CN117169441A - Method for detecting concentration of disinfectant in cold-chain environment based on electronic nose - Google Patents
Method for detecting concentration of disinfectant in cold-chain environment based on electronic nose Download PDFInfo
- Publication number
- CN117169441A CN117169441A CN202311135021.XA CN202311135021A CN117169441A CN 117169441 A CN117169441 A CN 117169441A CN 202311135021 A CN202311135021 A CN 202311135021A CN 117169441 A CN117169441 A CN 117169441A
- Authority
- CN
- China
- Prior art keywords
- disinfectant
- data
- svr
- cold
- electronic nose
- 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
- 239000000645 desinfectant Substances 0.000 title claims abstract description 63
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000004044 response Effects 0.000 claims abstract description 36
- 241000282461 Canis lupus Species 0.000 claims abstract description 32
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 31
- 238000007637 random forest analysis Methods 0.000 claims abstract description 23
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000000513 principal component analysis Methods 0.000 claims abstract description 7
- 238000012216 screening Methods 0.000 claims abstract description 6
- 241000282421 Canidae Species 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 238000001514 detection method Methods 0.000 abstract description 8
- 239000007789 gas Substances 0.000 description 25
- MHAJPDPJQMAIIY-UHFFFAOYSA-N Hydrogen peroxide Chemical compound OO MHAJPDPJQMAIIY-UHFFFAOYSA-N 0.000 description 10
- KFSLWBXXFJQRDL-UHFFFAOYSA-N Peracetic acid Chemical compound CC(=O)OO KFSLWBXXFJQRDL-UHFFFAOYSA-N 0.000 description 10
- 238000012360 testing method Methods 0.000 description 10
- 238000003066 decision tree Methods 0.000 description 7
- 238000011002 quantification Methods 0.000 description 6
- ZAMOUSCENKQFHK-UHFFFAOYSA-N Chlorine atom Chemical compound [Cl] ZAMOUSCENKQFHK-UHFFFAOYSA-N 0.000 description 5
- 239000000460 chlorine Substances 0.000 description 5
- 229910052801 chlorine Inorganic materials 0.000 description 5
- 150000003242 quaternary ammonium salts Chemical class 0.000 description 5
- 238000003491 array Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 239000003153 chemical reaction reagent Substances 0.000 description 2
- 238000004140 cleaning Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 241000894006 Bacteria Species 0.000 description 1
- 241000233866 Fungi Species 0.000 description 1
- 241000700605 Viruses Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000001030 gas--liquid chromatography Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 230000007794 irritation Effects 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 235000021485 packed food Nutrition 0.000 description 1
- 239000003755 preservative agent Substances 0.000 description 1
- 230000002335 preservative effect Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000002798 spectrophotometry method Methods 0.000 description 1
- 238000005507 spraying Methods 0.000 description 1
- 238000004659 sterilization and disinfection Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
- 235000013311 vegetables Nutrition 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Landscapes
- Investigating Or Analyzing Materials By The Use Of Fluid Adsorption Or Reactions (AREA)
Abstract
The invention relates to the technical field of gas sensing detection, and particularly discloses a method for detecting disinfectant concentration in a cold-chain environment based on an electronic nose. Simulating a cold chain environment, and preparing disinfectant samples with different concentrations; carrying a gas sensor array on the electronic nose equipment, and acquiring data of the prepared disinfectant sample to obtain sensor response data; performing data preprocessing on the sensor response data to obtain a gas response characteristic data set; the method comprises the steps of screening characteristic data and reducing dimension by adopting a random forest and principal component analysis algorithm, constructing an SVR model, introducing a wolf optimization algorithm to optimize parameters of the SVR model, constructing an RF-PCA-GWO-SVR model, and predicting the concentration of the disinfectant by adopting the RF-PCA-GWO-SVR model, so that the prediction precision, stability and generalization capability of a support vector regression algorithm are effectively improved.
Description
Technical Field
The invention relates to the technical field of gas sensing detection, and particularly discloses a method for detecting disinfectant concentration in a cold-chain environment based on an electronic nose.
Background
Chemical disinfectants play an important role in medical diagnosis, food processing, pharmacy, cleaning and disinfection, and the like, and mainly utilize chemical substances to kill or remove bacteria, viruses, fungi and other microorganisms. The disinfectant is easy to dissolve in water and volatilize, has strong corrosiveness and irritation, and brings challenges to detection and research of the disinfectant.
In recent years, the technologies of spectrophotometry, gas/liquid chromatography and the like show unique performance in the aspect of disinfectant detection, and the detection sensitivity is extremely high, but the technology has the advantages of higher use cost, complicated machine operation, single use scene, poor portability and difficult wide application in the actual scene of the cold chain market. Although the gas sensor method is inferior to the above technology in terms of accuracy and sensitivity, the gas sensor method is simple to operate, low in price, small in size and relatively easy to maintain in the later period, so that the gas sensor method has great value in practical application in many fields. Aiming at the actual demands of the cold-chain logistics environment such as cost, power consumption and the like, the gas sensor is used for detecting the concentration of each disinfectant gas in the cold-chain logistics.
There are many methods for quantifying the concentration of disinfectant or various volatile gases, and the obtained quantification performance is different after different quantification methods are input into different gas data sets. Each quantization algorithm has its own unique aspects, and the performance of the algorithm is analyzed at different angles to find an algorithm model suitable for the disinfectant gas, and the following are summarized as the advantages and disadvantages of RF (random forest), BP neural network (BPNN, back Propagation Neural Network) and SVR (support vector regression) algorithms.
(1) When the RF method is used for quantification, the method has the advantages that the method can effectively process high-dimensional data and nonlinear problems, and the generalization capability of the model can be improved through methods such as feature random sampling, subtree random selection and the like, but when the decision trees in the RF are more, a great amount of time and memory are consumed for training the model;
(2) The performance of the model can be improved by properly adjusting the weight and the bias through the quantitative analysis of the BPNN, but the model is easy to be fitted and the generalization capability is reduced due to the random setting of parameters;
(3) When the SVR algorithm is used for quantification, the optimal solution of the algorithm can be obtained by using a small amount of samples, but the training performance of a large sample or a data set with higher dimension can be poor.
In summary, under the conditions that the complex disinfectant gas characteristics are faced, and the number of samples is not very large, the SVR algorithm is selected to realize quantification of the disinfectant concentration, because compared with BPNN and RF, the parameters of the SVR model are relatively small, the quantification result is relatively easy to interpret and understand, verification and optimization in practical application are facilitated, the training speed of the SVR model is relatively high, and training and optimization of the model can be completed in a relatively short time. However, the key parameters in SVR can cause large errors in the model when empirically assigned.
Disclosure of Invention
In order to overcome the problems in the prior art, the invention provides a method for detecting the concentration of a disinfectant in a cold-chain environment based on an electronic nose.
The technical scheme for solving the technical problems is as follows:
the invention provides a method for detecting the concentration of a disinfectant in a cold-chain environment based on an electronic nose, which comprises the following steps:
simulating a cold chain environment, and preparing disinfectant samples with different concentrations;
carrying a gas sensor array on the electronic nose equipment, and acquiring data of the prepared disinfectant sample to obtain sensor response data;
performing data preprocessing on the sensor response data to obtain a gas response characteristic data set;
and (3) carrying out characteristic data screening and dimension reduction by adopting a random forest and principal component analysis algorithm, constructing an SVR model, introducing a gray wolf optimization algorithm to optimize parameters of the SVR model, constructing an RF-PCA-GWO-SVR model, and predicting the concentration of the disinfectant by adopting the RF-PCA-GWO-SVR model.
Further, the gas sensor array includes more than 2 gas sensors.
Further, the sensor response dataset is subjected to data preprocessing including filtering, denoising, de-baselining, feature extraction and normalization.
Further, the feature extraction includes: and extracting a relative steady-state average value, an integral value, a global average value and a standard deviation of sensor response data from the sensor response data after the baseline removal, wherein the relative steady-state average value is an average value in a steady-state interval when the gas sensor reaches the maximum response, and the integral value, the average value and the standard deviation are integral values, global average values and standard deviations of all data of each sample data curve.
Further, the characteristic data screening and dimension reduction are carried out by adopting a random forest and principal component analysis algorithm, and the method specifically comprises the following steps:
calculating the importance of features based on random forests, calculating the importance of each sensor, selecting the sensors meeting the requirements according to the importance of the sensors, selecting the sensors with high importance to construct a new sensor array and a feature matrix, adopting PCA to reduce the dimension, and constructing an optimal feature subset.
Further, a gray wolf optimization algorithm is introduced to optimize parameters of the SVR model, and the method specifically comprises the following steps:
step1: initializing SVR and GWO algorithm parameters;
setting the number N of the gray wolf population, the iteration times M, a search range A (A epsilon-a, a) and random variables C, C epsilon 0,2, and setting the optimizing range of SVR parameters (C, g), wherein C is a penalty coefficient of SVR, and g is the kernel width of SVR;
step2: initializing the position of a wolf population, training an SVR model, and calculating the adaptability of individual wolves;
step3: selecting 3 gray wolf individuals with optimal fitness and respectively named as alpha, beta and delta;
step4: updating the position and parameters a, A and c of each wolf, recalculating individual fitness of the wolves, and reserving an optimal fitness value;
step5: if the maximum iteration number is reached, the optimization is terminated, the position coordinates of alpha wolf, namely the optimal value of SVR parameters (C, g) are output, otherwise, the iteration is continued by returning to Step 2.
Further, the calculation formulas of a and c are as follows:
A＝2ar 1 -a
c＝2r 2
wherein r is 1 And r 2 Take [0,1 ]]Random numbers in between.
Compared with the prior art, the invention has the following technical effects:
(1) The method establishes the support vector regression model, then introduces a principal component analysis and random forest and gray wolf optimization algorithm to optimize input data and model parameters, and effectively improves the prediction precision, stability and generalization capability of the support vector regression algorithm;
(2) The method can realize the function of detecting the concentration of the disinfectant under the condition of limited artificial sense according to the actual application scene, has accurate and reliable result and wide applicability, and can provide a new idea for quantifying the concentration of the disinfectant in the cold chain environment.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of the invention for detecting disinfectant concentration in a cold-chain environment based on an electronic nose;
FIG. 2 is a flow chart of sensor data preprocessing in accordance with the present invention;
FIG. 3 is a graph showing the comparison of predicted values and actual values of the RF-PCA-GWO-SVR model for the peroxyacetic acid disinfectant;
FIG. 4 is a graph of predicted versus actual values of the RF-PCA-GWO-SVR model for hydrogen peroxide sterilant;
FIG. 5 is a graph of predicted versus actual values of the RF-PCA-GWO-SVR model for chlorine-containing disinfectants;
FIG. 6 is a graph of the predicted versus actual values of the RF-PCA-GWO-SVR model for quaternary ammonium salt disinfectant.
Detailed Description
In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention with reference to the accompanying drawings and preferred embodiments. The particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
Referring to fig. 1-6, the embodiment of the invention discloses a method for detecting disinfectant concentration in a cold-chain environment based on an electronic nose, which comprises the following steps:
step 1. The cold chain environment was simulated in a laboratory with the temperature controlled at 3 ℃ and the humidity controlled at 85% rh (Relative Humidity ).
The temperature range of the packaged food and vegetables and fruits transported by the cold chain is 0-7 ℃ and the humidity range is 85-90% RH during transportation and delivery, so that the cold chain environment with the temperature of 3 ℃ and the humidity of 85% RH is studied for transportation of common articles of people.
A high-low temperature damp-heat test box with the model of BPHS-060A is adopted for auxiliary test, the test box can control the temperature to be in the range of-20-150 ℃, the fluctuation is +/-0.5 ℃, the humidity is in the range of 30-95% RH, the fluctuation is +/-2% RH, and the size of a working chamber of the test box is 420 multiplied by 480mm.
Step2, preparing samples of the disinfectant of the peroxyacetic acid, the hydrogen peroxide, the chlorine and the quaternary ammonium salt with different concentrations, and mixing the samples according to 15ml/m 3 Is filled into 1500ml reagent bottles.
Samples of peracetic acid, hydrogen peroxide, chlorine-containing and quaternary ammonium salt disinfectants were prepared at various concentrations and were prepared at 15ml/m 3 Is filled into a reagent bottle of 1500 ml; the specific concentration of disinfectant measured: the concentration of the peroxyacetic acid disinfectant comprises 0.2%, 0.3%, 0.4% and 0.6%; the concentration of the hydrogen peroxide disinfectant comprises 2%, 3%, 4% and 6%; the chlorine-containing disinfectant comprises 125mg/L, 250mg/L, 500mg/L, 1000mg/L, 2000mg/L and 4000mg/L; the concentration of the quaternary ammonium salt disinfectant comprises 250mg/L, 500mg/L, 1000mg/L, 2000mg/L, 4000mg/L and 6000mg/L.
And 3, carrying a gas sensor array on the electronic nose equipment, and acquiring data of the prepared disinfectant sample to obtain sensor response data.
Selecting more than 2 gas sensor built-up arrays to be carried on the electronic nose equipment, combining a temperature sensor and a humidity sensor, carrying out data acquisition on the prepared disinfectant samples, acquiring more than 2 sensor response values for each disinfectant sample, and acquiring the sensor response value of each disinfectant sample to form a sensor response data set, namely, the sensor response data set comprises a response combination set of the gas sensor arrays to the disinfectant sample concentrations with different concentrations.
In a specific embodiment, 10 TGS series sensor arrays are selected and mounted on the electronic nose device, and the selected 10 TGS sensor arrays are respectively TGS2600, TGS2601, TGS2602, TGS2609, TGS2610, TGS2611, TGS2612, TGS2615, TGS2620, and TGS2630.
The specific steps of carrying out data acquisition on the disinfectant sample by using the electronic nose carrying the gas sensor array include: preheating a gas sensor for 72 hours, spraying a disinfectant sample into a headspace bottle, and placing the headspace bottle into a test chamber to generate a headspace for 20 minutes; the air inlet pipe of the detection device is inserted into the glass container through the preservative film, and after the disinfectant sample gas is extracted for 3min, air is introduced for 5min for gas washing; after all disinfectant samples are collected, ending the experiment and recording data; during testing, the test period of each disinfectant sample is 480s, wherein the detection time is 180s, the cleaning time is 300s, and the sampling frequency is 1Hz.
And 4, carrying out data preprocessing on the sensor response data to obtain a gas response characteristic data set.
The preprocessing includes filtering, denoising, baseline removal, feature extraction and normalization.
In this embodiment, this step specifically includes the following sub-steps:
step 4.1: the gas response dataset was smoothed and filtered using a Savitzky-Golay (S-G) smoothing filter.
Savitzky-Golay filtering is a digital signal processing technology, and the core idea is to perform k-order polynomial fitting on data points in a window with a certain length so as to obtain a fitted result. In a specific embodiment, the polynomial degree selects 2 degrees, and the number of window points is 50.
Step 4.2: subtracting the baseline voltage from the sensor response value to obtain response data after baseline removal.
For a sensor array consisting of 10 sensors, each disinfectant sample contains 10 sensor response values, and the average value of the last 5s of response voltage of the sensor is takenAs a baseline voltage, each response value in the current disinfectant sample is subtracted by the baseline voltage to obtain response data after baseline removal.
Step 4.3: and extracting a relative steady-state average value, an integral value, a global average value and a standard deviation of each disinfectant sample from the sensor response data after the baseline removal as original characteristic values, wherein the original characteristic values form an original characteristic matrix, and the original characteristic matrix forms a characteristic data set. The relative steady-state average value is an average value in a section of steady-state interval when the gas sensor reaches the maximum response, 161 th to 170 th second data of samples in the experiment are relatively stable, and 161 th to 170 th second data of each sample are selected to obtain an average value; the integral value, the global average value and the standard deviation are the integral value (accumulated value), the average value and the standard deviation of all data of each sample data curve, and the integral value, the average value and the standard deviation are taken as sample characteristics according to the data of 1 st to 480 seconds of each sample in the experiment.
Each disinfectant sample contains 10 sensor response values, so that a single disinfectant sample can yield a 4 x 10 feature matrix.
Step 4.4: and carrying out normalization processing on the characteristic data set to eliminate the influence of dimension.
The normalization formula is:
wherein,normalized sensor eigenvalues, x is the raw eigenvalue, x max Represents the maximum eigenvalue, x min Representing the minimum eigenvalue.
Step 4.5: 3/4 of the feature data set after normalization is used as a training set, and 1/4 is used as a test set.
And 5, carrying out characteristic data screening and dimension reduction by adopting a random forest and principal component analysis algorithm, constructing an SVR model, introducing a gray wolf optimization algorithm to optimize parameters of the SVR model, constructing an RF-PCA-GWO-SVR model, and predicting the concentration of the disinfectant by adopting the RF-PCA-GWO-SVR model.
Step 5.1: calculating the importance of the features based on the random forest, calculating the importance of each sensor, and selecting the sensors meeting the requirements according to the importance of the sensors; and selecting a sensor with high importance to construct a new feature matrix, adopting PCA to reduce the dimension, and constructing an input feature set of the SVR model of the optimal feature subset.
Random forests are a commonly used classification and regression algorithm, which is an ensemble learning algorithm that improves the accuracy of classification and regression by combining multiple decision trees. In random forests, there are many parameters that need to be adjusted, including the number of decision trees, the maximum depth of the tree, the minimum number of leaves, etc.
Determining that the number of decision trees in the random forest is 100, the minimum number of leaves (nodes) is 5, carrying out importance analysis on the normalized feature matrix based on the random forest, selecting the Gini importance of the random forest, and calculating the importance of each sensor.
Assume that a certain characteristic of a certain sensor is denoted as x t The importance of the random forest with M decision trees is calculated by the following characteristic importance calculation process:
assuming that the detection target has k categories, the proportion of the category i in the node m is P mi The Gini index Gini (m) for that node m is then expressed as:
assume thatFeature x t Z occurrences in branches of the nth decision tree, feature x t Importance VIM in nth tree nt The method comprises the following steps:
VIM in tz Gini (m) -Gini (o) -Gini (p), gini (o) and Gini (p) represent Gini indices of two new nodes after node m branches.
If there are M decision trees in RF, then x t Importance in RF can be expressed as:
the importance calculation is performed on the extracted characteristic values of the sensor responses, and the importance of all the characteristics of a certain sensor is summed to be the importance of the sensor.
And selecting a sensor with high importance to construct a new sensor array and a new feature matrix. In this embodiment, the importance of the sensors is sorted from large to small into TGS2602, TGS2615, TGS2612, TGS2603, TGS2609, TGS2610, TGS2600, TGS2611, TGS2630, TGS2620, and the sensors TGS2602, TGS2615, TGS2612, TGS2603, and TGS2609 are further selected to form an array. And adopting PCA to reduce the newly built feature matrix to 4 dimensions, and constructing an optimal feature subset as an input feature set of the SVR model.
Step 5.2: and GWO algorithm is adopted to select two parameters of penalty coefficient C and kernel width g of the SVR model.
The key parameter values of the SVR model comprise penalty factors and kernel widths g, the key parameters in the SVR are subjected to parameter optimization based on GWO, and the process comprises the following steps:
step1: initializing SVR and GWO algorithm parameters;
setting the number of gray wolf populations N=20, the iteration number M=300, gradually decreasing the value of a convergence factor a from 2 to 0 along with the increase of the iteration number, setting the optimizing range of SVR parameters (C, g) as [0.01,1000] in a searching range A (A epsilon-a, a ]) and a random variable C, C epsilon [0,2], wherein C is a penalty coefficient of SVR and g is the kernel width of SVR;
the formulas for A and c are as follows:
A＝2ar 1 -a
c＝2r 2
wherein r is 1 And r 2 Take [0,1 ]]Random numbers in between.
Step2: initializing the positions of the wolf population, wherein the position X of each wolf corresponds to the value of SVR parameters (C, g), training an SVR model, and calculating the adaptability of the individual wolves;
step3: selecting 3 gray wolf individuals with optimal fitness and respectively named as alpha, beta and delta, and respectively using X as the position vector α 、X β 、X δ Other wolf individuals are designated ω; update iterations are performed based on Step1-Step 3.
At the time of iteration, three wolves (alpha, beta, delta) with optimal population are reserved, and the positions are continuously updated by combining data, wherein the expression is as follows:
D α ＝c 1 ·X α -X
D β ＝c 2 ·X β -X
D δ ＝c 3 ·X δ -X
X 1 ＝X α -A 1 ·D α
X 2 ＝X β -A 2 ·D β
X 3 ＝X δ -A 3 ·D δ
wherein: x is X α For expressing the position vector of a in the population, X β Position vector for expressing beta in population, X δ A position vector for expressing delta within the population; x is used for expressing the position vector of the wolf, X 1 ，X 2 ，X 3 Respectively, are affected by the alpha layer wolf group, the beta layer wolf group and the delta layer wolf group, other gray wolvesThe position of the omega of the individual to be adjusted is finally determined by taking an average value; d (D) α 、D β 、D δ Position information for expressing candidate wolves; c 1 、c 2 、c 3 、A 1 、A 2 、A 3 Is a random vector, and ensures the randomness of the algorithm; when |A| > 1, the wolves in the wolf clusters can be scattered for searching, otherwise, the wolves can intensively search for the hunting object in a specific area.
Step4: updating the position and parameters a, A and c of each wolf, recalculating individual fitness of the wolves, and reserving an optimal fitness value;
step5: if the maximum iteration number is reached, the optimization is terminated, the position coordinates of alpha wolf, namely the optimal value of SVR parameters (C, g) are output, otherwise, the iteration is continued by returning to Step 2.
It should be noted that, from the initial stage of iteration to the end of iteration, the values of the parameters a, a and c are all random, which represents the randomness of the gray-wolf optimization algorithm and can prevent the gray-wolf optimization algorithm from sinking into the local optimum phenomenon to a certain extent.
The above calculations show that, for the peracetic acid, hydrogen peroxide, chlorine-containing and quaternary ammonium salt disinfectant data, the optimal values of the key parameters C of the SVR obtained in this example are 12.01, 1.622, 63.45, 14.65, and the optimal values of g are 3.04, 7.113, 1.27, 0.28, respectively, so that the SVR model of this example is constructed.
Step 5.3: and constructing an RF-PCA-GWO-SVR model by utilizing the optimal feature subset and the parameter values, testing by utilizing the testing set, and outputting the disinfectant concentration. The final prediction results are compared with the actual values in fig. 3, 4, 5 and 6.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.
Claims (7)
1. The method for detecting the concentration of the disinfectant in the cold-chain environment based on the electronic nose is characterized by comprising the following steps of:
simulating a cold chain environment, and preparing disinfectant samples with different concentrations;
carrying a gas sensor array on the electronic nose equipment, and acquiring data of the prepared disinfectant sample to obtain sensor response data;
performing data preprocessing on the sensor response data to obtain a gas response characteristic data set;
and (3) carrying out characteristic data screening and dimension reduction by adopting a random forest and principal component analysis algorithm, constructing an SVR model, introducing a gray wolf optimization algorithm to optimize parameters of the SVR model, constructing an RF-PCA-GWO-SVR model, and predicting the concentration of the disinfectant by adopting the RF-PCA-GWO-SVR model.
2. The method for detecting disinfectant concentration in a cold-chain environment based on an electronic nose according to claim 1, wherein the gas sensor array comprises more than 2 gas sensors.
3. A method of detecting disinfectant concentrations in cold chain environments based on electronic nose as claimed in claim 1, wherein the data preprocessing is performed on sensor response data sets, including filtering, denoising, de-baselining, feature extraction and normalization.
4. A method of detecting disinfectant concentrations in cold chain environments based on an electronic nose as claimed in claim 3, wherein the feature extraction includes, but is not limited to, the following four features: and extracting a relative steady-state average value, an integral value, a global average value and a standard deviation of sensor response data from the sensor response data after the baseline removal, wherein the relative steady-state average value is an average value in a steady-state interval when the gas sensor reaches the maximum response, and the integral value, the global average value and the standard deviation are integral values, average values and standard deviations of all data of each sample data curve.
5. The method for detecting the concentration of the disinfectant in the cold-chain environment based on the electronic nose according to claim 1, wherein the method is characterized in that a random forest and principal component analysis algorithm is adopted for screening and dimension reduction of characteristic data, and specifically comprises the following steps:
calculating the importance of the features based on the random forest, calculating the importance of each sensor, and selecting the sensors meeting the requirements according to the importance of the sensors; and selecting a sensor with high importance to construct a new sensor array and a feature matrix, and adopting PCA to reduce the dimension to construct an optimal feature subset.
6. The method for detecting the concentration of the disinfectant in the cold-chain environment based on the electronic nose according to claim 1, wherein the parameter of the SVR model is optimized by introducing a gray wolf optimization algorithm, and the method specifically comprises the following steps:
step1: initializing SVR and GWO algorithm parameters;
setting the number N of the gray wolf population, the iteration times M, a convergence factor a, a search range A (A E < -a > a < >) and a random variable C, setting the optimizing range of SVR parameters (C, g), wherein C is a punishment coefficient, and g is a kernel width;
step2: initializing the positions of the wolf population, wherein each position of the wolf population corresponds to the value of SVR parameters (C, g), training an SVR model, and calculating the adaptability of the individual wolves;
step3: selecting 3 gray wolf individuals with optimal fitness and respectively named as alpha, beta and delta;
step4: updating the position and parameters a, A and c of each wolf, recalculating individual fitness of the wolves, and reserving an optimal fitness value;
step5: if the maximum iteration number is reached, the optimization is terminated, the position coordinates of alpha wolf, namely the optimal value of SVR parameters (C, g) are output, otherwise, the iteration is continued by returning to Step 2.
7. The method for detecting disinfectant concentration in cold-chain environment based on electronic nose according to claim 6, wherein the calculation formulas of a and c are as follows:
A＝2ar 1 -a
c＝2r 2
wherein r is 1 And r 2 Take [0,1 ]]Random numbers in between.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311135021.XA CN117169441A (en) | 2023-09-05 | 2023-09-05 | Method for detecting concentration of disinfectant in cold-chain environment based on electronic nose |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311135021.XA CN117169441A (en) | 2023-09-05 | 2023-09-05 | Method for detecting concentration of disinfectant in cold-chain environment based on electronic nose |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117169441A true CN117169441A (en) | 2023-12-05 |
Family
ID=88939048
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311135021.XA Pending CN117169441A (en) | 2023-09-05 | 2023-09-05 | Method for detecting concentration of disinfectant in cold-chain environment based on electronic nose |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117169441A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109799269A (en) * | 2019-01-24 | 2019-05-24 | 山东工商学院 | Electronic nose gas sensor array optimization method based on behavioral characteristics different degree |
| CN112381290A (en) * | 2020-11-13 | 2021-02-19 | 辽宁工程技术大学 | Method for predicting coal gas content by random forest and gray wolf optimization |
| CN113022270A (en) * | 2021-02-04 | 2021-06-25 | 海伟环境科技有限公司 | Cold chain car vehicle-mounted ozone disinfection system |
| CN116415208A (en) * | 2023-04-10 | 2023-07-11 | 上海大学 | Dehumidification system humidity prediction method and system based on improved SVR |
-
2023
- 2023-09-05 CN CN202311135021.XA patent/CN117169441A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109799269A (en) * | 2019-01-24 | 2019-05-24 | 山东工商学院 | Electronic nose gas sensor array optimization method based on behavioral characteristics different degree |
| CN112381290A (en) * | 2020-11-13 | 2021-02-19 | 辽宁工程技术大学 | Method for predicting coal gas content by random forest and gray wolf optimization |
| CN113022270A (en) * | 2021-02-04 | 2021-06-25 | 海伟环境科技有限公司 | Cold chain car vehicle-mounted ozone disinfection system |
| CN116415208A (en) * | 2023-04-10 | 2023-07-11 | 上海大学 | Dehumidification system humidity prediction method and system based on improved SVR |
Non-Patent Citations (1)
| Title |
|---|
| ZHANG GUISHUAI，ET AL,: "Construction of an Electronic Nose for Disinfectant Concentration Detection in Cold Chain Environment", 《2022 IEEE SENSORS》, pages 1 - 4 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Discrimination of green tea quality using the electronic nose technique and the human panel test, comparison of linear and nonlinear classification tools | |
| Qiu et al. | Single-kernel FT-NIR spectroscopy for detecting supersweet corn (Zea mays L. saccharata sturt) seed viability with multivariate data analysis | |
| Jiang et al. | Quantitative detection of fatty acid value during storage of wheat flour based on a portable near-infrared (NIR) spectroscopy system | |
| Su et al. | Application of hyperspectral imaging for maturity and soluble solids content determination of strawberry with deep learning approaches | |
| CN106644983B (en) | Spectral wavelength selection method based on PLS-VIP-ACO algorithm | |
| Sexton et al. | A theoretical and real world evaluation of two Bayesian techniques for the calibration of variety parameters in a sugarcane crop model | |
| Sonobe et al. | Hyperspectral reflectance sensing for quantifying leaf chlorophyll content in wasabi leaves using spectral pre-processing techniques and machine learning algorithms | |
| Wu et al. | Nondestructive measurement of internal quality attributes of apple fruit by using NIR spectroscopy | |
| Ni et al. | Prediction of CO concentration in different conditions based on Gaussian-TCN | |
| Adiani et al. | Development of rapid method to assess microbial quality of minimally processed pomegranate arils using FTIR | |
| Chen et al. | Prediction of drought-induced components and evaluation of drought damage of tea plants based on hyperspectral imaging | |
| Mohammed et al. | Development and validation of innovative machine learning models for predicting date palm mite infestation on fruits | |
| Wu et al. | Rapid nondestructive detection of peanut varieties and peanut mildew based on hyperspectral imaging and stacked machine learning models | |
| Wang et al. | A comparison of multiple deep learning methods for predicting soil organic carbon in Southern Xinjiang, China | |
| Wang et al. | A detection method of two carbamate pesticides residues on tomatoes utilizing excitation-emission matrix fluorescence technique | |
| Zhao et al. | An apple fungal infection detection model based on BPNN optimized by sparrow search algorithm | |
| Wei et al. | Monitoring the freshness of pork during storage via near-infrared spectroscopy based on colorimetric sensor array coupled with efficient multivariable calibration | |
| Ye et al. | A new red wine prediction framework using machine learning | |
| Zhou et al. | Tomato storage quality predicting method based on portable electronic nose system combined with WOA-SVM model | |
| Shi et al. | Nondestructive detection of Panax notoginseng saponins by using hyperspectral imaging | |
| Li et al. | HSI combined with CNN model detection of heavy metal Cu stress levels in apple rootstocks | |
| CN117972433A (en) | Training method of mushroom room temperature prediction model, mushroom room temperature prediction method and device | |
| CN112630180B (en) | Ultraviolet/visible light absorption spectrum model for detecting concentration of organophosphorus pesticide in water body | |
| Yin et al. | Spoilage Monitoring and Early Warning for Apples in Storage Using Gas Sensors and Chemometrics | |
| Xiao et al. | Rapid detection of maize seed germination rate based on Gaussian process regression with selection kernel function |
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 |