Preparation Method of SERS Optical Fiber Probe and Pesticide Residue Detection Device Composed Thereof

Abstract

The present invention particularly relates to a preparation method of a SERS optical fiber probe, comprising the following steps: A. Using nano-chemical synthesis technology to prepare SiO2 or polystyrene microspheres with controllable particle sizes to obtain a nanostructure; B. Coating the nanostructure with a single-hole polymer containing pesticide molecular imprinting through a selective polymerization reaction on the surface of colloidal particles; C. Removing the pesticide molecular imprinting and the nanostructure by a solvent method to obtain a single-hole hollow polymer microsphere; D. Marking Raman tags to a position adjacent to the imprinting site by a surface modification method; E. Chemically modifying the polymer coating layer by a nano-surface modification method, and assembling a functional fluorescent sensitive material with nano-double-layer molecular imprinting onto the polymer probe. Coating a nano-active material sol on the surface of the optical fiber probe by a coating technology to obtain a polymer probe with a surface-active functional monomer coating; when a pesticide molecule binds to a molecular recognition site, the optical signal of the fluorescent group will change, and then high-sensitivity and rapid detection can be realized through the signal enhancement effect of the optical fiber.

Description

Technical Field

[0001] The invention relates to the technical field of pesticide residue detection, and in particular to a SERS optical fiber probe preparation method and a pesticide residue detection device constituted by the probe. Background Art

[0002] The use of pesticides plays an important role in preventing and controlling pests, diseases and weeds, increasing crop yields and ensuring national food security. Large-scale application of pesticides has brought economic benefits to modern agriculture, but it has also caused serious environmental pollution, increasing the amount of pesticide residues in the air, water and food, and posing a huge threat to the ecological environment and human health. The harm of pesticides to the human body is mainly reflected in the inhibition of cholinesterase activity in the nervous system, which makes it impossible to decompose acetylcholine in the human body. Short-term high-dose exposure to the human body can lead to acute poisoning, and when the human body is exposed to pesticide residues for a long time, it will cause many chronic diseases, such as various cancers, nervous system imbalances, and birth defects in infants and young children.

[0003] Pesticide residues have a wide range of impacts and are highly harmful, bringing huge negative impacts on food safety, human health, and agricultural product trade. The key technical measure to eliminate the above-mentioned adverse effects is to quickly and accurately analyze and detect pesticide residues, thereby monitoring the rational use of pesticides and preventing products with excessive pesticide residues from entering the circulation channels. Therefore, it is of great significance to conduct in-depth research on the detection technology of pesticide residues in agricultural products.

[0004] At present, the detection methods of pesticide residues can be divided into traditional detection methods (laboratory analysis methods) and enzyme-linked immunosorbent rapid detection methods. Traditional detection methods include gas chromatography, high-performance liquid chromatography, various improved chromatography, chromatography-mass spectrometry, capillary electrophoresis, etc. They are relatively mature and have high detection accuracy. Chromatography, mass spectrometry, and chromatography-mass spectrometry have been set as national standard detection methods. However, they have some application problems: cumbersome pretreatment, destructive to samples during analysis, large and expensive reagent consumption, long analysis time, and the detection instruments required for analysis and detection are bulky and difficult to move, which is not suitable for on-site online rapid detection. At present, there are many enzyme-linked immunosorbent rapid detection methods on the market. It is a new and practical detection method. The detection instrument used in this method is relatively cheap, the detection time is short, and it is easy to miniaturize. It has strong application value, but there are the following problems: limited detection concentration range, inability to quantitatively analyze, poor stability, and many interference factors. Summary of the invention

[0005] The primary purpose of the present invention is to provide a method for preparing a SERS optical fiber probe, which can prepare a probe that improves the Raman spectrum collection intensity and facilitates subsequent detection.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for preparing a SERS optical fiber probe, comprising the following steps: A. Using nano-chemical synthesis technology to prepare SiO2 or polystyrene microspheres with controllable particle sizes to obtain a nanostructure; B. Coating the nanostructure with a single-hole polymer containing a pesticide molecular imprint through a selective polymerization reaction on the surface of the colloidal particles; C. Removing the pesticide molecular imprint and the nanostructure through a solvent method to obtain a single-hole hollow polymer microsphere; D. Marking a Raman label to a position adjacent to the imprint site through a surface modification method; E. Chemically modifying the polymer coating layer by using a nano-surface modification method, and assembling a functional fluorescent sensitive material with a nano-double-layer molecular imprint onto the polymer probe.

[0007] Compared with the prior art, the present invention has the following technical effects: By optimizing the preparation conditions, a polymer probe or a probe array structure with a high specific surface area is obtained on the optical fiber. The nano-active material sol is coated on the surface of the optical fiber probe by using a coating technology to obtain a polymer probe with a surface-active functional monomer coating; when a pesticide molecule binds to the molecular recognition site, due to the existence of fluorescence energy transfer or photoinduced electron transfer effect, the optical signal of the fluorescent group will change, and then the high-sensitivity and rapid detection can be realized through the signal enhancement effect of the optical fiber.

[0008] Another objective of the present invention is to provide a method for detecting pesticide residues, which can more accurately detect the types and concentrations of pesticide residues.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for detecting pesticide residues, comprising the following steps: S100. Pretreating the agricultural products to be tested by using a SERS optical fiber probe; S200. Detecting the pretreated agricultural products to be tested by using an optical fiber Raman detection device to obtain their Raman spectra; S300. Sequentially passing the detected Raman spectra through a lock-in amplifier circuit and a programmable gain amplifier circuit to extract weak Raman signals of pesticide residues under a strong clutter background; S400. Processing the extracted Raman signals to achieve the characteristic recognition and quantitative detection of pesticide residues.

[0010] Compared with the prior art, the present invention has the following technical effects: By using advanced technologies such as surface-enhanced Raman spectroscopy, molecular imprinting, optical fiber probes, weak signal processing, and optical device integration, the rapid, quantitative analysis and detection of pesticide residues are realized. Compared with laboratory analysis methods such as liquid chromatography and gas chromatography, the detection time of this detection method is greatly shortened, the detection cost is also lower, and this detection method is easy to be miniaturized and is easy to promote and use. Description of the Drawings

[0011] Figure 1 It is a schematic diagram of the synthesis of a Au and Ag composite multi-layer shell Raman label;

[0012] Figure 2 It is a schematic diagram of the preparation of a polymer probe on an optical fiber;

[0013] Figure 3 It is a flowchart of the preparation of an optical fiber probe;

[0014] Figure 4 It is a schematic diagram of the structure of an optical fiber Raman detection device;

[0015] Figure 5 It is a schematic diagram of the structure of a lock-in amplifier circuit;

[0016] Figure 6 It is a schematic diagram of the structure of a programmable amplifier circuit. Specific implementation mode

[0017] The following combines Figures 1 to 6 to further describe the present invention in detail.

[0018] Referring to Figures 1 - 3 , a method for preparing a SERS optical fiber probe includes the following steps: A. Using nano-chemical synthesis technology to prepare SiO2 or polystyrene microspheres with controllable particle size to obtain a nanostructure; B. Coating the nanostructure with a single-hole polymer containing a pesticide molecular imprint through a selective polymerization reaction on the surface of the colloidal particles; C. Removing the pesticide molecular imprint and the nanostructure by a solvent method to obtain a single-hole hollow polymer microsphere; D. Marking Raman tags to positions adjacent to the imprinting sites through a surface modification method; E. Chemically modifying the polymer coating layer by a nano-surface modification method and assembling a functional fluorescent sensitive material with a nano-double-layer molecular imprint onto the polymer probe. By optimizing the preparation conditions, a polymer probe or a probe array structure with a high specific surface area is obtained on the optical fiber, and a polymer probe with a surface-active functional monomer coating is obtained by coating a nano-active material sol on the surface of the optical fiber probe using a coating technology; when a pesticide molecule binds to the molecular recognition site, due to the existence of fluorescence energy transfer or photoinduced electron transfer effects, the optical signal of the fluorescent group will change, and then through the signal enhancement effect of the optical fiber, highly sensitive and rapid detection can be achieved.

[0019] Using the above method, since there are two open surfaces, the effective number of binding sites for molecular imprinting is greatly increased, and moreover, the double-surface structure makes more imprinting sites closer to the surface, so the time required for diffusion and binding is shorter, improving the detection speed.

[0020] Referring to Figure 1, there are many kinds of structures of Raman tags and their preparation methods are diverse. In this embodiment, preferably, a multilayer shell structure Raman tag composed of Au and Ag is adopted. Specifically, in step D, the Raman tag is prepared according to the following steps: D1. Sodium borohydride is added to a mixed solution of sodium citrate and cobalt chloride, and a cobalt sol is obtained through reaction; D2. The cobalt sol is rapidly added to chloroauric acid solution to obtain Au nano-hollow microspheres; D3. Raman active molecules are added to achieve Raman labeling of the nano-probes on the outer side of the hollow microspheres; D4. Using the above particles as seeds, they are mixed with ascorbic acid solution, and under continuous stirring, a certain amount of silver nitrate solution is added drop by drop to obtain core / shell type Au@Ag bimetallic nano-hollow microspheres; D5. Repeat step D3 to continue Raman labeling of the hollow microspheres; D6. Then assemble Au and Ag nano-hollow shells in sequence to obtain a Raman tag with a multilayer shell structure.

[0021] Refer to Figure 2 , further, in this embodiment, preferably, in step E, the polymer probe is prepared according to the following steps: E1. Prepare a precursor solution, which is composed of three components: a photoinitiator, an amine co-reaction trigger, and a polymerizable monomer; E2. By controlling the polymerization conditions, a polymer probe is assembled at the end of the optical fiber. The length of the probe is 1 micron to 9 millimeters. Here, this step can also be extended and applied to the preparation of a polymer probe array for multi-core optical fibers. After preparing the probe, the surface area of the sensitive detection unit increases by several hundred to tens of thousands of times, effectively increasing the reaction area and improving the detection intensity of Raman spectroscopy; E3. Tetraethyl orthosilicate and 3-aminopropyltriethoxysilane are hydrolyzed under different conditions to form a viscous silica sol; E4. The silica sol is coated on the surface of the polymer probe by using a coating technique; E5. Through a further gelation process, a silica thin film is obtained. Since a large number of amino-functional monomers are contained on the surface of this thin film, various selective sensitive materials can be easily grafted. The analyte causes a change in the optical output signal by reacting or adsorbing with the sensitive film, so as to perform quantitative analysis on it.

[0022] The present invention also discloses a method for detecting pesticide residues by using the aforementioned SERS optical fiber probe, which includes the following steps: S100, pretreating the agricultural product to be detected by using the SERS optical fiber probe; S200, detecting the pretreated agricultural product to be detected by using an optical fiber Raman detection device to obtain its Raman spectrum; S300, sequentially passing the detected Raman spectrum through a phase-locked amplification circuit and a programmable amplification circuit to extract weak Raman signals of pesticide residues under a strong clutter background; S400, processing the extracted Raman signals to achieve characteristic recognition and quantitative detection of pesticide residues. Here, by using advanced technologies such as surface-enhanced Raman spectroscopy, molecular imprinting, optical fiber probes, weak signal processing, and optical device integration, rapid, quantitative analysis and detection of pesticide residues are realized. Compared with laboratory analysis methods such as liquid chromatography and gas chromatography, the detection time of this detection method is greatly shortened, the detection cost is also lower, and this detection method is easy to be miniaturized and is easy to promote and use.

[0023] Refer to Figure 4 , further, in the step S200, the optical fiber Raman detection device includes a semiconductor laser 10, a focusing lens 20, a first optical fiber 30, an optical fiber circulator 40, a second optical fiber 50, and a Raman spectrometer 60. The laser emitted from the semiconductor laser 10 is focused by the focusing lens 20 and then sequentially passes through the first optical fiber 30 and the optical fiber circulator 40 and is incident on the agricultural product to be detected. The reflected light sequentially passes through the optical fiber circulator 40 and the second optical fiber 50 and is incident on the Raman spectrometer 60. The Raman spectrometer 60 outputs the Raman spectrum curve of the received light. By optimizing the structure of the insertable optical fiber Raman spectrum detection probe, a surface-enhanced Raman signal detection system based on the structure of the optical fiber probe for Raman tags and active substrates is constructed. A multi-wavelength LED laser is coupled by using optical fiber coupling technology to form a multi-channel fusion selectable excitation light source system. The excitation light source, signal acquisition and other system units are integrated with the miniaturized Raman spectrometer 60, making the whole detection device small in structure and low in cost, and at the same time, it can reliably detect the solution to be detected.

[0024] The optical fiber circulator 40 here can also use a Y-shaped optical fiber. Specifically, in this embodiment, the optical fiber circulator 40 includes a first lens 41, a band-pass filter 42, a dichroic filter 43, a second lens 44, a low-pass filter 45, and a third lens 46. The light incident from the first lens 41 passes through the band-pass filter 42, the dichroic filter 43, and the second lens 44 in sequence and then enters the agricultural product to be measured. The reflected light passes through the second lens 44 and is then reflected by the dichroic filter 43 and passes through the low-pass filter 45 and the third lens 46 in sequence and enters the second optical fiber 50. Preferably, the optical axes of the first lens 41, the band-pass filter 42, and the second lens 44 coincide, the optical axes of the low-pass filter 45 and the third lens 46 coincide, and the optical axes of the first lens 41 and the third lens 46 are perpendicular to each other. The normal direction of the dichroic filter 43 forms an angle of 45 degrees with the optical axes of the first lens 41 and the third lens 46, and the center point of the dichroic filter 43 is located at the intersection of the optical axes of the first lens 41 and the third lens 46. The optical path with this structure is more concise.

[0025] Refer to Figure 5 and Figure 6 , Further, in the step S300, the lock-in amplifier circuit includes a signal channel, a reference channel, and a phase-sensitive detector. The input signals are processed by the signal channel and the reference channel respectively and then input into the phase-sensitive detector. The output end of the phase-sensitive detector is connected to the programmable gain amplifier circuit; the programmable gain amplifier circuit selectively amplifies the extracted weak signal to solve the automatic gain tuning of the signal. The lock-in amplifier circuit is a synchronous coherent detection circuit designed based on the correlation detection technology and using the principle of cross-correlation. By accurately locking the frequency of the target signal, it realizes the extraction of weak and useful signals under a complex background; at the same time, the programmable gain amplifier circuit can improve the reliability and accuracy of the collected data for the scattered non-uniformity of the signal and expand the dynamic measurement range of the sensor device.

[0026] Furthermore, the reference channel includes a trigger circuit, a phase shift circuit, and a square wave drive circuit. The output end of the trigger circuit is connected to the input end of the phase shift circuit, the output end of the phase shift circuit is connected to the input end of the square wave drive circuit, the output end of the square wave drive circuit is connected to one of the input ends of the phase-sensitive detector, and the output end of the phase-sensitive detector is filtered and output through a low-pass filter. The program-controlled amplifier circuit includes a pre-amplifier circuit, a post-amplifier circuit, an AGC automatic gain circuit, an ADUC812 single-chip microcomputer, and a peak detection circuit. The signal output from the phase-locked amplifier circuit is amplified by the pre-amplifier circuit and the post-amplifier circuit in turn and then output. The AGC automatic gain circuit is connected in series between the output end and the input end of the pre-amplifier circuit, and the peak detection circuit and the ADUC812 single-chip microcomputer are connected in series between the output end and the input end of the post-amplifier circuit. Through the above circuit design, the extraction of weak pesticide residue Raman signals under a strong clutter background can be easily realized.

[0027] Furthermore, a key control circuit is included, which is connected to the ADUC812 single-chip microcomputer for adjusting the gain range of the program-controlled amplifier circuit. After the key control circuit is set, the user can interact with the single-chip microcomputer through it, input some control instructions, or set the parameters of the single-chip microcomputer operation.

[0028] Furthermore, it includes a digital tube display circuit, which is connected to the ADUC812 single-chip microcomputer and is used to display the working status of the programmable amplifier circuit. The digital tube display circuit can provide display when the user interacts with the single-chip microcomputer, and can also provide status indication when the single-chip microcomputer is working.

[0029] The SERS spectrum of pesticide residues obtained by direct detection has a lot of noise, fluorescence background and interference information of other components, and the amount of SERS data is huge and redundancy is serious, which seriously affects the stability and reliability of the spectrum analysis model. In order to reliably parse the required information from the Raman spectrum, in this case, preferably, the step S400 includes the following steps: S410, the baseline drift and high-frequency noise of the collected Raman spectrum curve are preprocessed accordingly. After preprocessing, the dimension of the selected SERS data is still very high. If it is directly used for model training, there are disadvantages such as high computational complexity and long training time. Therefore, it is preferably included here that step S420 is used to reduce the dimension of the preprocessed spectrum data by principal component analysis. The dimension of the data after dimension reduction is low. When training the model, the computational complexity is low and the training time is short. At the same time, since the eigenvalues ​​with lower weights are removed, the accuracy of the trained model is still guaranteed. S430, the data after dimension reduction is substituted into the pesticide residue concentration prediction model to output the type and concentration information of pesticides in the agricultural products to be tested.

[0030] Specifically, in the step S410, the Raman spectrum is preprocessed by using the absolute value of the first derivative, multiplicative scatter correction, and standard normal variate in sequence. After the absolute value of the first derivative is processed, only the spectral transformation information is retained to remove the baseline drift. The multiplicative scatter correction and the standard normal variate eliminate the baseline drift and the fluorescence background caused by the size of solid particles, surface scattering, and the change of optical path. In the step S420, the following steps are included: S421, perform eigenvalue decomposition on the covariance matrix of the data set to obtain the corresponding eigenvalues and eigenvectors; S422, project the data set onto the eigenvector space to obtain its weights; S423, retain the components with a greater effect on the variance of the data set and delete the components with a smaller effect on the variance of the data set.

[0031] There are various ways to construct the pesticide residue concentration prediction model. Preferably, in the step S430 of this embodiment, the pesticide residue concentration prediction model is constructed according to the following steps: S431, use the known pesticide types and concentrations as samples, and process the samples according to S100 - S300, S410, and S421 - S423 in sequence to obtain the data after dimensionality reduction of the samples; S432, construct matrices X and Y, where matrix X is the matrix composed of the data after dimensionality reduction of the samples, and matrix Y is the matrix of pesticide types and concentrations; S433, after extracting the first components t1 and u1 from matrices X and Y respectively, perform partial least squares regression on the regression of X on t1 and the regression of Y on u1 respectively; S434, if the regression equation has reached a satisfactory accuracy, execute step S436, otherwise execute the next step; S435, use the residual information of X explained by t1 and the residual information of Y explained by u1 for the second round of component extraction, and so on until a satisfactory accuracy is reached and then execute the next step; S436, perform partial least squares regression on the regression of u (u1, u2,..., u q ) on t (t1, t2,..., t p ), and finally inverse transform it into the regression of Y on X. The regression model of matrix Y on X obtained is the pesticide residue concentration prediction model; in all steps, only the first sample selects the components with greater effects according to S421 - S423, and the remaining samples and the agricultural products to be tested directly perform dimensionality reduction on the data according to the components with greater effects determined for the first time. Using partial least squares regression to construct the pesticide residue concentration prediction model has a small amount of computation and accurate model construction.

[0032] Due to the wide variety of pesticide residues, the most representative and widely polluted malathion and fenitrothion are selected as experimental objects in this project. Typical agricultural products such as fruits, vegetables, and grains are chosen, and direct non-destructive Raman detection is carried out on their surfaces, and detection is carried out after pulverization pretreatment. The former is pretreated through the following steps: S111, uniformly coat the SERS optical fiber probe on the surface of agricultural products; S112, directly perform the next step after a certain kinetic reaction process. The latter is pretreated by the following steps: S121, pulverize the agricultural products to be measured; S122, mix and oscillate the pulverized agricultural products with a polar solvent; S123, filter the oscillated solution by suction to obtain a filtrate, concentrate it to dryness, and make up the volume with ultrapure water; S124, filter with a 0.45-micron membrane, mix the extract with the SERS optical fiber probe solution in a certain proportion, and adjust the pH to 1.0 - 5.0. For both of these two experimental objects, a Raman spectrometer with a laser wavelength of 785 nm needs to be selected, the laser power is set to 50 - 300 mw, and the scanning time is 2 - 20 s for laser scanning detection. A standard curve with pesticide concentration as the abscissa and characteristic peak height as the ordinate is plotted to analyze the pesticide content in the sample. This method provides a precise and simple Raman detection method, which can quickly, accurately, and quantitatively detect pesticide residues in agricultural product samples and can be used for rapid screening of a large number of samples on-site.

Claims

1. A preparation method of SERS optical fiber probe, characterized in that: It includes the following steps: A. Prepare SiO2 or polystyrene microspheres with controllable particle sizes by nano-chemical synthesis technology to obtain a nanostructure; B. Coat the nanostructure with a monolithic polymer containing pesticide molecular imprints through a selective polymerization reaction on the surface of colloidal particles; C. Remove the pesticide molecular imprints and the nanostructure by a solvent method to obtain monolithic hollow polymer microspheres; D. Mark Raman tags at positions adjacent to the imprint sites by surface modification methods; E. Chemically modify the polymer coating layer by nano-surface modification methods, and assemble the functional fluorescence-sensitive material with nano-double-layer molecular imprints onto the polymer probe.

2. The preparation method of the SERS optical fiber probe according to claim 1, wherein: In step D described above, the Raman tags are prepared according to the following steps: D1. Add sodium borohydride to a mixed solution of sodium citrate and cobalt chloride, and react to obtain cobalt sol; D2. Quickly add the cobalt sol to chloroauric acid solution to obtain Au nano-hollow microspheres; D3. Add Raman active molecules to achieve Raman labeling of the nano-probes on the outer side of the hollow microspheres; D4. Use the above particles as seeds, mix them with ascorbic acid solution, and dropwise add a certain amount of silver nitrate solution under continuous stirring to obtain core / shell type Au@Ag bimetallic nano-hollow microspheres; D5. Repeat step D3 to continue Raman labeling of the hollow microspheres; D6. Assemble Au and Ag nano-hollow shells in sequence to obtain Raman tags with a multi-layer shell structure.

3. The preparation method of the SERS optical fiber probe according to claim 2, wherein: In step E described above, the polymer probes are prepared according to the following steps: E1. Prepare a precursor solution, which is composed of three components: a photoinitiator, an amine co-reaction trigger, and a polymerizable monomer; E2. Assemble the polymer probes by controlling polymerization conditions, and the length of the probes is 1 μm to 9 mm; E3. Hydrolyze tetraethyl orthosilicate and 3-aminopropyltriethoxysilane under different conditions to form a viscous silica sol; E4. Coat the silica sol on the surface of the polymer probes by coating technology; E5. After a further gelation process, obtain a silica film.

4. A method for detecting pesticide residues using the SERS optical fiber probe described in claim 1, characterized in that: It includes the following steps: S100. Pretreat the agricultural products to be tested using a SERS optical fiber probe; S200. Detect the pretreated agricultural products to be tested using an optical fiber Raman detection device to obtain their Raman spectra; S300. Pass the detected Raman spectra through a lock-in amplifier circuit and a programmable gain amplifier circuit in sequence to extract weak Raman signals of pesticide residues under a strong clutter background; S400. Process the extracted Raman signals to achieve characteristic recognition and quantitative detection of pesticide residues.

5. The pesticide residue detection method according to claim 4, wherein: In step S200 described above, the optical fiber Raman detection device includes a semiconductor laser (10), a focusing lens (20), a first optical fiber (30), an optical fiber circulator (40), a second optical fiber (50), and a Raman spectrometer (60). The laser emitted from the semiconductor laser (10) is focused by the focusing lens (20) and then passes through the first optical fiber (30) and the optical fiber circulator (40) in sequence and is incident on the agricultural products to be tested. The reflected light passes through the optical fiber circulator (40) and the second optical fiber (50) in sequence and is incident on the Raman spectrometer (60). The Raman spectrometer (60) outputs the Raman spectral curve of the received light.

6. The pesticide residue detection method according to claim 5, characterized in that: In the said step S300, the lock-in amplifier circuit includes a signal channel, a reference channel, and a phase-sensitive detector. The input signals are respectively processed by the signal channel and the reference channel and then input into the phase-sensitive detector. The output end of the phase-sensitive detector is connected to a programmable gain amplifier circuit; the programmable gain amplifier circuit selectively amplifies the extracted weak signals to solve the automatic gain tuning of the signals.

7. The pesticide residue detection method according to claim 6, characterized in that: In the said step S400, the following steps are included: S410. Perform corresponding preprocessing on the baseline drift and high-frequency noise of the collected Raman spectrum curve; S420. Use principal component analysis to reduce the dimension of the preprocessed spectral data; S430. Substitute the data after dimension reduction into the pesticide residue concentration prediction model to output the information of the pesticide types and concentrations in the agricultural products to be tested.

8. The pesticide residue detection method according to claim 7, wherein: In the said step S410, the Raman spectrum is preprocessed by using the absolute value of the first derivative, multiplicative scatter correction, and standard normal variate in sequence. After the absolute value of the first derivative is processed, only the spectral transformation information is retained to remove the baseline drift. The multiplicative scatter correction and the standard normal variate eliminate the baseline drift and fluorescence background caused by the size of solid particles, surface scattering, and optical path changes; In step S420, the following steps are included: S421. Perform eigenvalue decomposition on the covariance matrix of the data set to obtain the corresponding eigenvalues and eigenvectors; S422. Project the data set onto the eigenvector space to obtain its weights; S423. Retain the components that have a greater effect on the variance of the data set and delete the components that have a smaller effect on the variance of the data set.

9. The pesticide residue detection method according to claim 8, wherein: In the said step S430, the pesticide residue concentration prediction model is constructed according to the following steps: S431. Take the known pesticide types and concentrations as samples, and process the samples in sequence according to S100 - S300, S410, S421 - S423 to obtain the data after dimension reduction of the samples; S432. Construct matrices X and Y, where matrix X is a matrix composed of the data after dimension reduction of the samples, and matrix Y is a matrix of pesticide types and concentrations; S433. After extracting the first components t1 and u1 from matrices X and Y respectively, partial least squares regression is respectively performed on the regression of X on t1 and the regression of Y on u1; S434. If the regression equation has reached a satisfactory accuracy, execute step S436, otherwise execute the next step; S435. Use the residual information of X explained by t1 and the residual information of Y explained by u1 to perform the second round of component extraction, and so on, until a satisfactory accuracy is reached and then execute the next step; S436, Partial least squares regression is implemented for u (u1, u2,..., u q ) with respect to t (t1, t2,..., t p ), and finally inverse-transformed into the regression of Y with respect to X. The regression model of matrix Y with respect to X thus obtained is the pesticide residue concentration prediction model; In all steps, only the first sample selects the components with greater effects according to S421 - S423, and the remaining samples and the agricultural products to be tested directly reduce the dimension of the data according to the larger components determined for the first time.

10. The pesticide residue detection method according to claim 9, characterized in that: The said step S100 includes the following steps: S111. Uniformly coat the SERS optical fiber probe on the surface of the agricultural product; S112. After a certain kinetic reaction process, directly execute the next step; Or step S100 includes the following steps: S121. Crush the agricultural product to be tested; S122. Mix the crushed agricultural product with a polar solvent and oscillate; S123. Filter the oscillated solution to obtain a filtrate, concentrate it to dryness, and make a constant volume with ultrapure water; S124. Filter with a 0.45-micron membrane, mix the extract with the SERS optical fiber probe solution in a certain proportion, and adjust the pH to 1.0 - 5.0.

Images