High-sensitivity carbendazim detection method based on terahertz metamaterial resonance enhancement

Abstract

The invention discloses a high-sensitivity carbendazim detection method based on terahertz metamaterial resonance enhancement, and provides a terahertz metamaterial sensor with a cross-shaped composite double-peak structure, a terahertz spectrum technology is utilized to collect transmission spectrums, the difference of different concentrations among terahertz spectrums is explored, and the detection accuracy is improved. The analysis finds that the transmission peak amplitude of the terahertz transmission spectrum of the structure is in a decreasing trend along with the increase of the concentration of a carbendazim solution, data is preprocessed through multiple scatter correction (MSC), feature extraction is performed by adopting an iterative information variable reservation (IRIV) method, and a quantitative detection model of the carbendazim terahertz transmission spectrum is established based on an LS-SVM (least squares support vector machine). The invention provides a novel approach for quantitative detection of carbendazim, and experimental results show that the RP of the model reaches 0.9825, the RMSEP is 0.2001, and the detection limit LOD is 0.672 mu g / ml. The invention provides a new method for rapid, lossless and high-sensitivity detection of carbendazim in food.

Description

Technical Field

[0001] The present invention relates to the technical field of chemical detection, and particularly to a highly sensitive detection method for carbendazim based on terahertz metamaterial resonance enhancement. Background Art

[0002] Carbendazim (CBZ) is a widely used fungicide belonging to the benzimidazole compound class. It is mainly used to control fungal diseases of various crops, such as apple ring rot, anthracnose, and fruit rust. As a systemic fungicide, carbendazim achieves its control effect by inhibiting the synthesis of the fungal cell wall and the mitosis process. However, with the increasing use of pesticides in agricultural production, the residue problem of carbendazim has gradually attracted attention in the field of food safety. Research shows that the residue of carbendazim on crops may pose potential hazards to consumer health. Long-term consumption of food containing carbendazim residues may cause endocrine disorders or other chronic diseases. To ensure that the carbendazim residue in food is within the legal and safe range, many countries and regions have established corresponding regulatory standards. For example, the national standard of China, "GB 2763-2021 Maximum Residue Limits of Pesticides in Foods", stipulates that the maximum allowable residue concentration of carbendazim in apples is 5 μg / ml. Therefore, researching a method for accurately, quickly, and non-destructively detecting carbendazim residues in agricultural products has very important practical significance and application value.

[0003] Currently, the commonly used methods for detecting pesticide residues at home and abroad include high-performance liquid chromatography, enzyme-linked immunosorbent assay, electrochemistry, near-infrared spectroscopy, Raman spectroscopy, etc. Although the existing methods have advantages in sensitivity, there are still many problems, such as high detection cost, long cycle, poor result stability, complicated experimental process, high cost of instrument equipment, and possible damage to samples. Therefore, there is an urgent need to develop a detection technology that is both highly sensitive and can be fast and non-destructive.

[0004] Terahertz (THz) waves generally refer to electromagnetic waves with a frequency range between 0.1 and 10 THz. Due to its high stability, high resolution, and rich optical properties, it has become an ideal tool in the fields of biological and chemical molecule detection. In recent years, more and more researchers have begun to explore the application of terahertz spectroscopy detection technology in fields such as food safety. However, when traditional terahertz spectroscopy technology is used to detect trace toxic substances, it usually faces problems such as insufficient sensitivity and inability to meet the detection standards. This may be because the low content of the substance leads to a weakened response of terahertz waves to it, restricting the resolution ability and detection accuracy of terahertz spectra. The emergence of metamaterial sensors has provided a new idea for improving the detection limit of terahertz detection technology.

[0005] Metamaterial sensors have special electromagnetic response characteristics. Through the resonance principle, they can enhance the sensitivity of terahertz spectroscopy technology to trace components. By analyzing the frequency shift and amplitude changes caused by the sample, the accuracy and sensitivity of the detection are significantly improved. Summary of the invention

[0006] In view of the above situation, the purpose of the present invention is to provide a highly sensitive detection method for carbendazim based on terahertz metamaterial resonance enhancement, to develop a new metamaterial structure, and to use the metamaterial to realize a method for quantitative detection of carbendazim, thereby providing a new approach for quantitative detection of carbendazim.

[0007] The present invention provides a highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement, wherein the terahertz metamaterial sensor has the following parameters: a period P1 of 70 μm, a coating metal of gold, and a silicon substrate; the terahertz metamaterial absorber is composed of a plurality of periodic structural units, the periodic structural unit is a square structure, the width of the periodic structural unit is 70 μm, four cross structures are arranged inside the periodic structural unit, a strip structure is arranged around the periodic structural unit, the width L1 of the cross structure is 30 μm, the cross structure is composed of two mutually perpendicular "I"-shaped structures, the width L3 of the end of the "I"-shaped structure is 4 μm, the width W2 of the middle of the "I"-shaped structure is 2 μm, the length L2 of the strip structure is 60 μm, and the width W1 of the strip structure is 2 μm; the refractive index of the silicon substrate is 3.335, and the thickness of the silicon substrate is 500 μm;

[0008] The detection method comprises the following steps:

[0009] (1) preparing a plurality of carbendazim standard solutions of different concentrations, and dropping the carbendazim standard solutions on the terahertz metamaterial absorber respectively, and placing the absorber in a drying oven for drying;

[0010] (2) placing the dried terahertz metamaterial absorber into a terahertz system and measuring it in transmission mode to collect the original terahertz spectrum information of the corresponding sample;

[0011] (3) using Mahalanobis distance to remove outliers in the original terahertz spectrum information and obtain the remaining spectrum data;

[0012] (4) Preprocess the remaining spectral data using multivariate scattering correction, and then use the iterative information retention variable method to extract features from the preprocessed data;

[0013] (5) A quantitative detection model of carbendazim terahertz transmission spectrum was established based on the extracted features and LS-SVM;

[0014] (6) Obtain the terahertz spectral information of the carbendazim-containing sample to be detected and input it into the quantitative detection model to obtain the content of carbendazim in the sample to be detected.

[0015] Compared with the prior art, the high-sensitivity detection method for carbendazim based on terahertz metamaterial resonance enhancement provided by the present invention proposes a terahertz metamaterial sensor with a "cross"-shaped composite double-peak structure. The transmission spectrum is collected using terahertz spectroscopy technology, and the differences between terahertz spectra at different concentrations are explored. It is found through analysis that the amplitude of the transmission peak of the terahertz transmission spectrum of this structure shows a downward trend as the concentration of the carbendazim solution increases. The data is preprocessed by multiplicative scatter correction (MSC), the iterative retained information variable method (IRIV) is used for feature extraction, and a quantitative detection model for the terahertz transmission spectrum of carbendazim is established based on LS-SVM, providing a new approach for the quantitative detection of carbendazim. The experimental results show that the R P reaches 0.9825, the RMSEP is 0.2001, and the detection limit LOD is 0.672 μg / ml. The present invention provides a new method for the rapid, non-destructive, and highly sensitive detection of carbendazim in food.

[0016] In addition, according to the high-sensitivity detection method for carbendazim based on terahertz metamaterial resonance enhancement provided by the present invention, it also has the following technical features:

[0017] Among them, the terahertz metamaterial absorber is prepared by the following preparation method:

[0018] First, select a silicon substrate. The surface of the substrate is cleaned to remove impurities and undergoes oxidation treatment to enhance the adhesion of the photoresist. Subsequently, a chromium layer is deposited on the silicon substrate using thin-film deposition technology, and then a metal layer is deposited to enhance the adhesion between the metal layer and the silicon substrate. Then, the silicon wafer is coated with a layer of photoresist with a thickness of 1-2 μm using the spin-coating process. After coating, pre-baking is carried out, baking at 100 °C for 120 seconds to remove the solvent in the photoresist, ensuring its hardening and readiness for exposure. After exposure, it enters the development process, and the development time is controlled within 30 seconds to remove the photoresist in the unexposed part, leaving a pattern. After development, solvent is used for glue application to remove the residual photoresist. Then, etching operations are carried out using reactive ion etching or ion beam etching. After etching is completed, the photoresist mask is removed by ultrasonic cleaning with acetone to ensure that there are no residues on the chip surface. Finally, post-treatment such as surface oxidation or metal deposition is carried out to enhance conductivity and corrosion resistance.

[0019] Among them, the specific content of step (1) includes:

[0020] The concentration of the carbendazim standard solution was selected to be 100 μg / ml, and ultrapure deionized water was used to dilute the carbendazim standard solution using a pipette to prepare 21 carbendazim solutions with different concentration gradients. The prepared sample solutions were placed on a vortex oscillator for mixing and shaking for 3 minutes to ensure that the carbendazim standard solution could be fully diluted in deionized water and stored in a volumetric flask.

[0021] Wherein, the step (2) specifically includes:

[0022] Before the experiment, the system was preheated by an air compressor with an air dryer, and dry air was continuously filled into the optical cavity. The humidity was monitored by a thermometer and hygrometer to keep it below 10% and the temperature constant at 25±0.5℃. A pipette was used to draw 20μl of carbendazim sample solution in the order of sample concentration from low to high and drop it on the terahertz metamaterial absorber. Before collecting the sample spectrum, the terahertz metamaterial absorber with the sample was placed in a drying oven, and the temperature of the drying oven was set to 50℃ for 30min. Finally, the terahertz metamaterial absorber with the sample was placed in the terahertz system and measured in transmission mode. After each sample was placed, wait for 2 minutes and start measuring after the system stabilized. Five points were taken for each sample, and 10 measurements were performed at each point. A total of 1050 spectra were obtained for 21 concentration samples.

[0023] Wherein, the step (3) specifically includes:

[0024] The variables were normalized by the covariance matrix, the correlation between the features was identified, and the Mahalanobis distance threshold was set to 0.99, and then the outliers in the original terahertz spectrum information were eliminated to obtain the remaining spectrum data.

[0025] Wherein, in step (5), the predicted value and the true value of the established quantitative detection model of the terahertz transmission spectrum of carbendazim are verified by using F-test and t-test. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0027] Figure 1 Schematic diagram of the structure of the terahertz metamaterial absorber;

[0028] Figure 2 This is a simulation diagram of the terahertz transmission spectrum;

[0029] Figure 3 The electromagnetic field distribution and terahertz transmission spectrum of the metamaterial structure simulated by FDTD: (a) the spatial distribution of the electric field intensity at 1.22 THz; (b) the spatial distribution of the electric field intensity at 2.30 THz;

[0030] Figure 4 is the transmission spectrum of carbendazim solutions with different concentrations on the surface of the metamaterial sensor;

[0031] Figure 5 is the Mahalanobis distance distribution map of different spectral data;

[0032] Figure 6 is 1 st Variable selection results of the data processed by 1

[0033] Figure 7 Variable selection results of the data processed by MSC through different feature extraction methods;

[0034] Figure 8 is the fitting graph of the predicted value and the true value of the LS - SVM model for the THz spectrum of the carbendazim solution. Detailed implementation manners

[0035] To make the objectives, features, and advantages of the present invention more obvious and understandable, the following provides a detailed description of the specific implementation manners of the present invention with reference to the accompanying drawings. Several embodiments of the present invention are given in the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0036] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the description of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0037] Please refer to Figures 1 to 3 , a terahertz metamaterial absorber provided in an embodiment of the present invention. In this embodiment, the structural parameters of the metamaterial absorber are designed by simulating parameters using FDTD Solutions software. Since the metamaterial is composed of periodic structural units, its overall transmission or reflection characteristics can be represented by the performance of a single periodic unit. Therefore, in the simulation analysis, the simulation efficiency is effectively improved by simulating a periodic unit of the metamaterial.

[0038] Specifically, the terahertz metamaterial sensor is a terahertz metamaterial sensor with a "cross"-shaped composite double-peak structure, having the following parameters: the period P1 is 70 μm, the coating metal is gold, and a silicon substrate is used; the terahertz metamaterial absorber is composed of multiple periodic structural units, the periodic structural unit is a square structure, the width of the periodic structural unit is 70 μm, four cross-shaped structures 10 are provided inside the periodic structural unit, a strip structure 20 is provided on each of the four sides of the periodic structural unit, the width L1 of the cross-shaped structure 10 is 30 μm, the cross-shaped structure 10 is composed of two perpendicular "I"-shaped structures 11, the width L3 of the end of the "I"-shaped structure 11 is 4 μm, the width W2 of the middle of the "I"-shaped structure 11 is 2 μm, the length L2 of the strip structure 20 is 60 μm, the width W1 of the strip structure 20 is 2 μm; the refractive index of the silicon substrate is 3.335, and the thickness of the silicon substrate is 500 μm.

[0039] The processing process of the metamaterial sensor is a precise operation. First, a high-quality silicon substrate is selected to ensure its smoothness and flatness. The surface of the substrate is cleaned to remove impurities and then oxidized to enhance the adhesion of the photoresist. Subsequently, a chromium layer of about 5 nm is deposited on the silicon substrate using thin-film deposition technology, and then a metal layer of about 100 nm is deposited to enhance the adhesion between the metal layer and the silicon substrate. Then, through the spin-coating process, the silicon wafer is coated with a layer of photoresist, and the rotation speed is controlled at about 3000 rpm to ensure that the photoresist is evenly distributed, and the thickness is usually 1-2 μm. After coating, pre-baking (soft baking) is carried out, and it is baked at 100 °C for 120 seconds to remove the solvent in the photoresist, ensure its hardening and prepare for exposure. After exposure, it enters the development process, and the development time is controlled at 30 seconds to remove the unexposed part of the photoresist and leave the pattern. After development, a glue removal treatment is carried out using a solvent (such as acetone) to remove the residual photoresist. Then, etching operations are carried out using reactive ion etching (RIE) or ion beam etching (IBE). After etching is completed, the photoresist mask needs to be removed by ultrasonic cleaning with acetone to ensure that there are no residues on the chip surface. Finally, to improve the performance of the metamaterial chip, post-treatment such as surface oxidation or metal deposition is usually carried out. The surface oxidation temperature is usually controlled at 1000 °C, and metal deposition uses evaporation or sputtering technology to enhance conductivity and corrosion resistance and further improve the performance of the chip.

[0040] After the metamaterial chip is processed, the terahertz full-band simulation of the terahertz metamaterial sensor with a "cross"-shaped composite double-peak structure is carried out using FDTD Solutions. Figure 2 It is a simulation diagram of the terahertz transmission spectrum. The simulation results show that the metamaterial sensor has two characteristic peaks in the terahertz band, which are 1.22 THz and 2.30 THz respectively.

[0041] As Figure 3 shown in Figure 3 , it is the spatial distribution of the electric field intensity around the "cross"-type composite terahertz metamaterial sensor calculated by FDTD. Among them Figure 3 (a) is the spatial distribution of the electric field intensity at 1.22 THz, Figure 3 (b) is the spatial distribution of the electric field intensity at 2.30 THz. The electric field intensity gradually changes from blue to red. The red area represents the highest value of the electric field intensity, and the dark red part corresponds to the maximum value of the electric field intensity. The area with the strongest electric field intensity is regarded as the terahertz hot spot, and the hot spot area usually enhances the signal of the terahertz wave. Through in-depth analysis of the simulation results, it can provide a theoretical basis for selecting an appropriate resonance frequency and further optimize the performance of the metamaterial sensor.

[0042] After being simulated by FDTD software, this metamaterial structure theoretically has the ability to detect trace substances. In order to verify the detection ability of the metamaterial sensor chip processed based on this metamaterial structure for trace carbendazim solution, subsequently, starting from the actual situation, carbendazim solutions with different concentration gradients will be configured, and through methods such as THz spectrum acquisition, algorithm model construction and optimization, etc., to illustrate the actual application effect and significance of this metamaterial structure.

[0043] Another embodiment of the present invention provides a highly sensitive detection method for carbendazim based on terahertz metamaterial resonance enhancement. The terahertz metamaterial absorber is the above-mentioned terahertz metamaterial absorber. The detection method includes the following steps (1) to (6):

[0044] (1) Configure multiple carbendazim standard solutions with different concentrations, and drop the carbendazim standard solutions on the terahertz metamaterial absorber respectively, and put them into the drying oven to dry.

[0045] The concentration of the carbendazim standard solution is 100 μg / ml. In order to detect the residue of trace carbendazim solution, a pipette gun is used to suck ultrapure deionized water to dilute the carbendazim standard solution. According to the national standard requirements, 21 carbendazim solutions with different concentration gradients are prepared. The prepared sample solution is placed on a vortex oscillator and mixed and shaken for 3 min to ensure that the carbendazim standard solution can be fully diluted in deionized water, and stored in a volumetric flask. The specific concentration gradients of the carbendazim solution are shown in Table 1.

[0046] Table 1 Concentration gradient table of carbendazim solution

[0047]

[0048] (2) Put the dried terahertz metamaterial absorber into the terahertz system and measure it in transmission mode to collect the original terahertz spectral information of the corresponding sample.

[0049] The terahertz experimental equipment used for the detection of trace carbendazim solution residues is the TAS7500 terahertz time-domain spectrometer of Advantest Corporation, Japan. This system consists of three main parts: a femtosecond laser, a terahertz emitter, and a terahertz detector. The spectral measurement range is 0.1 - 5.0 THz, and the resolution of the instrument is 7.6 GHz. To avoid interference from the surrounding environment and obtain stable information of the sample, before the experiment, the system is preheated by an air compressor with an air dryer, and dry air is continuously filled into the optical cavity. Monitored by a hygrometer, the humidity is maintained below 10%, and the temperature is kept constant at 25 ± 0.5 °C. To reduce the influence of the concentration residue on the metamaterial chip, a pipette is used to aspirate 20 μl of carbendazim sample solution in ascending order of sample concentration and drop it on the metamaterial chip. Since water strongly absorbs terahertz waves, before collecting the sample spectrum, to ensure that the solution on the metamaterial chip can be fully dried, the metamaterial chip with the sample is placed in an oven. The oven temperature is set at 50 °C, and the drying time is 30 min. Finally, the metamaterial chip with the sample is placed in the terahertz system and measured in transmission mode. Wait 2 minutes after each sample is placed, and start the measurement after the system stabilizes. To reduce the influence of random errors on the experimental results, 5 points are taken for each sample, and each point is measured 10 times. A total of 1050 spectra are obtained for 21 concentration samples.

[0050] First, collect the terahertz time-domain spectra of 21 different concentrations of carbendazim, and obtain the frequency-domain signals by fast Fourier transform (FFT), expressed as formula (1):

[0051]

[0052] Among them, A(ω) represents the electric field amplitude, is the phase difference between the reference signal and the sample signal, and E(t) is the terahertz time-domain signal. The refractive index and absorption coefficient of the sample are obtained through formula (2) and formula (3).

[0053]

[0054] Among them, n(ω) is the refractive index, α(ω) is the absorption coefficient, ω is the frequency, k(ω) is the extinction coefficient, ρ(ω) is the amplitude ratio function, d is the sample thickness, and c is the speed of light in vacuum.

[0055] Subsequently, the terahertz transmission spectrum of carbendazim is extracted by comparing the sample spectrum and the reference spectrum.

[0056] T = (A S / A R ) 2 (4)

[0057] Among them, A S and AR They are the amplitudes of the sample signal and the reference signal respectively.

[0058] The detection model for the carbendazim solution content consists of the related coefficient of the prediction set (R P ), the root mean square error of the prediction set (RMSEP), the related coefficient of the correction set (R C ), and the root mean square error of the correction set (RMSEC) to evaluate the model. Among them, R P and R C are between 0 and 1. The closer R P and R C are to 1, the higher the accuracy of the model. The closer and smaller the values of RMSEP and RMSEC are, the more stable the model is.

[0059]

[0060]

[0061] The limit of detection (LOD) refers to the minimum concentration required for a substance to be possibly detected. The detection limit (LOD) with a confidence interval of 99.86% can be calculated based on the slope of the fitting curve between the true value and the predicted value established from the THz transmission spectrum and the variance of the prediction error.

[0062]

[0063] Among them, σ is the standard error of the predicted concentration, m is the slope of the fitting curve, and in the model, the value of RMSEP is equal to σ.

[0064] (3) Use the Mahalanobis distance to eliminate the outliers in the original terahertz spectral information and obtain the remaining spectral data.

[0065] Among them, the "cross"-type composite metamaterial combined with the THz-TDS technology is used to detect and analyze the carbendazim solution. Figure 4This is the transmission spectrum of the carbendazim solution at a certain concentration on the surface of the metamaterial sensor. Considering that there is a lot of noise interference in the frequency bands below 1.0THz and above 3.0THz, in order to ensure the accuracy of the measurement, the transmission spectrum of the 1.0-3.0THz frequency band is intercepted. The 263 spectral data contained in this frequency band are used for subsequent modeling. Although the content of carbendazim on the surface of the "cross" composite metamaterial is extremely low, after signal enhancement sensing detection, there is an obvious amplitude response law in the transmission peak. That is, the amplitude of the transmission peak shows a downward trend with the increase of the concentration of the carbendazim solution. The reason for this phenomenon may be that when the concentration of the carbendazim solution increases, more carbendazim molecules in the solution will absorb the terahertz wave, and the refractive index of the metamaterial surface will change, resulting in the weakening of the energy of the transmission wave and the decrease of the transmission peak amplitude.

[0066] During the spectral data collection process, changes in the measurement environment and the stability of the spectral instrument may cause abnormal spectra in some samples. These abnormal data will interfere with model training and affect the final prediction accuracy, so they need to be screened out before modeling.

[0067] The Mahalanobis distance (MD) used in the present invention is to remove outliers by normalizing the variables through the covariance matrix, and can identify the correlation between the features. In spectral data processing, the absorption values ​​of different wavelengths are usually correlated, and the Mahalanobis distance can effectively detect those abnormal samples with a large degree of deviation from the overall distribution. In addition, spectral data are usually high-dimensional, and the Mahalanobis distance can measure the degree of deviation of the sample from the overall distribution in a multidimensional space. Therefore, it is more commonly used for spectral outlier detection than one-dimensional statistical methods (such as Z-score, mean ± standard deviation).

[0068] The Mahalanobis distance threshold is set to 0.99, and points exceeding this value are considered outliers. Figure 5 This is the result of elimination using the Mahalanobis distance algorithm. A total of 115 abnormal spectral data were eliminated, leaving 935 spectral data.

[0069] The original 1050 data and the remaining 935 spectral data after the Mahalanobis distance was removed were used to establish PLS (partial least squares method) and LS-SVM (least squares support vector machine) models, and the model effects before and after the abnormal data were removed were compared. The results are shown in Table 2. The results show that for both models, the RP and RMSEP are improved after the outliers are removed. Therefore, using the Mahalanobis distance method to remove outliers in spectral data is an effective method to improve the prediction accuracy and robustness of the model, which can reduce the interference of abnormal data on the modeling process, thereby improving the reliability of the analysis results.

[0070] Table 2 Verification results of the two models before and after outliers were removed

[0071]

[0072] (4) Use multiplicative scatter correction to preprocess the remaining spectral data, and then adopt the iterative informative variable retention method to extract features from the preprocessed data.

[0073] In spectral analysis, the original data is often affected by various external factors, such as ambient light interference, baseline drift, light scattering effect, instrument noise, etc. These factors may cause spectral signal distortion, reduce the reliability of the data, and thus affect the accuracy of modeling and analysis. Therefore, before data modeling, it is necessary to preprocess the spectral data to improve the data quality and make it more suitable for subsequent analysis.

[0074] In this application, four common preprocessing methods, namely discrete cosine transform (DCT), first derivative (1 st D), multiplicative scatter correction (MSC), and wavelet transform (WT), are adopted. They can effectively reduce and remove interference factors such as random noise, background interference, baseline drift, and light scattering effect, and improve the stability and accuracy of modeling.

[0075] The preprocessed data is divided into a calibration set and a prediction set according to a ratio of about 3:1 by the KS (Kennard-Stone) algorithm, and then input into the PLS model. The results are shown in Table 3. It can be seen from the results that the PLS model effect of the data preprocessed by the DCT and WT algorithms has decreased. This may be because in the process of removing noise, smoothing data, and correcting the baseline, the preprocessing method may lose high-frequency detail information, resulting in the weakening of the key features of the original spectral data or the distortion of the data distribution, making it difficult for the model to adapt and the accuracy and stability of the model to decline. In contrast, the data preprocessed by the 1 st D and MSC algorithms can better retain the characteristics of the original data, while reducing the interference such as noise in the original spectrum and improving the PLS model effect. The data preprocessed by the 1 st D has the best effect, with the prediction correlation coefficient R P reaching 0.8961 and the root mean square error of prediction RMSEP reaching 0.4435.

[0076] Table 3 Verification results of PLS models with different preprocessing methods

[0077]

[0078] Similarly, the calibration set and prediction set divided by the KS algorithm are input into the LS-SVM model, and the results are shown in Table 4. The results show that for the preprocessed data, the performance in the LS-SVM model has both improvements and declines. The best preprocessing method is MSC, and its prediction correlation coefficient R P reaches 0.9766, and the root mean square error RMSEP reaches 0.2349. At this time, its parameter combination is γ = 1192.861, σ 2 = 73.2734. In this embodiment, MSC is finally selected as the preprocessing method.

[0079] Table 4 Verification results of the LS-SVM model with different preprocessing methods

[0080]

[0081] After preprocessing the original data, the model has a better performance. However, in spectral data, the data dimension is very high, including 263 variables, and only a few of these variables are of practical significance for target analysis. High-dimensional data not only increases the computational complexity but also easily leads to the "curse of dimensionality", that is, the model is easily affected by data noise, resulting in overfitting. Feature extraction can reduce the dimension of high-dimensional data, remove redundant information, and retain the most representative features, thereby reducing the computational complexity and improving the analysis efficiency. In addition, feature extraction helps the model focus on learning the parts that are meaningful to the target variable by removing irrelevant or redundant bands, enhances the generalization ability of the model, avoids overfitting, and improves its performance on new data. Therefore, in order to remove redundant, noisy, and irrelevant information, this application will adopt feature extraction methods to help extract key information in the data, improve the analysis accuracy, and enhance the stability of the model.

[0082] This application will use five common feature extraction algorithms for the preprocessed data: Iteratively Retaining Informative Variables (IRIV), Princpalcomponent analysis (PCA), Successive projections algorithm (SPA), Competitive adaptive reweighting sampling (CARS), Genetic Algorithm (GA), etc. for feature extraction, and compare the model effects of each feature extraction algorithm. Figure 6 is 1 st Variables selected after feature extraction of the processed data.

[0083] The spectral data after feature extraction is input into the PLS model, and the model results are shown in Table 5. The results show that the best model is CARS-PLS, with its prediction correlation coefficient RP reaching 0.9484 and the root mean square error of prediction RMSEP reaching 0.3175. It can be seen from the table that after PCA and GA feature extraction, the model performance decreases slightly. This may be because after the data undergoes feature extraction and dimensionality reduction, the class distribution of the data or the non-linear relationship between features is not well preserved, losing key information and causing the model to be unable to effectively learn the discriminant features of the original data. In addition, if the number of features after dimensionality reduction is too small, the expressive ability of the model may be limited, possibly introducing noise or losing the discrimination of the data.

[0084] Table 5 Verification results of PLS models with different feature extraction methods

[0085]

[0086]

[0087] Similarly, Figure 7 are the variables selected after feature extraction of the data processed by MSC.

[0088] The spectral data after feature extraction is input into the LS-SVM model, and the model results are shown in Table 6. The results show that: The performance of the LS-SVM model is generally better than that of the PLS model. Among them, the MSC-IRIV-LS-SVM model based on the RBF kernel function has the best performance, with its prediction correlation coefficient R P reaching 0.9825 and the root mean square error of prediction RMSEP reaching 0.2001. At this time, its parameter combination is γ = 10.9260, σ 2 = 2.3702.

[0089] Table 6 Verification results of LS-SVM models with different feature extraction methods

[0090]

[0091] (5) Establish a quantitative detection model for carbendazim terahertz transmission spectra based on the extracted features and LS-SVM.

[0092] The present invention finally selects to use multiplicative scatter correction to preprocess the remaining spectral data, then uses the iterative retained information variable method to extract features from the preprocessed data, and establishes a quantitative detection model for carbendazim terahertz transmission spectra based on the extracted features and LS-SVM (abbreviated as the MSC-IRIV-LS-SVM model in the present invention).

[0093] The MSC-IRIV-LS-SVM model has the best comprehensive effect. The reasons are as follows: 1) After solutions with different concentrations are dropped on the surface of the terahertz metamaterial and dried, the thickness of the residues is different, which will cause a scattering effect. The MSC preprocessing algorithm is often used to correct the influence of the scattering effect on the spectrum. Therefore, MSC can improve the effect of the spectrum in the model; 2) IRIV optimizes the feature subset through iteration, and LS-SVM optimizes the model parameters through iteration. The two have a hierarchical progressive relationship in the optimization logic, making the entire modeling process more in line with the data characteristics. Therefore, IRIV-LS-SVM can effectively improve the prediction accuracy and generalization ability of the model; 3) Compared with other feature extraction algorithms such as CARS and SPA, IRIV not only considers the importance ranking of variables in the feature selection process, but also avoids local optima through the reverse variable selection strategy, making the selected feature subset more stable and robust. After combining with LS-SVM, the model can be fitted in a better feature space, thereby further improving the modeling effect.

[0094] Figure 8 It is the fitting diagram of the predicted value and the true value of the optimal model LS-SVM for the transmission spectrum of carbendazim solution in terahertz metamaterials. The slope of the fitting curve reaches 0.8933, and the LOD is 0.672 μg / ml calculated according to formula (9). It verifies the high accuracy and feasibility of the "cross"-shaped composite metamaterial structure and the MSC-IRIV-LS-SVM model in trace carbendazim residues.

[0095] In order to explore the statistical significance of the optimal model MSC-IRIV-LS-SVM model, the present invention uses F-test and t-test to verify the predicted value and the true value. The verification results are shown in Table 7 and Table 8. The results show that in the F-test, "F = 0.994" < "one-tailed critical value = 1.213", "(F ≤ f) = 0.52" > 0.05. In the t-test, "t = 0.071" < "two-tailed critical value = 1.964", "(T ≤ t) double-tail = 0.944" > 0.05. There is no significant difference between the predicted value and the true value of the carbendazim content. Therefore, the MSC-IRIV-LS-SVM model has good statistical robustness and prediction effectiveness.

[0096] Table 7 Results of F-test

[0097]

[0098] Table 8 Results of t-test

[0099]

[0100] (6) Obtain the terahertz spectral information of the carbendazim-containing sample to be detected and input it into the quantitative detection model to obtain the content of carbendazim in the sample to be detected.

[0101] In summary, the highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement provided by the present invention proposes a terahertz metamaterial sensor with a "cross"-shaped composite double-peak structure. The transmission spectrum is collected by terahertz spectroscopy technology, and the differences between terahertz spectra at different concentrations are explored. It is found through analysis that the amplitude of the transmission peak of the terahertz transmission spectrum of this structure shows a downward trend with the increase of the carbendazim solution concentration. The data is preprocessed by multiplicative scatter correction (MSC), the iterative retained information variable method (IRIV) is used for feature extraction, and a quantitative detection model of the terahertz transmission spectrum of carbendazim is established based on LS-SVM, providing a new way for the quantitative detection of carbendazim. The experimental results show that the R P reaches 0.9825, the RMSEP is 0.2001, and the detection limit LOD is 0.672 μg / ml. The present invention provides a new method for the rapid, non-destructive and highly sensitive detection of carbendazim in food.

[0102] The above-described embodiments merely represent several implementation manners of the present invention, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims

1. A highly sensitive detection method for carbendazim based on terahertz metamaterial resonance enhancement, characterized in that: The terahertz metamaterial sensor has the following parameters: the period P1 is 70 μm, the coating metal is gold, and a silicon substrate is used; the terahertz metamaterial absorber is composed of a plurality of periodic structural units, the periodic structural unit is a square structure, the width of the periodic structural unit is 70 μm, four cross structures are arranged inside the periodic structural unit, a strip structure is arranged around each of the four sides of the periodic structural unit, the width L1 of the cross structure is 30 μm, the cross structure is composed of two mutually perpendicular "I"-shaped structures, the width L3 of the end of the "I"-shaped structure is 4 μm, the width W2 of the middle of the "I"-shaped structure is 2 μm, the length L2 of the strip structure is 60 μm, and the width W1 of the strip structure is 2 μm; the refractive index of the silicon substrate is 3.335, and the thickness of the silicon substrate is 500 μm; The detection method comprises the following steps: (1) preparing a plurality of carbendazim standard solutions of different concentrations, and dropping the carbendazim standard solutions on the terahertz metamaterial absorber respectively, and placing the absorber in a drying oven for drying; (2) placing the dried terahertz metamaterial absorber into a terahertz system and measuring it in transmission mode to collect the original terahertz spectrum information of the corresponding sample; (3) using Mahalanobis distance to remove outliers in the original terahertz spectrum information and obtain the remaining spectrum data; (4) Preprocess the remaining spectral data using multivariate scattering correction, and then use the iterative information retention variable method to extract features from the preprocessed data; (5) A quantitative detection model of carbendazim terahertz transmission spectrum was established based on the extracted features and LS-SVM; (6) Obtaining terahertz spectrum information of the sample containing carbendazim to be tested, and inputting it into the quantitative detection model to obtain the content of carbendazim in the sample to be tested.

2. The highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement according to claim 1 is characterized in that: The terahertz metamaterial absorber is prepared by the following preparation method: First, a silicon substrate is selected, and the surface of the substrate is cleaned to remove impurities and oxidized to enhance the adhesion of the photoresist; then, a layer of chromium is plated on the silicon substrate using thin film deposition technology, followed by a metal layer to enhance the adhesion between the metal layer and the silicon substrate; Then, by using a spin coating process, the silicon wafer is coated with a layer of photoresist with a thickness of 1-2 μm; After coating, a pre-bake is performed at 100°C for 120 seconds to remove the solvent in the photoresist and ensure that it is hardened and ready for exposure; After exposure, the development process begins. The development time is controlled at 30 seconds to remove the photoresist in the unexposed part, leaving the pattern. After development, a solvent is used for photoresist treatment to remove the residual photoresist; Then, an etching operation is performed by using reactive ion etching or ion beam etching; After etching is completed, the photoresist mask is removed by acetone ultrasonic cleaning to ensure that there is no residue on the chip surface; finally, surface oxidation or metal deposition post-treatment is performed to enhance conductivity and corrosion resistance.

3. The highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement according to claim 1 is characterized in that: The step (1) specifically comprises: The concentration of the carbendazim standard solution was selected to be 100 μg / ml, and ultrapure deionized water was used to dilute the carbendazim standard solution using a pipette to prepare 21 carbendazim solutions with different concentration gradients. The prepared sample solutions were placed on a vortex oscillator for mixing and shaking for 3 minutes to ensure that the carbendazim standard solution could be fully diluted in deionized water and stored in a volumetric flask.

4. The highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement according to claim 1 is characterized in that: The step (2) specifically comprises: Before the experiment, the system was preheated by an air compressor with an air dryer, and dry air was continuously filled into the optical cavity. The humidity was monitored by a thermometer and hygrometer to keep it below 10% and the temperature constant at 25±0.5℃. A pipette was used to draw 20μl of carbendazim sample solution in the order of sample concentration from low to high and drop it on the terahertz metamaterial absorber. Before collecting the sample spectrum, the terahertz metamaterial absorber with the sample was placed in a drying oven, and the temperature of the drying oven was set to 50℃ for 30min. Finally, the terahertz metamaterial absorber with the sample was placed in the terahertz system and measured in transmission mode. After each sample was placed, wait for 2 minutes and start measuring after the system stabilized. Five points were taken for each sample, and 10 measurements were performed at each point. A total of 1050 spectra were obtained for 21 concentration samples.

5. The highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement according to claim 1, characterized in that: The step (3) specifically comprises: The variables were normalized by the covariance matrix, the correlation between the features was identified, and the Mahalanobis distance threshold was set to 0.99, and then the outliers in the original terahertz spectrum information were eliminated to obtain the remaining spectrum data.

6. The highly sensitive detection method of carbendazim based on terahertz metamaterial resonance enhancement according to claim 1, characterized in that: In step (5), the predicted value and the true value of the established quantitative detection model of the terahertz transmission spectrum of carbendazim are verified by using F-test and t-test.

Images