating shaft, in a side of the termination of stirring pipe, be provided with some the first pores (32), in another termination of stirring pipe and the opposing side of the first pore, be provided with some the second pores (33).

7. the lemon tea beverage Quality Analysis Methods based on smart electronics nasus system, the device that adopts claim 1-6 any one to describe, is characterized in that: comprise the following steps:

Step 1: 15～25 ℃ of experimental situation temperature are set, and humidity is 45%-55%, and the sensor array of gas production unit is cleaned, is passed into pure air in the chamber of each sensor, and operation 8-12min, makes each sensor in original state;

Step 2: gas is carried out to pre-service, lemon tea sample to be measured is got to 20ml, pour in sample chamber, first gas collection unit is cleaned, after cleaning, by air intake opening, pass into carrier gas, the escaping gas that drives sample to produce by the air pump 20-30min that circulates with carrier gas in gas collection unit;

Step 3: by gas production unit collected specimens gas, gas is drained in the chamber of each sensor in gas production unit, by control module, control each sensor the gas in chamber is detected, be 40-60s detection time, and each sensor sends to processing unit by the information detecting;

Step 4: processing unit is processed the response curve that obtains each sensor to information, and at 30 points of each response curve up-sampling, the data that each curve sampling is obtained are as input data I nput(t), substitution non-linear stochastic resonance model calculates signal to noise ratio snr, and this non-linear stochastic resonance model algorithm is as follows:

Stochastic resonance system comprises three factors: bistable system, and descriptive system feature carried out by power-actuated overdamping Brownian movement of cycle particle with one in input signal and external noise source in bistable state potential well,

V (x) is non-linear symmetric potential function, and ξ (t) is white Gaussian noise, and its auto-correlation connection function is: E[ξ (t) ξ (0)]=2D δ (t), a is input signal strength, f ₀be frequency modulating signal, D is noise intensity, and a, b are all real parameters,

$V\left(x\right)=\frac{1}{8}{\mathrm{ax}}^4-\frac{1}{4}{\mathrm{bx}}^2$

Therefore above formula can change into:

Obtaining signal to noise ratio (S/N ratio) is:

$\mathrm{SNR}=2\left[\underset{\mathrm{\&Delta;\&omega;}\&RightArrow;0}{\lim }\underset{\mathrm{\&Omega;}-\mathrm{\&Delta;\&omega;}}{\overset{\mathrm{\&Omega;}+\mathrm{\&Delta;\&omega;}}{\&Integral;}}S\left(\mathrm{\&omega;}\right)\mathrm{d\&omega;}\right]/S_N\left(\mathrm{\&Omega;}\right)$

S (ω) is signal spectral density, S _n(Ω) be the noise intensity in signal frequency range;

Get this signal to noise ratio (S/N ratio) peak of curve as signal to noise ratio (S/N ratio) eigenwert;

Step 5: bring input variable into a kind of Nonlinear state space model

In formula:

σ is input variable, signal to noise ratio (S/N ratio) eigenwert, ε be intermediate transfer parameter, τ be initial phase,

for output variable, κ are that real parameter, η are that real parameter, Γ are the real parameter of correcting,

Then define residual error variable:

for the actual output of system,

for Systems Theory output,

Defining classification master pattern again:

$\mathrm{\&Delta;}=\frac{1}{L}\underset{\mathrm{\&psi;}=N-L+1}{\overset{N}{\mathrm{\&Sigma;}}}{e}^T\left(\mathrm{\&psi;}\right)e\left(\mathrm{\&psi;}\right)$

In formula, L is data length, and just Δ is compared with each threshold value Thr in predefined threshold library, if had

can judge that sample is type under this threshold value, obtain this sample quality information, if need to re-start type judgement.

8. a kind of lemon tea beverage Quality Analysis Methods based on smart electronics nasus system according to claim 7, it is characterized in that each threshold value of described threshold library is for obtaining in advance, its process is: obtain in advance every class sample, then use analytical equipment to detect every class sample, detecting data input accidental resonance model, analyze, obtain signal to noise ratio (S/N ratio) eigenwert, again every class sample is taken multiple measurements, get the mean value of the signal to noise ratio (S/N ratio) eigenwert that such sample repeatedly obtains as the threshold value Thr of such sample of judgement, the threshold value of all kinds of samples has formed threshold library jointly.

9. according to a kind of lemon tea beverage Quality Analysis Methods based on smart electronics nasus system described in claim 7 or 8, before it is characterized in that the sampled data in step 4 is calculated to to-noise ratio, be first normalized, processing procedure is: the data that each sensor response curve is sampled are as one group of detection data, by every group of sampled value substitution formula y=log detecting in data ₁₀(x) calculate, x is the sampled value before normalized.

10. a kind of lemon tea beverage Quality Analysis Methods based on smart electronics nasus system according to claim 9, before it is characterized in that every group of detection data to be normalized, first carry out dealing of abnormal data, the steps include: every group of sampled value Input(t detecting in data), here be designated as W, meet normal distribution: W～N (μ, σ ²), μ is the average of sampled value W in every group of data, σ is the standard deviation of sampled value W in every group of data, through deriving, has:

P(|W-μ|＞3σ)≤2-2Φ(3)＝0.003

By the average μ of every group of data, standard deviation sigma and each sampled value W substitution formula | W-μ | > 3 σ, will meet formula | W-μ | the sampled value W of > 3 σ removes as abnormal data.

Images