CN110567909A - A method for detecting sex pheromone content in trap chip - Google Patents

Abstract

The invention provides a method for detecting the content of sex pheromone in a trap chip, comprising: (1) extracting a standard trap chip with an organic solvent to obtain a standard solution, and diluting the standard solution into a plurality of gradient solutions of gradient concentrations; 2) Utilize the near-infrared spectrometer to collect the spectra of the standard solution and multiple gradient solutions; (3) based on the semi-supervised least squares support vector machine regression algorithm (QPSO‑LSS3VR), the sample spectra obtained in step (2) are compared with the corresponding Concentration values are fitted, and a quantitative detection model for the change of sex pheromone content in the sex pheromone trap is established; (4) the model obtained in step (3) is used for quantitative analysis of the sex pheromone content in the trap chip to be tested. The method is simple and fast.

Description

A method for detecting sex pheromone content in trap chip

technical field

The invention relates to the field of spectrum detection, in particular to a method for detecting the content of sex pheromones in trap chips.

Background technique

Tobacco bin pests refer to a variety of pests that harm tobacco leaves and tobacco products in tobacco bins and tobacco storage places. In my country, there are more than 30 types of tobacco bin pests, among which the most harmful are tobacco beetle and tobacco powder borer. In some areas, the big grain robbers have caused serious damage. Tobacco bin pests cause great losses to tobacco leaves during storage period, ranging from lowering the grade, to causing the tobacco leaves to be completely unusable. Worldwide losses due to infestation by tobacco beetles and tobacco meal borers are estimated to be approximately 1% per year. The direct insect damage rate of stored tobacco leaves in my country is 1.64% per year.

At present, the prevention and control methods of tobacco storage pests are to use pheromone traps to monitor the distribution and density of tobacco beetles and tobacco powder borers, to do a good job in cleaning and sanitation, to use insecticides and insect growth regulators rationally, and to fumigate with phosphine . It can be seen that the stability of the sex pheromone content of the pheromone trap is very important. If the sex pheromone content of the sex pheromone trap is low, the sex pheromone trap will fail early, and the insect situation will not be reflected faithfully.

Contents of the invention

The present disclosure provides a method for detecting the content of sex pheromone in a trap chip.

In some aspects, a method of detecting pheromone content in a trap chip is provided, comprising:

(1) Extract the standard trap chip with an organic solvent to obtain a standard solution, and dilute the standard solution into multiple gradient solutions of gradient concentrations;

(2) Utilize near-infrared spectrometer to collect the spectra of standard solution and multiple gradient solutions;

(3) Based on the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR), the sample spectrum obtained in step (2) is fitted with the corresponding ratio concentration value, and the sex pheromone content in the sex pheromone trap is established Change quantitative detection model;

(4) utilize the model that step (3) obtains to quantitatively analyze the sex pheromone content in the trap chip to be tested;

The organic solvent is composed of acetone and NN-dimethylformamide, and the volume ratio of acetone and NN-dimethylformamide is 18.5˜19.5:1.

Preferably, the volume ratio of acetone and NN-dimethylformamide is 19:1.

The specific organic solvent formulation, that is, the specific volume ratio of acetone and NN-dimethylformamide is very critical for obtaining accurate detection of the sex pheromone content in the trap chip. Deviating from the above formula cannot obtain accurate detection results.

In some embodiments, between steps (2) and (3), the step of preprocessing the spectrum is also included, and the method of preprocessing is selected from: first derivative, second derivative, vector normalization, multivariate signal correction and One or more of spectral smoothing.

In some embodiments, between steps (2) and (3), a step of preprocessing the spectrum is also included, and the preprocessing method includes one or more of the following:

- Using multiple signal correction (MSC) to eliminate the difference caused by sample inhomogeneity;

-Using first-order differential processing to eliminate the influence of baseline drift and obtain higher resolution and clearer spectral profile changes than the original spectrum; and

-Adopt the Savitzky-Golay filter with a segment length of 9 and an interval of 5 to smooth the spectrum to eliminate high-frequency noise and retain useful low-frequency information.

In some embodiments, the method for establishing the quantitative detection model of sex pheromone content change in the sex pheromone trap in step (3) comprises the following steps:

input spectral data;

Set the initial parameters, execute the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR), and estimate the unlabeled samples;

The semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) was implemented to obtain the optimal model parameters and establish a spectral quantitative analysis model.

In some embodiments, between steps i and ii, a cross-validation method is used, and about 10% of the samples are reserved as test samples; after the model is established, the test samples are input into the model for detection, and the performance of the model is evaluated.

In some embodiments, step (3) includes the following operations,

Establish a quantitative model: add 190 samples with different acetone ratios, scan the spectrum, preprocess the spectral data, select the best spectral segment in the wavenumber range of 8634 to 4102 cm ^-1 , and use the cross-validation method to reserve 20 samples as test samples;

Set the initial parameters: the number of iterations M=100, the number of individuals in the initialization group N=25, p=0.6, the search range is: α=[0,100], γ[0,1000], λ=[0,1000];

Execute the QPSO-LSS3VR algorithm to estimate unlabeled samples. The model detection performance is evaluated by using the detection root mean square error RMSEC and the coefficient of determination R ² ; after calculation, α = [0, 3.9], γ [0, 118], λ = [0, 42.5] is optimal;

Among them, n is the sample size, is the predicted value of the test sample, and y _i is the measured value of the test sample.

The coefficient of determination (R ² ) was used as the standard for evaluating the correlation between the measured value and the predicted value of the calibration set data (the closer the value is to 1, the better the correlation is). A model with good predictive performance has an R ² value closer to 1 and a lower RMSEC value.

In some embodiments, the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) is designed as follows:

a) Use TQ software to reduce the dimension of the sample spectrum, and map the high-dimensional data to the low-dimensional space, which is convenient for distance measurement;

b) Calculate the distance d(n _i , m _j ) between the sample n _i in the unmarked sample set N and the sample m _j in the marked sample set M.

c) According to the KNN algorithm, solve the set M' of k marked neighbors of each unlabeled sample n _i .

d) Take the average value of all marked samples in M', and estimate the initial estimated value of the unmarked sample n _i : T _n =T _l ×(1+rand(0,1))

Among them, _Tl represents the average value of the marked samples in the set.

e) For the unlabeled samples selected for training, replace the labeled value with the detection value currently given by the model; for the unlabeled samples that are not selected for the next iteration, keep the value unchanged.

In some embodiments, step (1) includes, select 200 qualified trap chips just produced, add 2000ml of organic solvent (consisting of acetone and NN-dimethylformamide, acetone and NN-dimethylformamide The volume ratio is 19:1), after soaking and extracting for 24 hours, the solution is poured out and stirred evenly, and divided into 200 parts on average, 10 parts of which are used as the reference liquid, and a certain amount of acetone is added to the remaining 190 parts to prepare a mixture containing 70 parts. For the prepared solution of ~99.9% by volume of the reference solution, set the value of the reference solution as 100, and the value of the proportioning solution shall be the corresponding value based on the percentage containing the reference solution.

In some embodiments, step (2) includes, using the sample prepared above, using the liquid transmission sampling module of the near-infrared spectrometer to collect the sample spectrum, and the collected wavenumber ranges from 10000 to 3800 cm ^-1 , with the built-in background of the instrument as a reference, the sample and For reference, 70 scans were used with a resolution of 8 cm ^-1 .

In some aspects, a method of determining the effectiveness of a trap chip is provided, comprising the steps of:

Extracting the trap chip to be tested with an organic solvent to obtain a solution to be tested;

Use the method described in any one of the present disclosure to detect the content of sex pheromone in the solution to be tested to obtain the detection content;

Comparing the detected content with the preset threshold, when the detected content is less than the preset threshold, it is determined that the trap chip is invalid, and when the detected content is greater than the preset threshold, it is determined that the trap chip is valid.

In some embodiments, Least squares support vector machines (LS-SVM) are a machine learning method. The entire content of the document Suykens, Johan A K. Least square support vector machines [J]. International Journal of Circuit Theory & Applications, 2002, 27(6): 605-615. is hereby cited.

In some embodiments, the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) is a semi-supervised SVR algorithm based on quantum particle swarm optimization. Literature research on near-infrared spectroscopy modeling method based on semi-supervised and transfer learning [D]. Ocean University of China, 2012. The entire content is cited here.

In some embodiments, the modeling steps of the near-infrared spectrum quantitative analysis based on the semi-supervised least squares support vector machine regression algorithm are:

Step 1: Input the spectral data, perform preprocessing such as noise reduction on the spectrum;

The second step: using the cross-validation method, reserve 10% of the samples as test samples;

Step 3: Set initial parameters, execute the QPSO-LSS3VR algorithm, and estimate unlabeled samples;

Step 4: Execute the QPSO-LSS3VR algorithm to obtain the best model parameters and establish a spectral quantitative analysis model;

Step 5: Input the test sample into the model for detection and evaluate the model performance.

Glossary

Sex pheromones are chemical substances secreted by animals to communicate sexual information between the sexes of the same species. Acts as a sex attractant.

Beneficial effect

Adopting this technical solution, for the first time, the problem of the inability to measure the concentration of pheromone in the sex pheromone trap is solved, and then the effectiveness of the sex pheromone trap is evaluated, and the operation is simple and fast, and it is non-destructive to the sex pheromone trap , can be reused and other advantages.

The overall innovation of the entire process of the disclosed method, the inventors have achieved the best results by using specific spectral acquisition parameters, specific spectral preprocessing parameters, specific QPSO-LSS3VR algorithm parameters and specific quantitative model building parameters.

The making of modeling samples in the disclosed method is also unique, for example, some embodiments adopt organic solvent (consisting of acetone and NN-dimethylformamide, the volume ratio of acetone and NN-dimethylformamide is 18.5～19.5: 1) Extracting the standard trap chip to obtain a standard solution, and diluting the standard solution into multiple gradient solutions with gradient concentrations.

Description of drawings

Fig. 1 is a schematic diagram of some embodiments of the method for detecting the content of sex pheromones in a trap chip according to the present invention.

Detailed ways

Embodiments of the present invention will be described in detail below in conjunction with examples, but those skilled in the art will understand that the following examples are only used to illustrate the present invention, and should not be considered as limiting the scope of the present invention. Those who do not indicate the specific conditions in the examples are carried out according to the conventional conditions or the conditions suggested by the manufacturer. The reagents or instruments used were not indicated by the manufacturer, and they were all commercially available conventional products.

Fig. 1 shows a schematic flow chart of a method for detecting sex pheromone content in a trap chip of the present disclosure, including:

Step 101, extracting the standard trap chip with an organic solvent to obtain a standard solution, and diluting the standard solution into multiple gradient solutions with gradient concentrations;

Step 102, using a near-infrared spectrometer to collect the spectra of the standard solution and multiple gradient solutions;

Step 103, based on the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR), the sample spectrum obtained in step (2) is fitted with the corresponding ratio concentration value, and the sex pheromone content in the sex pheromone trap is established Change quantitative detection model;

Step 104, using the model obtained in step (3) to carry out quantitative analysis of the sex pheromone content in the trap chip to be tested;

The organic solvent is a solution obtained by mixing acetone and NN-dimethylformamide (DMF) at a volume ratio of 19:1.

Embodiments of the present invention will be described in detail below in conjunction with Example 1

Example 1

The following details the method for measuring the change of sex pheromone concentration in the sex pheromone trap:

(1) Modeling sample production: Select 200 qualified trap chips just produced, add 2000ml of solvent (mixture of acetone and NN-dimethylformamide (DMF) at a volume ratio of 19:1), soak and extract for 24 hours , pour out the solution and stir evenly, divide it into 200 parts on average, use 10 parts as the reference solution, and add a certain amount of acetone to the remaining 190 parts to prepare a preparation solution containing 70%-99.9% of the reference solution, and set the value of the reference solution is 100, and the proportioning liquid value is taken as the corresponding value according to the percentage containing the base liquid.

(2) Collect sample spectrum: Use the liquid transmission sampling module of the near ^- infrared spectrometer to collect the sample spectrum of the sample prepared above. scans with a resolution of 8cm ^-1 ;

(3) Spectral preprocessing: In order to reduce noise and baseline drift in near-infrared spectrograms, one or more combinations of vector normalization, first-order derivative, second-order derivative, multivariate signal correction and spectral smoothing methods are used to The near-infrared spectrum is preprocessed; in the example, the sample spectrum is obtained by the following method to obtain ideal results:

aUsing multiple signal correction (MSC) to eliminate the difference caused by sample inhomogeneity;

b Adopt first-order differential processing to eliminate the influence of baseline drift and obtain higher resolution and clearer spectral profile changes than the original spectrum;

c. Use the Savitzky-Golay filter smooth spectrum with a segment length of 9 and an interval of 5 to eliminate high-frequency noise and retain useful low-frequency information;

(4) QPSO-LSS3VR algorithm design:

a) All sample spectra are processed by TQ software for dimensionality reduction, and high-dimensional data is mapped to low-dimensional space, which is convenient for distance measurement;

b) Calculate the distance d(n _i , m _j ) between the sample n _i in the unmarked sample set N and the sample m _j in the marked sample set M.

c) According to the KNN algorithm, solve the set M' of k marked neighbors for each unlabeled sample n _i .

d) Take the average value of all marked samples in M', and estimate the initial estimated value of the unmarked sample n _i : T _n =T _l ×(1+rand(0,1))

Among them, _Tl represents the average value of the marked samples in the set.

e) For the unlabeled samples selected for training, replace the labeled value with the detection value currently given by the model; for the unlabeled samples that are not selected for the next iteration, keep the value unchanged.

(5) Establish a quantitative model: add 190 samples with different acetone ratios, scan the spectrum, preprocess the spectral data, select the best spectral segment in the wavenumber range of 8634 to 4102 cm ^-1 , and use the cross-validation method to pre-process the spectral data. Leave 20 samples as test samples; set initial parameters: number of iterations M=100, number of individuals in the initialization group N=25, p=0.6, search range: α=[0,100], γ[0,1000], λ = [0, 1000]. Execute the QPSO-LSS3VR algorithm to estimate unlabeled samples. The model detection performance is evaluated by using the detection root mean square error RMSEC and the coefficient of determination R ² ; after calculation, α = [0, 3.9], γ [0, 118], λ = [0, 42.5] is optimal; the spectral quantitative analysis model was established according to the optimized parameters, and the model results: RMSEC=5.61, R ² =0.9665.

20 prepared test samples were added to the mathematical model as the verification set samples, and the test results were compared with the actual prepared ratios. The results are shown in Table 1 below.

Table 1 Verification results of the near-infrared model for detecting changes in sex pheromone content

As shown in Table 1, the average relative error obtained by the detection method of the present invention is only 5.18%, which has the advantage of high accuracy.

It can be seen from the results that the near-infrared data model established by using the semi-supervised learning method based on quantum particle swarm optimization-least squares support vector machine regression algorithm can quickly and accurately detect Changes in the sex pheromone content in the sex pheromone trap, so as to determine whether the efficacy of the drug has expired.

Although the specific implementation of the present invention has been described in detail, those skilled in the art will understand that: according to all the teachings that have been disclosed, various modifications and changes can be made to the details, and these changes are all within the protection scope of the present invention . The full scope of the invention is given by the appended claims and any equivalents thereof.

Claims (10)

1. A method of detecting the amount of a sex pheromone in a trap chip comprising:

(1) Extracting the standard trapper chip by using an organic solvent to obtain a standard solution, and diluting the standard solution into a plurality of gradient solutions with gradient concentrations;

(2) Collecting spectra of a standard solution and a plurality of gradient solutions by using a near-infrared spectrometer;

(3) Fitting the sample spectrum obtained in the step (2) with a corresponding ratio concentration value based on a semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) to establish a quantitative detection model of the change of the content of the sex pheromone in the sex pheromone trap;

(4) Carrying out quantitative analysis on the content of the sex pheromone in the trapper chip to be detected by using the model obtained in the step (3);

The organic solvent is composed of acetone and NN-dimethylformamide, and the volume ratio of the acetone to the NN-dimethylformamide is 18.5-19.5: 1.

2. The method of claim 1, further comprising between steps (2) and (3) a step of pre-processing the spectrum by a method selected from the group consisting of: one or more of first order derivatives, second order derivatives, vector normalization, multivariate signal correction, and spectral smoothing.

3. the method of claim 1, further comprising, between steps (2) and (3), a step of pre-processing the spectrum, the pre-processing method comprising one or more of:

-eliminating the differences due to sample inhomogeneities by using Multivariate Signal Correction (MSC);

The influence of baseline drift is eliminated by adopting first-order differential processing, and spectral profile change with higher resolution and clearer resolution than the original spectrum is obtained; and

-smoothing the spectrum using Savitzky-Golay filtering with a segment length of 9 and an interval of 5, eliminating high frequency noise and preserving useful low frequency information.

4. the method according to claim 1, wherein the method for establishing the model for quantitatively detecting the change in the content of the sex pheromone in the sex pheromone trap in the step (3) comprises the following steps:

i. Inputting spectral data;

Setting initial parameters, executing a semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR), and estimating unmarked samples;

And iii, executing a semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) to obtain the optimal model parameters, and establishing a spectrum quantitative analysis model.

5. The method of claim 4, wherein between steps i and ii, about 10% of samples are reserved for testing by adopting a cross-validation method; after the model is established, the test sample is input into the model for detection, and the performance of the model is evaluated.

6. The method of claim 1, step (3) comprising the operations of,

Establishing a quantitative model: adding 190 parts of sample into different acetone ratios, scanning the spectrum, preprocessing the spectrum data, and selecting 8634-4102 cm^-1In the optimal spectrum section in the wave number range, 20 samples are reserved as test samples by adopting a cross validation method;

Setting initial parameters: the iteration number M is 100, the number N of the initialized population is 25, p is 0.6, and the search range is: α ═ 0, 100], γ [0, 1000], λ ═ 0, 1000 ];

Executing a QPSO-LSS3VR algorithm, estimating an unmarked sample, and evaluating the model detection performance by selecting a detection root mean square error RMSEC and a decision coefficient R2; the calculation results show that α ═ 0, 3.9], γ [0, 118], and λ ═ 0, 42.5 are optimal;

Establishing a spectrum quantitative analysis model according to the optimized parameters, wherein the model result is as follows: RMSEC ═ 5.61, R²＝0.9665。

7. The method of claim 1, the semi-supervised least squares support vector machine regression algorithm (QPSO-LSS3VR) designed as follows:

a) the TQ software is adopted to perform dimensionality reduction processing on the sample spectrum, and high-dimensional data is mapped to a low-dimensional space, so that distance measurement is facilitated;

b) Computing a sample N in a set N of unlabeled samples_iAnd the samples M in the marked sample set M_jDistance d (n) of_i，m_j)。

c) solving each unlabeled sample n according to KNN algorithm_ik marked with a nearthe neighbors are grouped by M'.

d) Taking the average value of all marked samples in M' and estimating the unmarked sample n_iThe initial estimate of (d) is: t is_n＝T_l×(1+rand(0,1))

wherein, T_lRepresenting the mean of the labeled samples in the collection.

e) For the unlabeled sample selected for training, replacing the labeled value with the detection value currently given by the model; the values are kept unchanged for unlabeled samples that are not selected for the next iteration.

8. the method according to claim 1, wherein the step (1) comprises selecting 200 qualified trap chips which are just produced, adding 2000ml of organic solvent (acetone and NN-dimethylformamide mixed proportioning solution), soaking and extracting for 24 hours, pouring out the solution, uniformly stirring, evenly dividing the solution into 200 parts, taking 10 parts as reference solution, adding a certain amount of acetone into the other 190 parts respectively to prepare a preparation solution containing 70-99.9 vol% of reference solution, setting the reference solution value as 100, and taking the proportioning solution value as the corresponding value according to the percentage of the reference solution.

9. the method according to claim 1, wherein the step (2) comprises collecting the sample spectrum with a liquid transmission sampling module of a near infrared spectrometer, wherein the wave number is 10000-3800 cm^-1Taking the background in the instrument as reference, the sample and reference are scanned for 70 times, and the resolution is 8cm^-1。

10. A method of determining the validity of a trap chip, comprising the steps of:

extracting a trapper chip to be detected by using an organic solvent to obtain a solution to be detected;

Detecting the content of the sex pheromone in a solution to be detected by using the method of any one of claims 1 to 9 to obtain a detected content;

And comparing the detected content with a preset threshold, judging that the trap chip is invalid when the detected content is less than the preset threshold, and judging that the trap chip is valid when the detected content is greater than the preset threshold.