Method for rapidly detecting trace antibiotics in water

Abstract

The invention discloses a method for rapidly detecting trace antibiotics in water, which solves the problems of tedious detection process and long consumed time by combining preparation of a surface enhanced Raman spectrum and a deep learning algorithm. On the other hand, Raman spectrum libraries of different antibiotics are established through deep learning, newly collected fingerprints are directly imported into the libraries for judgment, and the accuracy of identification results can be better guaranteed through comparison and combination of a deep learning method and evaluation indexes.

Description

Technical Field

[0001] The present invention relates to the field of Raman spectroscopy detection, and specifically to a method for rapidly detecting different antibiotics in water using Raman spectroscopy, analyzing the distribution of characteristic peaks in the spectrum, comparing the performance of deep learning models in detecting antibiotics in water, and using a deep learning model combined with a non-negative elastic network to predict the proportion of elements in mixed antibiotics. Background Art

[0002] Water is an essential resource for human life. However, with the development of modern agriculture and animal husbandry, chemicals such as pesticides and antibiotics have become widely used. These substances can enter water bodies and pose a potential threat to human health. Antibiotic residues are of particular concern. The widespread use of antibiotics has led to increased concentrations in water, potentially contributing to the development and spread of drug resistance. Currently, commonly used methods for detecting antibiotics in water include high-performance liquid chromatography (HPLC), liquid chromatography-tandem mass spectrometry (LC-MS / MS), spectroscopic techniques, and biosensors. However, these methods suffer from complex operations, the need for specialized personnel, lengthy analysis times, and expensive equipment, limiting their widespread adoption in practical applications. In recent years, surface-enhanced Raman spectroscopy (SERS) combined with deep learning (DL) algorithms has been widely used for the rapid detection of antibiotics in water. SERS technology offers advantages such as high sensitivity and rapid analysis, enabling highly sensitive detection of antibiotics in water. Furthermore, deep learning algorithms can automatically process and analyze SERS signals, improving detection accuracy and speed. This combined approach provides a new approach for predicting antibiotic concentrations in water. By collecting and processing SERS signals and then using deep learning algorithms to establish a prediction model, quantitative analysis of antibiotics in water can be achieved. Summary of the Invention

[0003] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and provide a method for rapidly detecting trace antibiotics in water. The method solves the problem of cumbersome and time-consuming detection process by combining surface-enhanced Raman spectroscopy with deep learning algorithms.

[0004] The technical solution adopted in the present invention is as follows:

[0005] A method for rapidly detecting trace amounts of antibiotics in water comprises the following steps:

[0006] (1) Preparation of standard solutions: Weigh the standard products of doxycycline, ciprofloxacin, and levofloxacin, add them to water, and mix them evenly to obtain standard solutions of doxycycline, ciprofloxacin, and levofloxacin. Prepare standard solutions of doxycycline, ciprofloxacin, and levofloxacin of different concentrations, respectively, and set aside.

[0007] (2) Sample preparation: adding sodium citrate solution to a silver nitrate solution heated to boiling to react with stirring, centrifuging the resulting reaction solution, and resuspending the resulting precipitate in deionized water to obtain a Raman enhancement substrate with negatively charged nanosilver particles. The three antibiotic standard solution samples of different concentrations obtained in step (1) are mixed with the Raman substrate solution to obtain high-sensitivity Raman signal detection samples of different concentrations;

[0008] (3) SERS detection: The antibiotic SERS samples of different concentrations obtained in step (2) are titrated onto a silicon wafer and air-dried, and Raman spectral sampling is performed multiple times and at multiple locations to obtain the corresponding Raman spectra of the antibiotic samples of different concentrations, thereby constructing a Raman spectral database of antibiotics of different concentrations;

[0009] (4) Detection limit selection: Calculate the signal intensity of the Raman spectra of the antibiotic samples with different concentrations obtained in step (3) to determine the minimum concentration signal detection limit;

[0010] (5) Mixed sample preparation: The three antibiotic samples with the lowest concentration detection limit determined in step (4) are mixed and ultrasonically treated, and then mixed with the high-sensitivity Raman signal detection sample prepared in step (2) after ultrasonic treatment to obtain a SERS detection sample of the mixed antibiotic sample;

[0011] (6) Data clustering: cluster analysis is performed on the SERS detection samples of the mixed antibiotic sample obtained in step (5), and the sample category is determined by comparing the differences between different antibiotics to form a Raman vector classification quadrant. The Raman spectra of different antibiotics are distinguished by observing the clustering results of the spectral sample points in the classification quadrant;

[0012] (7) Data enhancement: The three antibiotic samples with the lowest concentration detection limit determined in step (4) were averaged and their spectra were added in random proportions. Random Gaussian noise was further added to obtain their enhanced Raman spectra, generating sufficient data to train the deep neural network and improve the generalization ability of the model.

[0013] (8) Data identification: Use a deep learning algorithm to fit the Raman spectra after data enhancement in step (7). The Raman spectrum data set of all antibiotic samples is divided into a training set, a validation set, and a test set by random sampling. Different hyperparameter combinations are adjusted to find the optimal model parameter combination;

[0014] (9) Identification of mixed antibiotics: Convolutional neural networks and non-negative elastic nets are combined to identify the ratio of antibiotics in mixed antibiotics.

[0015] In the present invention, in step (1), the standard solutions of three antibiotics, doxycycline, ciprofloxacin and levofloxacin, are prepared as follows:

[0016] 9.618 mg of doxycycline, 6.6268 mg of ciprofloxacin, and 7.2274 mg of levofloxacin powder were added to water to prepare three antibiotic single substance stock solutions with a final concentration of 1×10-4 M. Subsequently, a stepwise dilution method was used to prepare antibiotic standard solutions with concentrations ranging from 1×10-4 M to 1×10-8 M for later use.

[0017] Preferably, the three antibiotic single substance stock solutions are prepared into 1×10 -4 M, 1×10 -5 M, 1×10 -6 M, 1×10 -7 M, 1×10 -8 M concentration solution, set aside.

[0018] In the present invention, in step (2), when preparing a high-sensitivity Raman signal detection sample for three antibiotics, silver nitrate is dissolved in water and stirred and heated to boiling, wherein the concentration of silver nitrate in the silver nitrate solution is 0.5 to 1.5 mol / L, preferably 1 mol / L. The reducing agent is sodium citrate, and the sodium citrate is prepared into a sodium citrate solution with a concentration of 0.5 to 1.5 wt%, preferably 1 wt%.

[0019] In a preferred embodiment, the method for preparing samples for high-sensitivity Raman signal detection of three antibiotics is as follows:

[0020] (a) heating a 1 mol / L silver nitrate solution to boiling, adding 8 mL of a 1 wt% sodium citrate solution while stirring, and stirring at 600-800 rpm for 30-60 min to obtain a negatively charged silver nanoparticle solution; then centrifuging at 6000-8000 rpm for 5-10 min, discarding the supernatant, and resuspending the resulting precipitate in deionized water to obtain a negatively charged silver nanoparticle solution;

[0021] (b) 10 to 20 μL of the negatively charged silver nanoparticle solution obtained in step (a) was mixed with an equal amount of standard antibiotic solutions of varying concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then dried naturally in a safety cabinet for testing.

[0022] In a more preferred embodiment, the method for preparing samples for high-sensitivity Raman signal detection of three antibiotics is as follows:

[0023] (a) A 1 mol / L silver nitrate solution was heated to boiling. 8 mL of a 1 wt% sodium citrate solution was added while stirring. The mixture was stirred at 650 rpm for 40 min to obtain a negatively charged silver nanoparticle solution. The solution was then centrifuged at 7000 rpm for 7 min. The supernatant was discarded and the resulting precipitate was resuspended in 100 μL of deionized water to obtain a negatively charged silver nanoparticle solution, which served as the Raman enhancement substrate.

[0024] (b) 15 μL of the Raman enhancement substrate solution obtained in step (a) was mixed with an equal amount of antibiotic standard solutions of different concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then naturally dried in a safety cabinet for detection.

[0025] For the present invention, in step (3), SERS detection is suitable for collecting surface enhanced Raman signals of the sample to be tested, including mixing the solution samples of different concentrations of antibiotics obtained in step (2) with a high-sensitivity Raman signal detection substrate, dripping appropriate amounts onto a silicon wafer multiple times, and naturally drying in a safety cabinet for detection; thereby detecting and obtaining a Raman spectrum fingerprint.

[0026] In the present invention, in step (3), the multi-site Raman spectrum sampling is 80 to 120 sites; SERS detection uses Anton Paar TM Cora100 handheld Raman spectrometer, sampling parameters are: Raman spectrum excitation wavelength of 785nm, excitation power of 25mW, spectral resolution of 1nm, spectral wave number resolution of 10cm-1, detection spectrum range of 400-2300cm-1. Preferably, the detection spectrum range is 530-1800cm-1. -1 .

[0027] In the present invention, in step (4), the average Raman spectra of antibiotic standard solutions of different concentrations are calculated, the differences in Raman spectral signal intensities at different concentrations are compared, and the minimum concentration detection limit is determined to be 1×10-7M based on the distribution of characteristic peak differences.

[0028] In the present invention, in step (5), the mixed sample preparation process is specifically as follows:

[0029] (a) After determining the minimum detection limit, the three antibiotic solutions were mixed in pairs at the minimum detection limit and all three antibiotics were mixed together, and sonicated for 10 min to obtain a homogenous sample.

[0030] (b) 10-20 μL of the mixed antibiotic solution sample prepared in step (a) was mixed with an equal amount of antibiotic standard solutions of varying concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then dried naturally in a safety cabinet for testing.

[0031] In a preferred embodiment, the mixed sample preparation process is as follows:

[0032] (a) After determining the minimum concentration detection limit, 0.5×10 -7 M doxycycline, 0.5×10 -7 M ciprofloxacin and 0.5×10 -7 M levofloxacin solutions were mixed in pairs, and then mixed with the three antibiotics and sonicated for 10 min to obtain a homogenous sample;

[0033] (b) 15 μL of the mixed antibiotic solution sample prepared in step (a) was mixed with an equal amount of antibiotic standard solutions of varying concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then dried naturally in a safety cabinet for testing.

[0034] In the present invention, in step (6), principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) operations are performed on the obtained SERS detection sample of the mixed antibiotic sample. The specific operation steps are as follows:

[0035] (a) All seven antibiotic spectral data after preprocessing were input into the PCA unsupervised learning algorithm model. Dimensionality reduction and clustering visualization were performed on the spectral data. The principal components PCA1 to PCA7 after dimension reduction were selected as principal components. The clustering results of the Raman spectral principal component sample points in the classification quadrant were observed.

[0036] (b) All seven preprocessed antibiotic spectral data were input into the OPLS-DA supervised learning algorithm model, and the clustering results were evaluated using the three indicators R2X, R2Y and Q2. The larger the indicator, the better the model performance.

[0037] In the present invention, in step (7), different pure spectra are added to the input spectral data of three pure antibiotics by random proportions, and random Gaussian noise is further added to obtain enhanced Raman spectra. The enhanced spectra are of two types: positive and negative. The positive spectrum is a random combination of the spectra of other interfering compounds and the spectrum of the selected compound to enhance the positive spectrum data, and the proportion of the selected compound in each enhanced spectrum is set to no less than 10%. The negative spectrum is a random selection of spectra outside the spectrum of the selected compound and a random generation of their ratios.

[0038] Preferably, in step (7), data augmentation can generate sufficient frequency spectrum to train the deep neural network, thereby improving the generalization ability of the model.

[0039] Preferably, the maximum number of components used for data enhancement is set to 2, and the noise rate is set to 0.005. Based on these parameters, each pure antibiotic spectrum is enhanced to 30,000.

[0040] For the present invention, in step (8), a total of three pure antibiotic identification models are generated, each deep learning model is a 1-classification model, the deep learning algorithm is a convolutional neural network (CNN), and the deep learning process includes a computer memory, a computer processor, and a computer program stored in the computer memory and executable on the computer processor, the computer memory stores a model optimization parameter range, and the computer processor implements the following steps when executing the computer program:

[0041] a) dividing all SERS Raman signals after data preprocessing into different data subsets;

[0042] b) Pre-set the deep learning training parameter range, fit the model, and select the optimal parameter combination;

[0043] c) using the trained deep learning model to classify and predict the SERS spectra of the mixed antibiotics to obtain a model file;

[0044] Step a) involves randomly partitioning a pre-built antibiotic SERS spectral database into a training set, a validation set, and a test set in an 8:1:1 ratio. The training set and validation set are used to train and validate the model, respectively. The test data is not used in model training or validation; it is input into the training model file as unfamiliar data to test the model's ability to discriminate against unknown data types.

[0045] Step b) includes: data classification and prediction using a model parameter combination adjusted by hyperparameters to perform model training, selecting the optimal model by comparing the model loss on the validation set, and classifying and labeling the antibiotic spectral data using the trained optimal classifier and storing it in a database.

[0046] In the present invention, in step (9), two antibiotics and three antibiotics are mixed in different proportions to obtain a mixed antibiotic SERS sample.

[0047] Preferably, doxycycline, ciprofloxacin and levofloxacin solutions are mixed in pairs, the two antibiotics are mixed in a ratio of 1:9, 2:8, 3:7, 4:6, 5:5, 6:4, 7:3, 8:2 and 9:1, and the three antibiotics are mixed in a ratio of 1:3:6, 1:8:1, 1:1:8, 2:4:4, 2:6:2, 3:1:6, 3:3:4, 4:5:1, 4:1:5, 6:2:2 and 8:1:1, and 15 μL of the prepared mixed solution sample is mixed with an equal amount of the Raman enhanced substrate solution in step (2), and ultrasonicated for 10 minutes to obtain a uniform antibiotic mixed solution sample.

[0048] Among them, a convolutional neural network is used to predict the presence of single-element antibiotics, and the data is processed with optional preprocessing steps - baseline correction and spectral smoothing. The non-negative elastic network algorithm is used to determine the relative proportion of single antibiotics in the mixture of antibiotics. The specific non-negative elastic network algorithm is expressed as follows:

[0049]

[0050] The beneficial effects of the present invention are:

[0051] (1) The present invention uses deep learning combined with SERS method to identify three antibiotic elements and a mixture of three antibiotic elements in water, providing a rapid identification and classification method for antibiotics in water, providing a fast and effective analysis approach for the application of Raman spectrometers, and reflecting the application advantages of Raman spectrometers.

[0052] (2) The present invention establishes a Raman spectrum library of different antibiotics through deep learning, and directly imports the newly collected fingerprint spectrum into the library for judgment. The comparison and combination of deep learning methods and evaluation indicators can better ensure the accuracy of identification results. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 Establishing an analysis process combining deep learning algorithm for the Raman spectroscopy fingerprint of antibiotics in the present invention;

[0054] Figure 2 This is a linear relationship diagram between the lowest concentration detection limit and the concentration gradient of antibiotics in the present invention;

[0055] Figure 3 The average Raman spectrum of the mixed antibiotics in the present invention;

[0056] Figure 4 This is a clustering result diagram of SERS sample points of different antibiotics obtained by PCA and OPLS-DA data cluster analysis before and after data normalization in the present invention;

[0057] Figure 5Visualization of the prediction results using CNN and NEN for two antibiotics and a mixture of three antibiotics;

[0058] Figure 6 This is a graph showing the prediction results of the model when two antibiotics in different proportions pass through the model in the present invention;

[0059] Figure 7 This is a diagram showing the prediction results of the model when different proportions of the three antibiotics in the present invention are passed. DETAILED DESCRIPTION

[0060] Raman spectroscopy is based on the principle of inelastic scattering. That is, when incident light from a laser light source illuminates a substance, it is scattered by the molecules of the substance. A very small portion of the scattered light has a frequency different from the incident light. The change in the scattered light frequency depends on the structural characteristics of the illuminated substance. Different substances produce scattered light of specific frequencies under the same laser irradiation. Therefore, Raman spectroscopy can be used to achieve fast, simple, repeatable and non-destructive detection of material composition.

[0061] Artificial intelligence technology provides an efficient and accurate implementation for Raman spectroscopy-based material composition detection. Existing Raman spectroscopy deep learning algorithms are oriented toward specific substances to be tested. They transform the material identification problem of Raman spectroscopy into a deep learning classification problem. Deep learning models are trained based on standard Raman spectra of known substances, and the trained models are used to accurately identify test samples.

[0062] This specification and claims do not use differences in names as a way to distinguish components, but use differences in components' functions as the criteria for distinction. As mentioned in the description and claims throughout the text, "including" is an open-ended term and should be interpreted as "including but not limited to". "Approximately" means that within an acceptable error range, those skilled in the art can solve the technical problem within a certain error range and basically achieve the technical effect. In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention is further described in detail below in conjunction with the embodiments so that those skilled in the art can implement it with reference to the text of the specification. The equipment or raw materials used in the embodiments can all be obtained from the market.

[0063] instrument

[0064] Raman spectrometer (Anton Paar TM The models and manufacturers of the instruments used in the implementation case are detailed in Table 1.

[0065] Table 1 Experimental instrument information

[0066]

[0067] Drugs and reagents

[0068] Silver nitrate (Sinopharm Chemical Reagent), sodium citrate (Sinopharm Chemical Reagent), sodium chloride (Sinopharm Chemical Reagent), doxycycline (Shanghai McLean), ciprofloxacin (Shanghai McLean), and levofloxacin (Shanghai McLean).

[0069] Specific embodiment: Rapid detection of trace antibiotics in water using the method of the present invention

[0070] 1. Method

[0071] (1) Prepare standard solution: add 9.618 mg of doxycycline, 6.6268 mg of ciprofloxacin and 7.2274 mg of levofloxacin powder to water to prepare a final concentration of 1×10 -4 M of the three antibiotics; then, the concentration range of the prepared -4 M~1×10 -8 M antibiotic standard solution for later use.

[0072] (2) Sample preparation:

[0073] The sample preparation method for high-sensitivity Raman signal detection of three antibiotics is as follows:

[0074] (a) 33.72 mg of silver nitrate was weighed and dissolved in 200 ml of deionized water to obtain a 1 mol / L silver nitrate solution. The solution was heated to boiling using a magnetic stirrer. 8 mL of a 1 wt% sodium citrate solution was added at once while stirring. The solution was stirred at 650 r / min for 40 min to obtain a negatively charged silver nanoparticle reaction solution, wherein the silver nanoparticles had a diameter of less than 11 nm. 1 mL of the resulting silver nanoparticle reaction solution was then centrifuged at 7000 r / min for 7 min. The supernatant was discarded and the resulting precipitate was resuspended in 100 μL of deionized water to obtain a negatively charged silver nanoparticle solution, which was the Raman enhancement substrate solution. The resulting Raman enhancement substrate solution was uniformly milky white and was stored in the dark at room temperature until use.

[0075] (b) 15 μL of the Raman enhancement substrate solution obtained in step (a) was mixed with an equal amount of antibiotic standard solutions of different concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then naturally dried in a safety cabinet for detection.

[0076] (3) SERS detection: The antibiotic SERS samples of different concentrations obtained in step (2) were titrated onto a silicon wafer and dried, and Raman spectral sampling was performed multiple times and at multiple locations to obtain the corresponding Raman spectra of the antibiotic samples of different concentrations, thereby constructing a Raman spectral database of antibiotics of different concentrations. The parameters of Raman spectral sampling are: Raman spectral excitation wavelength of 785nm, excitation power of 25mW, spectral resolution of 1nm, and spectral wavenumber resolution of 10cm -1 , the detection spectrum range is 400-2300cm -1 The actual spectral range is 530-1800cm -1 .

[0077] (4) Detection limit selection: Calculate the average Raman spectra of antibiotic standard solutions with different concentrations, compare the differences in Raman spectral signal intensities at different concentrations, and determine the minimum concentration detection limit as 1×10 -7 M.

[0078] (5) Preparation of mixed samples: After determining the minimum concentration detection limit, the three antibiotic solutions were mixed in pairs and together at the minimum concentration detection limit, and ultrasonicated for 10 min to obtain a uniform sample. 15 μL of the prepared mixed solution sample was taken and mixed evenly with 15 μL of negatively charged nanosilver particle substrate solution. The mixed sample was titrated on a clean silicon wafer to form circular spots, and then naturally dried in a safety cabinet for SERS signal detection.

[0079] (6) Data clustering:

[0080] The specific steps are as follows:

[0081] (a) All seven antibiotic spectral data after preprocessing were input into the PCA unsupervised learning algorithm model, and the spectral data were subjected to dimensionality reduction clustering visualization. The principal components PCA1 and PCA2 were selected to determine the classification quadrants, and the clustering of spectral sample points in the classification quadrants was observed.

[0082] (b) The spectral data of all seven preprocessed antibiotics were input into the OPLS-DA supervised learning algorithm model. By comparing the distribution differences of the spectral sample points of different antibiotics, the classification box plots of the Raman vectors of different antibiotics were formed. The clustering results were evaluated using the three indicators of R2X, R2Y and Q2.

[0083] (7) Data augmentation: The average spectra of the three antibiotic samples with the lowest detection limits determined in step (4) were taken, and different pure spectra were added in random proportions. Random Gaussian noise was further added to obtain their enhanced Raman spectra. 30,000 spectral data were generated to train the deep neural network and improve the generalization ability of the model. The maximum number of components used for data augmentation was set to 2, and the noise rate was 0.005.

[0084] (8) Data Identification: A deep learning algorithm is used to fit the Raman spectra after data enhancement in step (7). The Raman spectral dataset of all antibiotic samples is randomly divided into a training set, a validation set, and a test set, and a classifier is trained. The sample data is classified and labeled using the trained classifier and then stored in the database. The deep learning parameter settings are detailed in Table 2 below.

[0085] Table 2 Deep learning network parameter settings

[0086]

[0087] (9) Identification of mixed antibiotics: Mix two or three antibiotics in different proportions. Use a convolutional neural network to predict the presence of single antibiotics. Optional preprocessing steps, such as baseline correction and spectral smoothing, are used to process the data. The non-negative elastic network algorithm is used to determine the relative proportions of the single antibiotics in the mixed antibiotics. The specific expression of the non-negative elastic network algorithm is:

[0088]

[0089] The lasso ratio of the non-negative elastic network parameters used to predict the proportion of single substances in mixed antibiotics was set to 0.96 and the maximum iterator was set to 1.

[0090] 2. Results

[0091] Figure 2 The concentration gradient and linear relationship diagrams of three simple antibiotics show that as the concentration decreases, the spectral intensity decreases linearly, and at 1×10 -8 At the concentration of M, the characteristic peaks of the antibiotic spectrum could not be well displayed in the full Raman spectrum, and the minimum concentration detection limit was determined to be 1×10 -7 M.

[0092] Figure 3 The average Raman spectra and corresponding characteristic peaks of the mixture of three antibiotics and the mixture of three antibiotics in pairs.

[0093] Figure 4The PCA algorithm was used to classify the SERS spectral data of three single antibiotics and a mixture of four antibiotics, but some spectral data points overlapped, and the OPLS-DA algorithm was unable to effectively classify the different SERS datasets into distinct clusters. After data normalization, both the PCA and OPLS-DA models were able to distinguish more antibiotic types, but they still could not accurately distinguish different antibiotics.

[0094] Figure 5 Visualizations of the prediction results for a mixture of two and three antibiotics using a CNN and a non-negative elastic network. (A-C) shows a two-antibiotic mixture, with the black line indicating a 10% absolute error margin. (D) shows a mixture of three antibiotics. The same colored graphs represent a set of experimental data, with circles representing actual proportions and triangles representing model predictions. This demonstrates that the CNN and non-negative elastic network can accurately predict the proportions of the individual antibiotics in the mixture.

[0095] Figure 6 The following figure shows the prediction results of the model for different ratios of two antibiotics. The prediction results of the mixed antibiotic ratio using different preprocessing methods are compared, where BC represents baseline correction and SS represents spectral smoothing.

[0096] Figure 7 The following figure shows the prediction results of the model for different proportions of three antibiotics. The prediction results of the mixed antibiotic ratio using different preprocessing methods are compared, where BC represents baseline correction and SS represents spectral smoothing.

[0097] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for rapid detection of trace antibiotics in water, characterized in that: It includes the following steps: (1) Preparation of standard solutions: Weigh the standard products of doxycycline, ciprofloxacin, and levofloxacin, add them to water, and mix them evenly to obtain standard solutions of doxycycline, ciprofloxacin, and levofloxacin. Prepare standard solutions of doxycycline, ciprofloxacin, and levofloxacin of different concentrations, respectively, and set aside. (2) Sample preparation: adding a sodium citrate solution to a silver nitrate solution heated to boiling to react with stirring, centrifuging the resulting reaction solution, and resuspending the resulting precipitate in deionized water to obtain a Raman enhancement substrate with negatively charged nanosilver particles. The three antibiotic standard solution samples of different concentrations obtained in step (1) are mixed with the Raman substrate solution to obtain high-sensitivity Raman signal (SERS) detection samples of different concentrations; (3) SERS detection: The antibiotic SERS detection samples of different concentrations obtained in step (2) are titrated onto a silicon wafer and air-dried, and Raman spectral sampling is performed multiple times and at multiple locations to obtain the corresponding Raman spectra of the antibiotic samples of different concentrations, thereby constructing a Raman spectral database of antibiotics of different concentrations; (4) Detection limit selection: Calculate the signal intensity of the Raman spectra of the antibiotic samples with different concentrations obtained in step (3) to determine the minimum concentration signal detection limit; (5) Mixed sample preparation: The three antibiotic samples with the lowest concentration detection limit determined in step (4) are mixed and ultrasonically treated, and then mixed with the high-sensitivity Raman signal detection sample prepared in step (2) after ultrasonic treatment to obtain a SERS detection sample of the mixed antibiotic sample; (6) Data clustering: cluster analysis is performed on the SERS detection samples of the mixed antibiotic sample obtained in step (5), and the sample category is determined by comparing the differences between different antibiotics to form a Raman vector classification quadrant. The Raman spectra of different antibiotics are distinguished by observing the clustering results of the spectral sample points in the classification quadrant; (7) Data enhancement: The three antibiotic samples with the lowest concentration detection limit determined in step (4) were averaged and their spectra were added in random proportions. Random Gaussian noise was further added to obtain their enhanced Raman spectra, generating sufficient data to train the deep neural network and improve the generalization ability of the model. (8) Data identification: Use a deep learning algorithm to fit the Raman spectra after data enhancement in step (7). The Raman spectrum data set of all antibiotic samples is divided into a training set, a validation set, and a test set by random sampling. Different hyperparameter combinations are adjusted to find the optimal model parameter combination; (9) Identification of mixed antibiotics: Convolutional neural networks and non-negative elastic nets are combined to identify the ratio of antibiotics in mixed antibiotics.

2. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (1), the standard solutions of the three antibiotics, doxycycline, ciprofloxacin and levofloxacin, are prepared as follows: 9.618 mg of doxycycline, 6.6268 mg of ciprofloxacin, and 7.2274 mg of levofloxacin powder were added to water to prepare a final concentration of 1×10 -4 M of the three antibiotics; then, the concentration range of the prepared -4 M~1×10 -8 M antibiotic standard solution for later use.

3. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (2), the method for preparing samples for high-sensitivity Raman signal detection of the three antibiotics is as follows: (a) heating a 1 mol / L silver nitrate solution to boiling, adding 8 mL of a 1 wt% sodium citrate solution while stirring, and stirring at 600-800 rpm for 30-60 min to obtain a negatively charged silver nanoparticle solution; then centrifuging at 6000-8000 rpm for 5-10 min, discarding the supernatant, and resuspending the resulting precipitate in deionized water to obtain a negatively charged silver nanoparticle solution; (b) 10 to 20 μL of the negatively charged silver nanoparticle solution obtained in step (a) was mixed with an equal amount of standard antibiotic solutions of varying concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then dried naturally in a safety cabinet for testing.

4. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (3), the number of sites in the multi-site Raman spectrum sampling is 80 to 120 sites; the parameter conditions of the Raman spectrum sampling are: the excitation wavelength of the Raman spectrum is 785nm, the excitation power is 25mW, the spectral resolution is 1nm, and the spectral wavenumber resolution is 10cm -1 , the detection spectrum range is 400-2300cm -1 .

5. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (4), the average Raman spectra of antibiotic standard solutions at different concentrations are calculated, and the differences in Raman spectral signal intensities at different concentrations are compared. Based on the distribution of characteristic peak differences, the minimum concentration detection limit is determined to be 1×10-7M.

6. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (5), the mixed sample preparation process is as follows: (a) After determining the minimum detection limit, the three antibiotic solutions were mixed in pairs at the minimum detection limit and all three antibiotics were mixed together, and sonicated for 10 min to obtain a homogenous sample. (b) 10-20 μL of the mixed antibiotic solution sample prepared in step (a) was mixed with an equal amount of antibiotic standard solutions of varying concentrations. The mixed sample was dropped onto a clean silicon wafer to form circular spots, which were then dried naturally in a safety cabinet for testing.

7. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (6), principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA) operations are performed on the obtained SERS detection sample of the mixed antibiotic sample. The specific operation steps are as follows: (a) All seven antibiotic spectral data after preprocessing were input into the PCA unsupervised learning algorithm model. Dimensionality reduction and clustering visualization were performed on the spectral data. The principal components PCA1 to PCA7 after dimension reduction were selected as principal components. The clustering results of the Raman spectral principal component sample points in the classification quadrant were observed. (b) All seven preprocessed antibiotic spectral data were input into the OPLS-DA supervised learning algorithm model, and the clustering results were evaluated using the three indicators R2X, R2Y and Q2. The larger the indicator, the better the model performance.

8. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (7), for the input three pure antibiotic spectral data, different pure spectra are added in random proportions, and random Gaussian noise is further added to obtain their enhanced Raman spectra; There are two types of enhanced spectra: positive and negative. A positive spectrum is a random combination of spectra of other interfering compounds and the spectrum of the selected compound to enhance the positive spectrum data. The proportion of the selected compound in each enhanced spectrum is set to no less than 10%. Negative spectrum randomly selects spectra from outside the spectrum of the selected compound and randomly generates their ratio.

9. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (8), the deep learning algorithm is a convolutional neural network (CNN); the Raman spectroscopy data set is divided into a training set, a validation set, and a test set in an 8:1:1 ratio; the training set and the validation set are used for fitting and verification, and the test set data is only used to test the model performance, and the optimal model parameter combination is found by adjusting different hyperparameter combinations.

10. The method for rapid detection of trace antibiotics in water according to claim 1, characterized in that: In step (9), the three antibiotics are mixed in different proportions, two antibiotics and three antibiotics are mixed, and a convolutional neural network is used to predict the presence of single antibiotics. The data is processed using optional preprocessing steps - baseline correction and spectral smoothing. The non-negative elastic network algorithm is used to determine the relative proportions of the single antibiotics in the mixed antibiotics. The non-negative elastic network algorithm is specifically expressed as:

Images