Veterinary drug residue intelligent detection system and method based on multi-dimensional spectral characteristics

Abstract

The invention provides an intelligent veterinary drug residue detection system and method based on multi-dimensional spectral characteristics, and relates to the technical field of veterinary drug residue detection, spectral response data of a to-be-detected sample solution in a visible light band and a near-infrared band is collected, and the residue discrimination degree of veterinary drug residues under any wavelength when the concentration dependence change exists is determined; abnormal wave band combinations with quantitative response to veterinary drug residues are screened out from the visible light wave band and the near-infrared wave band respectively, the residual peak deviation of the veterinary drug residues in the region of interest of the spectrum is determined according to the abnormal wave band combinations, and trace residue marks in the multiple spectrum intensity distribution modes are determined; and performing in-situ nondestructive screening on veterinary drug residues in the to-be-detected animal-derived food sample according to the residue discrimination and the trace residue mark. According to the method, on the premise that the dual-band spectrum collaborative response is enhanced, the veterinary drug residue characteristics can be directionally distinguished and detected, so that the veterinary drug residue screening accuracy is improved.

Description

Technical Field

[0001] This application relates to the field of veterinary drug residue detection technology, and more specifically, to an intelligent veterinary drug residue detection system and method based on multidimensional spectral characteristics. Background Technology

[0002] Veterinary drug residue detection refers to the process of detecting animal-derived foods such as meat, eggs, and dairy products. This involves sample pretreatment (e.g., purification and extraction, matrix removal) to enrich target residues, followed by the use of techniques such as spectral analysis, chromatographic analysis, and immunoassay, combined with chemometric methods to extract characteristic signals of veterinary drug residues. The process quantifies and analyzes residue concentrations to accurately identify residues of commonly used veterinary drugs such as sulfonamides, quinolones, and tetracyclines. Its core purpose is to strictly control the amount of veterinary drug residues in food to meet safety standards, prevent the chronic toxicity and allergic reactions caused by excessive residues, meet food safety regulatory requirements, and ensure the quality and safety of food in the distribution process and public health.

[0003] However, current intelligent detection methods for veterinary drug residues based on multidimensional spectral features fail to address sample matrix interference through targeted purification processes and are largely limited to single-band spectral analysis or simple feature extraction. They lack in-depth exploration of the synergistic response of the visible and near-infrared bands and fail to construct residue discrimination criteria and multi-modal trace residue marker systems strongly correlated with concentration changes. This results in veterinary drug residue characteristic signals being masked by matrix noise, making it difficult to accurately capture the spectral differences corresponding to trace residues. Consequently, the specificity and sensitivity of the detection are insufficient, leading to false positives or missed detections. Therefore, how to enhance the synergistic response of dual-band spectral features for targeted differentiation and detection of veterinary drug residue characteristics to improve the accuracy of veterinary drug residue screening is a challenge facing the industry. Summary of the Invention

[0004] This application provides an intelligent detection system and method for veterinary drug residues based on multidimensional spectral features. It can perform targeted differentiation and detection of veterinary drug residue features under the premise of enhancing the synergistic response of dual-band spectra, thereby improving the accuracy of veterinary drug residue screening.

[0005] In a first aspect, this application provides an intelligent detection method for veterinary drug residues based on multidimensional spectral characteristics, the detection method comprising the following steps:

[0006] The spectral response data of the sample solution to be tested are collected in the visible light band and the near-infrared band. The veterinary drug residue characteristics containing spectral intensity information are determined based on the spectral response data. Then, the veterinary drug residue characteristics are distinguished and verified to obtain the residue discrimination degree when there is a concentration-dependent change in veterinary drug residue at any wavelength.

[0007] Abnormal band combinations that have a quantitative response to veterinary drug residues are screened from the visible light band and the near-infrared band, respectively. The residue peak deviation of veterinary drug residues in the region of interest of the spectrum is determined based on the abnormal band combinations, and then trace residue markers in multiple spectral intensity distribution patterns are determined based on the residue peak deviation.

[0008] Based on the aforementioned residue discrimination criteria and all trace residue markers, in-situ non-destructive screening was performed on animal-derived food samples to be tested for veterinary drug residues.

[0009] In this embodiment, collecting the spectral response data of the test sample solution in the visible and near-infrared bands specifically includes:

[0010] The configuration of the light source and detector for the visible light band and the near-infrared band is determined based on the optical properties of the sample solution to be tested.

[0011] The measurement parameters of the spectrometer are determined according to the configuration of the light source detector, and the raw spectral signal is acquired;

[0012] The spectral response data of the test sample solution in the visible and near-infrared bands are determined by comparing the original spectral signal with a reference standard.

[0013] In this embodiment, the veterinary drug residue features refer to a feature dataset that includes spectral intensity information and is related to the concentration of veterinary drug residues.

[0014] In this embodiment, the distinction and verification of the veterinary drug residue characteristics to obtain the residue discrimination degree when there is a concentration-dependent change in veterinary drug residue at any wavelength specifically includes:

[0015] The spectral intensity benchmark at any wavelength is determined based on multiple standard sample solutions with known gradient concentrations.

[0016] Determine the difference in spectral intensity between the standard sample solution and the zero-concentration reference sample at the same wavelength;

[0017] The sensitivity ratio of spectral changes to concentration changes at each wavelength is determined based on the spectral intensity reference and the spectral intensity difference.

[0018] The residue discrimination criterion is determined based on the response sensitivity ratio when there are concentration-dependent changes in veterinary drug residues at any wavelength.

[0019] In this embodiment, the residue discrimination refers to a quantitative indicator that measures the ability of any wavelength to distinguish veterinary drug residues of different concentrations.

[0020] In this embodiment, the screening of abnormal band combinations that have a quantitative response to veterinary drug residues from the visible light band and the near-infrared band specifically includes:

[0021] Establish a baseline for optical analysis of standard samples containing gradient concentrations of veterinary drug residues and blank control samples;

[0022] Collect full-spectrum spectral response data of the standard sample and the blank control sample in the visible and near-infrared bands;

[0023] By comparing the full-spectrum spectral response data of the standard sample and the blank control sample, and based on the benchmark of the optical analysis, the abnormal band combination that has a quantitative response to veterinary drug residues is determined.

[0024] In this embodiment, the abnormal band combination refers to a set of bands that show a significant quantitative response to changes in veterinary drug residue concentration.

[0025] In this embodiment, determining trace residue markers in multiple spectral intensity distribution modes based on the residual peak deviation specifically includes:

[0026] Based on the residual peak deviation and the spectral intensity distribution of samples of different concentrations in the abnormal band, the deviation correlation information between the specific distribution patterns is determined;

[0027] The spectral identification factor that appears stably in multiple spectral intensity distribution patterns and has a quantitative relationship with the peak deviation value is determined by the deviation correlation information.

[0028] Trace residue markers were identified in multiple spectral intensity distribution patterns based on all spectral identification factors.

[0029] In this embodiment, the in-situ non-destructive screening of veterinary drug residues in animal-derived food samples to be tested, based on the residue discrimination and all trace residue markers, specifically includes:

[0030] The quantitative detection conditions for in-situ spectral acquisition of the sample surface to be tested are determined based on the residual discrimination and all trace residual markers.

[0031] The detection matching degree of the animal-derived food sample to be tested at the set wavelength is determined by the quantitative detection conditions.

[0032] The residual risk level of veterinary drug residues in the animal-derived food sample to be tested is determined by the detection matching degree and the preset matching degree threshold, and the in-situ screening conclusion is output.

[0033] Secondly, this application provides an intelligent detection system for veterinary drug residues based on multidimensional spectral features, used to execute an intelligent detection method for veterinary drug residues based on multidimensional spectral features. The detection system includes:

[0034] The pretreatment module is used to purify the animal-derived food samples to be tested, and obtain the purified sample solution to be tested.

[0035] The differentiation and verification module is used to collect the spectral response data of the sample solution to be tested in the visible light band and the near-infrared band, determine the veterinary drug residue characteristics containing spectral intensity information based on the spectral response data, and then differentiate and verify the veterinary drug residue characteristics to obtain the residue differentiation degree when there is a concentration-dependent change in veterinary drug residue at any wavelength.

[0036] The band analysis module is used to screen out abnormal band combinations that have a quantitative response to veterinary drug residues from the visible light band and the near-infrared band, determine the residue peak deviation of veterinary drug residues in the region of interest of the spectrum based on the abnormal band combinations, and then determine trace residue markers in multiple spectral intensity distribution patterns based on the residue peak deviation.

[0037] The residue detection module is used to perform in-situ non-destructive screening of veterinary drug residues in animal-derived food samples to be tested based on the residue discrimination and all trace residue markers.

[0038] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0039] The animal-derived food samples to be tested are purified to obtain a purified test sample solution. Spectral response data of the test sample solution in the visible and near-infrared bands are collected. Based on the spectral response data, veterinary drug residue characteristics containing spectral intensity information are determined. These veterinary drug residue characteristics are then differentiated and verified to obtain the residue discrimination degree when there is a concentration-dependent change in veterinary drug residue at any wavelength. Abnormal band combinations that have a quantitative response to veterinary drug residues are screened from the visible and near-infrared bands respectively. Based on the abnormal band combinations, the residue peak deviation of veterinary drug residues in the region of interest is determined. Then, trace residue markers in multiple spectral intensity distribution patterns are determined based on the residue discrimination degree and all trace residue markers. In-situ non-destructive screening of veterinary drug residues in the animal-derived food samples to be tested is performed based on the residue discrimination degree and all trace residue markers.

[0040] Therefore, this application demonstrates an improvement in the accuracy of veterinary drug residue detection in scenarios involving complex animal-derived food matrices and trace amounts of veterinary drug residues. Specifically, by purifying the animal-derived food samples, interfering substances such as proteins, fats, and pigments are effectively removed from the sample matrix, maximizing the retention of veterinary drug residues without altering their chemical properties, providing a pure sample solution for dual-band spectral acquisition. By acquiring the spectral response data of the sample solution in the visible and near-infrared bands, veterinary drug residue characteristics containing spectral intensity information are extracted and differentiated, calculating the residue discrimination at any wavelength, thus achieving dual-band spectral acquisition. The complementary use of band spectral information quantifies the response sensitivity of each wavelength to changes in veterinary drug residue concentration, optimizing the concentration correlation loss problem caused by simple feature extraction. By screening anomalous band combinations that have a quantitative response to veterinary drug residues from dual bands, residue peak deviation is determined and trace residue markers in multiple spectral intensity distribution patterns are constructed, achieving accurate capture of trace residue spectral differences. Based on residue discrimination and all trace residue markers, in-situ non-destructive screening of the test samples is carried out, realizing non-destructive, targeted rapid detection, outputting clear residue risk levels and screening conclusions, and improving the efficiency and reliability of on-site batch screening of veterinary drug residues.

[0041] In summary, the technical solution adopted in this application can perform targeted differentiation and detection of veterinary drug residue characteristics under the premise of enhancing the synergistic response of dual-band spectral signals, thereby improving the accuracy of veterinary drug residue screening. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this embodiment of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is an exemplary flowchart of an intelligent detection method for veterinary drug residues based on multidimensional spectral features provided in this application;

[0044] Figure 2 This is a flowchart illustrating the process for determining the characteristics of veterinary drug residues provided in this application;

[0045] Figure 3 This is a flowchart illustrating the determination of residual peak deviation provided in this application;

[0046] Figure 4 This is a module structure diagram of an intelligent veterinary drug residue detection system based on multidimensional spectral characteristics provided in this application. Detailed Implementation

[0047] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0048] This application provides an intelligent detection system and method for veterinary drug residues based on multidimensional spectral features. The core of this system involves purifying animal-derived food samples to obtain a purified sample solution; collecting spectral response data of the sample solution in the visible and near-infrared bands; determining veterinary drug residue characteristics containing spectral intensity information based on the spectral response data; further distinguishing and verifying these characteristics to obtain a residue discrimination degree when veterinary drug residues exhibit concentration-dependent changes at any wavelength; selecting anomalous band combinations that quantitatively respond to veterinary drug residues from both the visible and near-infrared bands; determining the residue peak deviation of veterinary drug residues within the region of interest based on these anomalous band combinations; and then determining trace residue markers under multiple spectral intensity distribution patterns based on the residue discrimination degree and all trace residue markers; and finally performing in-situ non-destructive screening of veterinary drug residues in the animal-derived food samples based on the residue discrimination degree and all trace residue markers.

[0049] Example 1: To better understand the above technical solution, the following will provide a detailed description of the technical solution in conjunction with the accompanying drawings and specific implementation methods. (Refer to...) Figure 1 As shown in the figure, this is an exemplary flowchart of a smart detection method for veterinary drug residues based on multidimensional spectral features according to this embodiment of the present application. The detection method includes the following steps:

[0050] In step S1, the animal-derived food sample to be tested is purified to obtain a purified sample solution.

[0051] In practice, during purification, 5.0 g of the animal-derived food sample to be tested can be weighed, processed to a homogeneous state, and placed in a 50 mL centrifuge tube. 10 mL of acetonitrile is added and vortexed for 5 minutes. Then, 4 g of anhydrous magnesium sulfate and 1 g of sodium chloride are added and vortexed for 2 minutes for salting out. After centrifugation at 8000 rpm for 10 minutes, 4 mL of the upper acetonitrile extract is taken and transferred to a purification tube containing 150 mg of anhydrous magnesium sulfate, 50 mg of N-propylethylenediamine, and 50 mg of C18 adsorbent. The tube is vortexed for 2 minutes and centrifuged at 6000 rpm for 5 minutes. 3 mL of the upper clear liquid is taken and diluted to the mark of a 25 mL volumetric flask with acetonitrile. After shaking well, the purified sample solution to be tested is obtained. Other purification methods can also be used in other embodiments, which are not limited here.

[0052] It should be noted that, in this application, animal-derived food samples refer to food raw material samples from animal-processed products that need to be tested for whether veterinary drug residues exceed the standard; the sample solution to be tested refers to a clear liquid sample that has been purified to remove interfering matrices and can be used for spectroscopic detection to analyze veterinary drug residues.

[0053] In step S2, spectral response data of the sample solution to be tested are collected in the visible light band and near-infrared band. Based on the spectral response data, veterinary drug residue characteristics containing spectral intensity information are determined. Then, the veterinary drug residue characteristics are distinguished and verified to obtain the residue discrimination degree when there is a concentration-dependent change in veterinary drug residue at any wavelength.

[0054] In this embodiment, the spectral response data of the sample solution to be tested in the visible and near-infrared bands can be acquired by the following steps:

[0055] The configuration of the light source and detector for the visible light band and the near-infrared band is determined based on the optical properties of the sample solution to be tested.

[0056] The measurement parameters of the spectrometer are determined according to the configuration of the light source detector, and the raw spectral signal is acquired;

[0057] The spectral response data of the test sample solution in the visible and near-infrared bands are determined by comparing the original spectral signal with a reference standard.

[0058] In practice, firstly, three sets of veterinary drug standard solutions with known concentrations are taken as samples. The absorption spectra of these solutions in the 200-3000nm range are obtained using a full-band scanning device to determine the distribution range of the characteristic absorption peaks of veterinary drug residues. For the visible light band, a deuterium lamp with an emission range covering the electronic transition characteristic peaks is selected, paired with a silicon detector whose response range matches that of the deuterium lamp. For the near-infrared band, a tungsten lamp with an emission range covering the vibration-rotation characteristic peaks is selected, paired with an indium gallium arsenide detector whose response range matches that of the tungsten lamp. Preliminary experiments are conducted to verify that this configuration can completely capture the characteristic peaks, thus determining the light source and detector configurations for the visible light and near-infrared bands. Then, based on the determined light source and detector configuration, the wavelength range is taken as the intersection of the light source emission and the detector response, with the visible light band at 400-760nm and the near-infrared band at 760-2500nm; the resolution is determined based on the half-width at half-maximum (WHM) of the characteristic peak in the preliminary experiment, taking 1 / 5 of the WHM to ensure the peak shape is complete; the number of scans is set to 3, and the average of multiple scans is taken to reduce instrument noise; the integration time is adjusted based on the signal intensity in the preliminary experiment, and determined with 80% of the signal saturation value as the benchmark; the spectrometer is set according to the above parameters, and the sample solution to be tested is placed in the instrument and the acquisition is started to obtain the original spectral signal. Finally, the reference standard is the original spectral signal of a blank solution that is completely identical to the solvent type and concentration of the sample solution to be tested. The original spectral signal and the reference standard signal are collected three times with the same measurement parameters and averaged. The absorbance of the original spectral signal and the reference standard signal at each wavelength is calculated using the formula: absorbance = lg(intensity of reference standard signal / intensity of original spectral signal). After the calculation is completed for each wavelength, the absorbance data of all wavelengths in the dual bands are integrated, and the resulting dataset is used as the spectral response data of the sample solution to be tested in the visible and near-infrared bands.

[0059] It should be noted that, in this application, the visible light band refers to the band used to capture the electronic transition spectral information of veterinary drug molecules; the near-infrared band refers to the band used to capture the vibrational-rotational spectral information of veterinary drug molecules; the light source and detector configuration refers to the combination of light source and detector selected to match the optical characteristics of the sample solution to be tested; the raw spectral signal refers to the raw light intensity data of the sample solution to be tested in the dual bands directly acquired by the spectrometer; the reference standard refers to the blank control spectral signal used to eliminate background interference; and the spectral response data refers to the set of dual-band spectral characteristics reflecting veterinary drug residues in the sample solution to be tested.

[0060] Preferably, in this embodiment, the veterinary drug residue characteristics containing spectral intensity information are determined based on the spectral response data, with reference to... Figure 2 As shown in the figure, this is a flowchart illustrating the process of determining veterinary drug residue characteristics in some embodiments of this application. In this embodiment, the determination of veterinary drug residue characteristics can be achieved through the following steps:

[0061] In step S21, the standard absorption bands of veterinary drug residues for comparison are determined based on the standard spectral response data of the pure matrix.

[0062] In step S22, the actual absorption intensity of the sample solution to be tested within the characteristic absorption band of veterinary drug residues is determined based on the spectral response data of the sample solution to be tested.

[0063] In step S23, the spectral difference data caused by veterinary drug residues are determined based on the veterinary drug residue characteristic absorption band reference and the actual absorption intensity.

[0064] In step S24, the spectral difference data is fitted with a standard concentration curve to obtain veterinary drug residue characteristics containing spectral intensity information.

[0065] In practice, firstly, a pure matrix sample identical to the matrix of the sample solution to be tested is selected, and its dual-band standard spectral response data is obtained according to the aforementioned purification and spectral acquisition methods. In the preliminary experiment, a series of veterinary drug standard solutions of known concentrations are used, and their spectral response data are obtained and compared with the pure matrix standard spectrum. Bands that show significant changes with concentration and have no matrix absorption are selected as candidate bands. Using the absorption intensity of each wavelength in the pure matrix standard spectrum within the candidate bands as a baseline, the wavelength range and corresponding baseline intensity of the candidate bands are determined, and integrated to obtain the veterinary drug residue characteristic absorption spectrum baseline. Next, the dual-band spectral response data of the sample solution to be tested is extracted. Based on the wavelength range corresponding to the determined veterinary drug residue characteristic absorption spectrum band, absorbance is converted for each wavelength within the spectrum band. The conversion formula is that absorbance equals the logarithm of the ratio of the base-10 reference light intensity to the original signal light intensity at that wavelength. After conversion, the absorbance values ​​corresponding to all wavelengths within the spectrum band are compiled, and the resulting dataset is taken as the actual absorption intensity of the sample solution to be tested within the veterinary drug residue characteristic absorption spectrum band. Then, baseline intensity data from the characteristic absorption bands of veterinary drug residues were extracted to ensure that its wavelength range was completely consistent with the wavelength range of the actual absorption intensity of the sample under test. The difference between the actual absorption intensity and the corresponding baseline intensity was calculated for each wavelength, by subtracting the baseline intensity from the actual absorption intensity at that wavelength. The validity of all calculated wavelength differences was verified, and abnormal differences caused by instrument noise were removed. The verified difference dataset was then integrated as spectral difference data caused by veterinary drug residues. Finally, a series of veterinary drug standard solutions of known concentrations were prepared, and spectral difference data for each concentration was obtained. Using veterinary drug residue concentration as the independent variable and the mean of the corresponding spectral difference data as the dependent variable, a linear regression algorithm was used to construct a standard concentration curve. The coefficient of determination of the curve was calculated to verify linear correlation and ensure that the coefficient of determination met the requirements for quantitative analysis. The spectral difference data of the sample under test was substituted into the standard concentration curve to obtain concentration-related spectral intensity correlation data. This data was integrated with the corresponding characteristic band information to obtain the veterinary drug residue characteristics containing spectral intensity information.

[0066] It should be noted that, in this application, spectral intensity information refers to the spectral absorption intensity data and derived difference data of the test sample solution at each wavelength in the visible and near-infrared bands after calibration with a reference standard; the standard spectral response data of the pure matrix refers to the dual-band spectral data of a blank sample that is completely identical to the matrix of the test sample solution and has no veterinary drug residues; the veterinary drug residue characteristic absorption band reference refers to the range of spectral absorption bands unique to veterinary drug residues and the corresponding baseline intensity determined based on the standard spectrum of the pure matrix; the veterinary drug residue characteristic absorption band refers to a specific wavelength range in the visible and near-infrared bands of veterinary drug molecules, where the spectral absorption intensity changes significantly with the concentration of veterinary drug and there is no obvious absorption in the pure matrix; the actual absorption intensity refers to the actual spectral absorption intensity of the test sample solution at each wavelength within the characteristic absorption band of veterinary drug residues; the spectral difference data refers to the difference in spectral intensity between the test sample and the pure matrix within the characteristic absorption band caused by veterinary drug residues; the standard concentration curve refers to the fitting curve constructed by the spectral difference data of a series of known concentration veterinary drug standard solutions and the corresponding concentrations; and the veterinary drug residue characteristics refer to a feature dataset containing spectral intensity information and related to the concentration of veterinary drug residues.

[0067] In this embodiment, the distinction and verification of the veterinary drug residue characteristics to obtain the residue discrimination degree when there is a concentration-dependent change in veterinary drug residue at any wavelength can be achieved by the following steps:

[0068] The spectral intensity benchmark at any wavelength is determined based on multiple standard sample solutions with known gradient concentrations.

[0069] Determine the difference in spectral intensity between the standard sample solution and the zero-concentration reference sample at the same wavelength;

[0070] The sensitivity ratio of spectral changes to concentration changes at each wavelength is determined based on the spectral intensity reference and the spectral intensity difference.

[0071] The residue discrimination criterion is determined based on the response sensitivity ratio when there are concentration-dependent changes in veterinary drug residues at any wavelength.

[0072] In practice, firstly, 5-8 gradient veterinary drug standard solutions of known concentrations are prepared, covering the range from trace to low concentrations. Each concentration standard solution is processed, and spectral response data is collected six times for each concentration. The spectral intensity values ​​of all repetitions at each wavelength are extracted, and the arithmetic mean at each wavelength is calculated. Outliers other than three times the standard deviation are removed, and the mean is recalculated. The final mean values ​​corresponding to each wavelength are integrated into a dataset, which serves as the spectral intensity baseline for each known gradient concentration standard sample solution at any wavelength. Next, a zero-concentration reference sample with a matrix completely identical to the standard sample solution is prepared. Spectral response data is collected six times, and the mean at each wavelength is calculated to obtain the spectral intensity baseline of the zero-concentration reference sample. The spectral intensity baselines of each gradient concentration standard sample are extracted, and the difference between the standard sample spectral intensity baseline and the zero-concentration reference sample spectral intensity baseline at the same wavelength is calculated for each wavelength. The differences corresponding to all wavelengths are integrated to obtain a dataset of spectral intensity differences between each concentration standard sample solution and the zero-concentration reference sample. Then, the spectral intensity differences and corresponding veterinary drug residue concentrations of standard samples at each concentration gradient are extracted. The concentration difference between two adjacent concentration gradients and the spectral intensity difference at the corresponding wavelength are calculated. The ratio of the spectral intensity difference to the concentration difference is calculated for each wavelength to obtain the response sensitivity ratio of adjacent concentration gradients at that wavelength. The arithmetic mean of the response sensitivity ratios of all adjacent concentration gradients at each wavelength is calculated, and this mean is the response sensitivity ratio of spectral change to concentration change at each wavelength. Finally, at least three sets of response sensitivity ratio data for different veterinary drug types are collected, and the global mean and standard deviation of all data are calculated. The global mean plus one standard deviation is used as the effective threshold for the response sensitivity ratio. Wavelengths with response sensitivity ratios greater than the effective threshold are selected, and the response sensitivity ratios of these wavelengths are normalized by dividing the response sensitivity ratio of that wavelength by the maximum value of the response sensitivity ratios of all wavelengths. The normalized value is the residue discrimination index at that wavelength. The residue discrimination indexes of all wavelengths are integrated to obtain the residue discrimination index when there is a concentration-dependent change in veterinary drug residues at any wavelength.

[0073] It should be noted that, in this application, "discrimination verification" refers to the operation of quantifying and verifying the ability of spectral signals at various wavelengths to identify differences in veterinary drug residue concentrations; "concentration-dependent change" refers to the regular and correlated trend of spectral intensity information corresponding to veterinary drug residues as the spectral concentration gradient changes; "standard sample solution" refers to a series of standard solutions of the same matrix containing different concentrations of veterinary drug residues; "spectral intensity benchmark" refers to the statistical value of the spectral intensity of standard sample solutions of known gradient concentrations at any wavelength; "zero concentration reference sample" refers to a blank sample that is completely identical to the sample matrix to be tested and has no veterinary drug residues; "spectral intensity difference" refers to the difference in spectral intensity between standard sample solutions of known gradient concentrations and the zero concentration reference sample at the same wavelength; "response sensitivity ratio" refers to the ratio of the change in spectral intensity at each wavelength to the change in veterinary drug residue concentration; and "residue discrimination" refers to a quantitative indicator that measures the ability of any wavelength to distinguish different concentrations of veterinary drug residues.

[0074] In step S3, abnormal band combinations that have a quantitative response to veterinary drug residues are screened from the visible light band and the near-infrared band, respectively. The residue peak deviation of veterinary drug residues in the region of interest of the spectrum is determined based on the abnormal band combinations, and then trace residue markers in multiple spectral intensity distribution patterns are determined based on the residue peak deviation.

[0075] In this embodiment, the following steps can be used to screen out abnormal band combinations that have a quantitative response to veterinary drug residues from the visible light band and the near-infrared band:

[0076] Establish a baseline for optical analysis of standard samples containing gradient concentrations of veterinary drug residues and blank control samples;

[0077] Collect full-spectrum spectral response data of the standard sample and the blank control sample in the visible and near-infrared bands;

[0078] By comparing the full-spectrum spectral response data of the standard sample and the blank control sample, and based on the benchmark of the optical analysis, the abnormal band combination that has a quantitative response to veterinary drug residues is determined.

[0079] In practice, firstly, the target veterinary drug type is selected, and the core indicators for determining the quantitative response are correlation coefficient, signal-to-noise ratio (SNR), and response stability. A series of gradient concentration standard samples are prepared through preliminary experiments. After collecting spectral data, each indicator is calculated: the correlation coefficient is calculated using Pearson correlation analysis to determine the linear correlation between spectral intensity and concentration; the SNR is the ratio of the signal peak value to the background noise; and the response stability is the coefficient of variation of spectral intensity measured multiple times at the same concentration. Based on the results of three sets of parallel preliminary experiments, thresholds are determined for the indicators: a correlation coefficient not lower than 0.95, a SNR not lower than 3, and a coefficient of variation not higher than 5%. These indicators and thresholds are then integrated to form the benchmark for optical analysis. Next, blank control samples with the same matrix as the test sample are prepared, along with gradient concentration standard samples covering trace to low concentrations. Three parallel samples are prepared for each group. A calibrated spectrometer is used, with the scanning range set to the complete visible and near-infrared bands, ensuring that parameters such as wavelength resolution, number of scans, and integration time are consistent with the spectral acquisition. Spectral data of the blank control samples and each gradient concentration standard sample are collected sequentially, with each sample collected three times. The average value is taken as the final full-spectrum spectral response data. Finally, the full-spectrum spectral response data were preprocessed, with Savitzky-Golay smoothing used to eliminate high-frequency noise, and baseline drift corrected using standard normal transformation. The spectral intensity differences between standard samples and blank control samples at each gradient concentration were calculated for each band, yielding concentration-intensity difference data for each band. For each band, the correlation coefficient, signal-to-noise ratio, and response stability were calculated and compared with the baseline thresholds for optical analysis. Bands meeting all three criteria were selected and categorized into visible light and near-infrared bands based on wavelength. Combination validation was performed within each group, and partial least squares regression was used to analyze the quantitative prediction capability of the combined bands. The combination with the smallest prediction error was retained as the anomalous band combination with a quantitative response to veterinary drug residues.

[0080] It should be noted that, in this application, quantitative response refers to a specific band within the visible and near-infrared bands, where the corresponding spectral intensity data exhibits a regular and quantifiable correlation with the veterinary drug residue concentration, and this correlation meets the preset optical analysis benchmark requirements; standard sample refers to a standard solution containing a series of continuous concentrations of veterinary drugs and consistent with the matrix of the sample to be tested; blank control sample refers to a sample with a matrix completely consistent with the standard sample and without any veterinary drug residues; the benchmark for optical analysis refers to the criteria for screening abnormal band combinations; full-spectrum spectral response data refers to the spectral intensity data of the sample covering the complete visible and near-infrared bands; abnormal band combination refers to a set of several bands that show a significant quantitative response to changes in veterinary drug residue concentration.

[0081] Preferably, in this embodiment, the deviation of the residual peak value of veterinary drug residues in the region of interest of the spectrum is determined based on the abnormal band combination, with reference to... Figure 3As shown in the figure, this is a flowchart illustrating the process of determining the residual peak deviation in some embodiments of this application. In this embodiment, the determination of the residual peak deviation can be achieved using the following steps:

[0082] In step S31, the standard residue characteristic spectrum is determined based on the spectral response of the standard concentration veterinary drug residue sample within the abnormal band combination;

[0083] In step S32, the difference in spectral intensity between the measured spectrum of the sample to be tested and the standard residual characteristic spectrum at each wavelength point within the abnormal band combination is determined.

[0084] In step S33, the spectral deviation characteristics of the measured spectrum within the region of interest are determined based on the spectral intensity difference.

[0085] In step S34, the residual peak deviation of veterinary drug residues in the region of interest of the spectrum is determined based on the spectral deviation characteristics.

[0086] In practice, firstly, three standard concentration veterinary drug residue samples with the same matrix as the test sample are prepared to ensure that the purification conditions are completely consistent with those of the test sample. Using a calibrated spectrometer, spectral response data for each sample are acquired within the wavelength range corresponding to the abnormal band combinations, according to the previously set dual-band acquisition parameters. Each sample is acquired six times. All acquired data are imported into analysis software. First, Savitzky-Golay smoothing is used to eliminate high-frequency noise. The smoothing window and polynomial order are determined based on pre-experiments to ensure peak shape integrity. Then, standard normal variable transformation is used to correct baseline drift. Finally, Gaussian fitting is performed on the processed data to determine parameters such as spectral peak shape, peak position, and peak intensity. The resulting smooth and stable spectral curve is the standard residue characteristic spectrum. Next, the test sample is treated using the exact same purification process as the standard concentration veterinary drug residue sample to obtain the purified test sample solution. Using the same spectrometer and acquisition parameters, spectral response data for the test sample is acquired within the wavelength range corresponding to the abnormal band combinations. After six repetitions, the average value is taken to obtain the measured spectrum of the test sample. The wavelength ranges of the measured spectrum and the standard residual characteristic spectrum are precisely aligned to ensure a one-to-one correspondence between each wavelength point. The difference between the measured spectral intensity and the corresponding wavelength intensity of the standard residual characteristic spectrum is calculated for each wavelength. The validity of all differences is verified, and outliers exceeding three times the standard deviation due to instantaneous instrument fluctuations are removed. The remaining valid differences are integrated to form the spectral intensity difference between the measured spectrum and the standard residual characteristic spectrum at each wavelength point within the abnormal band combination of the test sample. Then, the spectral intensity difference dataset is statistically analyzed, calculating the absolute value of the difference at each wavelength point. Wavelength points with an absolute value greater than twice the mean of all differences are selected; the continuous bands containing these wavelength points constitute the region of interest (ROI). All spectral intensity differences within the ROI are extracted, and the difference distribution curve is plotted using data analysis software. Spectral deviation characteristics are then extracted, including the overall distribution trend of the differences (increasing, decreasing, or fluctuating), the maximum and minimum values ​​of the differences, the wavelength position corresponding to the maximum value, the distribution range of the differences, and the coefficient of variation. This collection of information constitutes the spectral deviation characteristics of the measured spectrum within the ROI. Finally, the wavelength position corresponding to the maximum difference is extracted from the spectral deviation characteristics, and this wavelength is defined as the residual characteristic peak wavelength. The spectral intensity difference corresponding to this wavelength point is obtained, which is the deviation value of the measured spectrum at the characteristic peak wavelength. At the same time, the standard peak intensity of the standard residual characteristic spectrum at this peak wavelength is extracted, and the difference between the measured spectral peak intensity (standard peak intensity + wavelength difference) and the standard peak intensity is calculated, which is the residual peak deviation. The repeatability of this deviation value is verified by repeated calculations of 3 sets of parallel test samples to ensure that the coefficient of variation does not exceed the preset threshold. The finally determined stable value is the residual peak deviation of veterinary drug residues in the region of interest of the spectrum.

[0087] It should be noted that, in this application, the region of interest in spectroscopy refers to a specific wavelength range within the anomalous band combination where the spectral intensity difference is significant and the correlation with veterinary drug residues is highest; the standard concentration veterinary drug residue sample refers to a standard solution containing a known fixed concentration of veterinary drug and completely identical to the matrix of the sample to be tested; the standard residue characteristic spectrum refers to a characteristic spectral curve constructed based on the spectral response data of the standard concentration veterinary drug residue sample; the measured spectrum refers to the actual spectral response curve of the sample solution to be tested collected within the anomalous band combination; the spectral intensity difference refers to the intensity difference between the measured spectrum of the sample to be tested and the standard residue characteristic spectrum at the same wavelength point; the spectral deviation characteristic refers to the information on the distribution trend, extreme value position, and variation range of the spectral intensity difference within the region of interest; and the residue peak deviation refers to the quantitative difference between the peak value of the measured spectrum of the sample to be tested and the peak value of the standard residue characteristic spectrum.

[0088] In this embodiment, determining trace residual markers in multiple spectral intensity distribution modes based on the residual peak deviation can be achieved using the following steps:

[0089] Based on the residual peak deviation and the spectral intensity distribution of samples of different concentrations in the abnormal band, the deviation correlation information between the specific distribution patterns is determined;

[0090] The spectral identification factor that appears stably in multiple spectral intensity distribution patterns and has a quantitative relationship with the peak deviation value is determined by the deviation correlation information.

[0091] Trace residue markers were identified in multiple spectral intensity distribution patterns based on all spectral identification factors.

[0092] In practice, firstly, a series of veterinary drug standard samples with gradient concentrations ranging from trace to low concentrations are prepared. Spectral response data within abnormal band combinations are collected, with each concentration repeated six times. Cluster analysis is used to classify the spectral curves of all samples, identifying 3-5 specific spectral intensity distribution patterns based on peak shape (single-peak, double-peak), peak value change trend (increasing, decreasing), and intensity distribution gradient. The residual peak deviation for each concentration sample under each pattern is calculated, establishing a three-dimensional data matrix of "distribution pattern - veterinary drug concentration - residual peak deviation." Correlation analysis clarifies the quantitative correlation between deviation and concentration within the same pattern and the synergistic law of deviation changes between different patterns, integrating the deviation correlation information between specific distribution patterns. Then, candidate identification indicators are extracted from the deviation correlation information, including the mean, extreme values, coefficient of variation, peak position shift, and intensity gradient change rate of the residual peak deviation. Cross-pattern stability verification is performed on each candidate indicator, calculating its coefficient of variation across all distribution patterns, and eliminating unstable indicators with coefficients of variation exceeding a preset range. For the remaining stable indicators, a linear regression model was used to analyze their quantitative relationship with the residual peak deviation value. The coefficient of determination was used as the criterion to screen indicators whose coefficients of determination met the quantitative requirements. These indicators, possessing both stability and quantitative relationships, were integrated to obtain spectral identification factors that appeared stably in multiple spectral intensity distribution patterns and had a quantitative relationship with the peak deviation value. Finally, a large number of veterinary drug standard samples with trace concentrations (covering the target trace range) were collected, and their spectral intensity distribution patterns and corresponding spectral identification factor values ​​were obtained. Statistical analysis was performed on the identification factor values ​​under each distribution pattern, calculating the mean and standard deviation. The "mean - 1 standard deviation" was used as the trace threshold for the corresponding identification factor under that pattern. The trace thresholds of all identification factors under each pattern were bound to the spectral intensity distribution morphological characteristics of that pattern, forming a combined judgment unit of "distribution pattern + multiple identification factor thresholds". The recognition accuracy of each combined unit was verified through blind sample testing. Units with accuracy below the requirement were eliminated, and the remaining combined units together constituted trace residue markers under multiple spectral intensity distribution patterns.

[0093] It should be noted that, in this application, the spectral intensity distribution pattern refers to the spectral curve morphology of veterinary drug residue samples of different concentrations within abnormal band combinations, exhibiting significant characteristic differences; the spectral intensity distribution morphology refers to the peak shape, peak value variation trend, and intensity gradient distribution of spectral curves of veterinary drug residue samples of different concentrations within abnormal band combinations; the deviation correlation information between specific distribution patterns refers to the corresponding correlation data set between residue peak deviation and veterinary drug residue concentration and spectral morphological characteristics under different spectral intensity distribution patterns; the peak deviation value refers to the quantitative difference between the measured spectral peak value of the sample under test in the region of interest and the standard residue characteristic spectral peak value of the standard concentration veterinary drug residue sample; the spectral identification factor refers to a spectral characteristic index that appears stably in multiple spectral intensity distribution patterns and has a significant quantitative relationship with the residue peak deviation value; and the trace residue marker refers to the specific determination standard for trace veterinary drug residues.

[0094] In step S4, in-situ non-destructive screening of veterinary drug residues in the animal-derived food samples to be tested is performed based on the residue discrimination and all trace residue markers.

[0095] In this embodiment, the in-situ non-destructive screening of veterinary drug residues in animal-derived food samples to be tested, based on the residue discrimination and all trace residue markers, can be achieved using the following steps:

[0096] The quantitative detection conditions for in-situ spectral acquisition of the sample surface to be tested are determined based on the residual discrimination and all trace residual markers.

[0097] The detection matching degree of the animal-derived food sample to be tested at the set wavelength is determined by the quantitative detection conditions.

[0098] The residual risk level of veterinary drug residues in the animal-derived food sample to be tested is determined by the detection matching degree and the preset matching degree threshold, and the in-situ screening conclusion is output.

[0099] In practice, firstly, the residual discrimination data calculated in the previous stage are extracted, and wavelengths with residual discrimination greater than the effective threshold are selected to form a high-sensitivity wavelength pool. Combining the abnormal band combinations corresponding to all trace residual markers, wavelengths overlapping between the two are further selected from the high-sensitivity wavelength pool and determined as the set wavelength. Using the set wavelength as the core, the spectral acquisition parameter range is adopted, adjusting the spectrometer's scanning range (only covering the set wavelength), resolution (matching the peak shape requirements of the set wavelength), number of scans (balancing efficiency and signal-to-noise ratio), and integration time (ensuring sufficient signal strength). Pre-acquisition verification is performed using 3 sets of standard trace samples and 3 sets of blank samples. Parameters are adjusted until the acquired spectral data can stably extract the features corresponding to the trace residual markers. The extracted features are used as the quantitative detection conditions for in-situ spectral acquisition of the sample surface. Then, the in-situ spectrometer is started according to the quantitative detection conditions, and the spectrometer probe is vertically aligned with the sample surface (maintaining a fixed distance to avoid optical path changes). In-situ spectral response data at the set wavelength is directly acquired, and the acquisition is repeated 6 times, with the average value taken as the measured spectral data. The measured spectral data were preprocessed (Savitzky-Golay smoothing + standard normal transformation) to extract spectral identification factors (such as peak bias, intensity gradient, etc.) corresponding to trace residue markers. A weighted summation algorithm was used to calculate the detection matching degree: weights were assigned to each spectral identification factor (weights were determined based on their cross-mode stability and quantitative correlation; stable and highly correlated factors had higher weights). The fit score between the measured value of each identification factor and the trace residue marker threshold was calculated (higher fit scores were higher, ranging from 0 to 1). The weighted summation result was used as the detection matching degree of the animal-derived food sample to be tested at the set wavelength. Finally, a large number of trace veterinary drug residue samples (covering the target trace range) and blank samples were collected. The detection matching degree of each group of samples was calculated, and a preset matching degree threshold was determined through statistical analysis: the maximum matching degree of the blank sample was used as the "low-risk-medium-risk" threshold, and the minimum matching degree of the trace sample was used as the "medium-risk-high-risk" threshold. The detection match degree of the sample to be tested is compared with two thresholds: below the low-risk threshold is judged as "low risk" (no trace residue detected), between the two thresholds is judged as "medium risk" (suspected detection, requiring further verification), and above the high-risk threshold is judged as "high risk" (clearly detected trace residue). The residual risk level, detection match degree value, set wavelength information, and quantitative detection condition parameters are integrated to form and output an in-situ screening conclusion that includes data, judgment criteria, and risk warnings.

[0100] It should be noted that, in this application, in-situ non-destructive screening refers to a veterinary drug residue detection method that combines residue discrimination and trace residue markers without damaging the original physical form and chemical properties of the animal-derived food sample to be tested; in-situ spectral acquisition refers to the operation of directly pointing the spectrometer probe at the surface of the animal-derived food sample to be tested and collecting spectral response data within a set wavelength range based on quantitative detection conditions; quantitative detection conditions refer to the set of key parameters for in-situ spectral acquisition determined based on residue discrimination and trace residue markers; detection matching degree refers to an index that quantifies the degree of fit between the measured spectral characteristics of the sample and the trace residue markers; preset matching degree threshold refers to the critical value for classifying residue risk levels; residue risk level refers to the result of defining the potential hazard level of veterinary drug residues in the sample; and in-situ screening conclusion refers to the final detection result formed by integrating residue risk level, detection matching degree, and set wavelengths.

[0101] Therefore, this application demonstrates an improvement in the accuracy of veterinary drug residue detection in scenarios involving complex animal-derived food matrices and trace amounts of veterinary drug residues. Specifically, by purifying the animal-derived food samples, interfering substances such as proteins, fats, and pigments are effectively removed from the sample matrix, maximizing the retention of veterinary drug residues without altering their chemical properties, providing a pure sample solution for dual-band spectral acquisition. By acquiring the spectral response data of the sample solution in the visible and near-infrared bands, veterinary drug residue characteristics containing spectral intensity information are extracted and differentiated, calculating the residue discrimination at any wavelength, thus achieving dual-band spectral acquisition. The complementary use of band spectral information quantifies the response sensitivity of each wavelength to changes in veterinary drug residue concentration, optimizing the concentration correlation loss problem caused by simple feature extraction. By screening anomalous band combinations that have a quantitative response to veterinary drug residues from dual bands, residue peak deviation is determined and trace residue markers in multiple spectral intensity distribution patterns are constructed, achieving accurate capture of trace residue spectral differences. Based on residue discrimination and all trace residue markers, in-situ non-destructive screening of the test samples is carried out, realizing non-destructive, targeted rapid detection, outputting clear residue risk levels and screening conclusions, and improving the efficiency and reliability of on-site batch screening of veterinary drug residues.

[0102] In summary, the technical solution adopted in this application can perform targeted differentiation and detection of veterinary drug residue characteristics under the premise of enhancing the synergistic response of dual-band spectral signals, thereby improving the accuracy of veterinary drug residue screening.

[0103] Example 2: This application provides an intelligent detection system for veterinary drug residues based on multidimensional spectral characteristics, referring to... Figure 4 As shown in the figure, this is a modular structure diagram of a smart veterinary drug residue detection system based on multidimensional spectral features according to this embodiment of the present application. The detection system includes:

[0104] The pretreatment module 100 is used to purify the animal-derived food samples to be tested, and obtain the purified sample solution to be tested.

[0105] The differentiation and verification module 200 is used to collect the spectral response data of the sample solution to be tested in the visible light band and the near-infrared band, determine the veterinary drug residue characteristics containing spectral intensity information based on the spectral response data, and then differentiate and verify the veterinary drug residue characteristics to obtain the residue differentiation degree when there is a concentration-dependent change in veterinary drug residue at any wavelength.

[0106] The band analysis module 300 is used to screen out abnormal band combinations that have a quantitative response to veterinary drug residues from the visible light band and the near-infrared band, determine the residue peak deviation of veterinary drug residues in the region of interest of the spectrum based on the abnormal band combinations, and then determine trace residue markers in multiple spectral intensity distribution patterns based on the residue peak deviation.

[0107] The residue detection module 400 is used to perform in-situ non-destructive screening of veterinary drug residues in animal-derived food samples to be tested based on the residue discrimination and all trace residue markers.

[0108] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compactdisc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0110] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A multi-dimensional spectral feature-based intelligent detection method for veterinary drug residues, characterized in that, The detection method comprises the following steps: The animal-derived food sample to be detected is subjected to purification treatment to obtain a purified sample solution to be detected; Spectrum response data of the sample solution to be detected in the visible light band and the near-infrared band are collected, and a veterinary drug residue feature containing spectrum intensity information is determined according to the spectrum response data, and the veterinary drug residue feature is verified to obtain a residue discrimination degree when the veterinary drug residue exists in concentration-dependent change at any wavelength; An abnormal band combination with quantitative response to the veterinary drug residue is screened out from the visible light band and the near-infrared band, respectively, and a residue peak deviation of the veterinary drug residue in a spectrum region of interest is determined according to the abnormal band combination, and trace residue marks in multiple spectrum intensity distribution modes are determined from the residue peak deviation; The residue discrimination degree and all the trace residue marks are used for in-situ nondestructive screening of the veterinary drug residue in the animal-derived food sample to be detected.

2. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The spectrum response data of the sample solution to be detected in the visible light band and the near-infrared band specifically comprises: A light source detector configuration of the visible light band and the near-infrared band is determined according to the optical characteristics of the sample solution to be detected; Measurement parameters of a spectrometer are determined according to the light source detector configuration, and original spectrum signals are collected; The spectrum response data of the sample solution to be detected in the visible light band and the near-infrared band are determined through the original spectrum signals and a reference benchmark.

3. The method of claim 1, wherein the method comprises: The veterinary drug residue feature refers to a feature data set containing spectrum intensity information and related to the concentration of the veterinary drug residue.

4. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The veterinary drug residue feature is verified to obtain a residue discrimination degree when the veterinary drug residue exists in concentration-dependent change at any wavelength, specifically comprising: Spectrum intensity benchmarks of a standard sample solution with multiple known gradient concentrations at any wavelength are determined; Spectrum intensity difference values of the standard sample solution and a zero-concentration reference sample at the same wavelength are determined; A response sensitivity ratio of spectrum change to concentration change at each wavelength is determined according to the spectrum intensity benchmarks and the spectrum intensity difference values; The residue discrimination degree when the veterinary drug residue exists in concentration-dependent change at any wavelength is determined according to the response sensitivity ratio.

5. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The residue discrimination degree refers to a quantitative index for measuring the ability of differentiating veterinary drug residues with different concentrations at any wavelength.

6. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The abnormal band combination with quantitative response to the veterinary drug residue is screened out from the visible light band and the near-infrared band, specifically comprising: A benchmark of optical analysis of a standard sample containing gradient-concentration veterinary drug residues and a blank control sample is set; Full-spectrum spectrum response data of the standard sample and the blank control sample in the visible light band and the near-infrared band are collected; The abnormal band combination with quantitative response to the veterinary drug residue is determined by comparing the full-spectrum spectrum response data of the standard sample and the blank control sample, and according to the benchmark of optical analysis.

7. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The abnormal band combination refers to a set of several bands that have significant quantitative response to the concentration change of the veterinary drug residue.

8. The intelligent detection method of veterinary drug residues based on multi-dimensional spectral characteristics according to claim 1, characterized in that, The trace residue marks in multiple spectrum intensity distribution modes are determined from the residue peak deviation, specifically comprising: According to the residual peak deviation and the deviation correlation information between the spectral intensity distribution patterns of different concentration samples in the abnormal waveband, a specific distribution pattern is determined; Through the deviation correlation information, a spectral discrimination factor that is stable in multiple spectral intensity distribution patterns and has a quantitative relationship with the peak deviation value is determined; Based on all the spectral discrimination factors, trace residual marks in multiple spectral intensity distribution patterns are determined.

9. The multi-dimensional spectral feature-based intelligent detection method for veterinary drug residues according to claim 1, characterized in that, According to the residual discrimination degree and all the trace residual marks, the veterinary drug residues in the animal-derived food sample to be detected are in-situ and non-destructively screened, and the specific steps include: According to the residual discrimination degree and all the trace residual marks, the quantitative detection conditions for in-situ spectral collection of the sample surface to be detected are determined; Through the quantitative detection conditions, the detection matching degree of the animal-derived food sample to be detected at a set wavelength is determined; According to the detection matching degree and a preset matching degree threshold, the residual risk level of the veterinary drug residues in the animal-derived food sample to be detected is determined, and an in-situ screening conclusion is output.

10. A multi-dimensional spectral feature based intelligent detection system for veterinary drug residues, for performing a multi-dimensional spectral feature based intelligent detection method for veterinary drug residues according to any one of claims 1 to 9, characterized in that, The detection system includes: A pretreatment module for purifying the animal-derived food sample to be detected to obtain a purified sample solution to be detected; A discrimination verification module for collecting spectral response data of the sample solution to be detected in the visible light waveband and the near-infrared waveband, determining veterinary drug residue characteristics containing spectral intensity information according to the spectral response data, and then performing discrimination verification on the veterinary drug residue characteristics to obtain a residual discrimination degree when the concentration of the veterinary drug residues present at any wavelength changes; A waveband analysis module for screening abnormal waveband combinations that have quantitative responses to the veterinary drug residues from the visible light waveband and the near-infrared waveband, respectively, determining a residual peak deviation of the veterinary drug residues in a spectral region of interest according to the abnormal waveband combinations, and then determining trace residual marks in multiple spectral intensity distribution patterns from the residual peak deviation; A residual detection module for screening the veterinary drug residues in the animal-derived food sample to be detected in-situ and non-destructively according to the residual discrimination degree and all the trace residual marks.

Images