Intelligent clenbuterol concentration mapping method based on Raman scattering space-time dynamic attenuation characteristics

Abstract

The invention relates to an intelligent clenbuterol concentration mapping method based on Raman scattering space-time dynamic attenuation characteristics. According to the method, based on theoretical models of molecular dynamics and statistical physics, frequency-domain spectral information is inverted to a time-domain coherence function, a clenbuterol molecule'free state-cluster state 'double-pool dynamic model is established, and internal correlation between concentration and molecule aggregation state and coherence attenuation is revealed for the first time; abnormal samples can be automatically identified in detection of clenbuterol, and a quality scoring mechanism is provided; compared with traditional Raman spectrum detection, the sensitivity and the measuring range of the method are remarkably improved, the detection limit is reduced to 0.42 mu g / kg from 10-50 mu g / kg and is improved by 20-100 times, and the national standard requirement (1t, 0.5 mu g / kg) is met; the linear range of the measuring range is expanded from 1-2 orders of magnitude to 3-4 orders of magnitude (0.5-500 [mu] g / kg), and a high-residue full scene is covered by trace detection; the method can also be expanded to detection of other small molecules and does not depend on specific chemical properties of substances.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of Raman spectroscopy analysis and food safety testing, specifically a method for intelligent mapping of ractopamine concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering. Background Technology

[0002] Clenbuterol (a β-agonist compound, such as clenbuterol hydrochloride and ractopamine) is a type of feed additive that promotes lean meat growth in animals. However, its residues can cause serious harm to humans, such as palpitations and muscle tremors, and its use has been explicitly banned in my country. Establishing rapid, accurate, and on-site detection methods for clenbuterol is of great significance for ensuring food safety.

[0003] The existing methods for detecting clenbuterol mainly include the following categories: (1) Enzyme-linked immunosorbent assay (ELISA) has a high false positive rate, especially in the low concentration range (<1 μg / kg) where it is easily affected by matrix interference and cross-reaction. The antibody specificity is limited, and there is confusion in the recognition of β-agonist compounds with similar structures. This method also requires strict temperature and time control, and the detection window period is short and the field application is limited.

[0004] (2) Liquid chromatography-tandem mass spectrometry (LC-MS / MS) has a complex pretreatment process, high reagent consumption, severe matrix effect, high cost per sample, and long detection time, which cannot meet the needs of rapid on-site screening. In addition, its detection equipment is expensive and requires high technical skills from operators.

[0005] (3) Traditional Raman spectroscopy, which relies solely on linear fitting based on a single parameter such as peak intensity or peak width, makes it difficult to break through the bottleneck of 10 μg / kg in detection limit; it does not utilize the time-domain dynamics information of coherent attenuation; single-point or small-point measurements cannot reflect the uneven distribution of ractopamine in muscle tissue, resulting in poor detection repeatability; and the fluorescence background generated by a large number of chromophores such as hemoglobin and myoglobin in the meat will mask the narrow-band Raman peak of ractopamine (half-peak width approximately...). The relative error of the target signal can reach 30-50%, affecting the detection results.

[0006] Raman scattering is essentially a coherent evolution process of molecular vibrational excited states, and its frequency domain peak shape is a Fourier transform of the time domain coherence function. Traditional methods, however, only analyze the static peak shape in the frequency domain, completely discarding the rich dynamic information contained in the time-domain coherent decay: ① Molecules in different aggregation states (monomers, dimers, clusters) have significantly different coherent decay times. This is directly related to concentration; ② The coherent decay rate is extremely sensitive to the molecular microenvironment (degree of solvation, hydrogen bond network, charge distribution), and can serve as a "kinetic fingerprint" of concentration; ③ The temporal fluctuation characteristics reflect the time scale of molecular diffusion and concentration fluctuations, containing concentration information. Existing technologies ignore these temporal features, which is equivalent to discarding more than 50% of the effective information dimensions. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention aims to provide an intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering. This method, based on theoretical models of molecular dynamics and statistical physics, inverts frequency-domain spectral information to a time-domain coherence function, establishing a dual-pool dynamic model of clenbuterol molecules in their "free state-clustered state." This model reveals for the first time the intrinsic correlation between concentration, molecular aggregation state, and coherence decay. In the detection of clenbuterol, it can automatically identify abnormal samples and provide a quality scoring mechanism. Furthermore, this method can be extended to the detection of other small molecules without relying on the specific chemical properties of the substances.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for intelligent mapping of ractopamine concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering, characterized by the following steps: Step 1: Collect spatiotemporal multimodal Raman spectral data of the sample containing clenbuterol to be tested, and construct a four-dimensional Raman spectral data tensor. The sample spatial dimension The total number of points occupied by the sample in the three-dimensional space was determined by marking points. Spatial points; sample measurement time dimension The duration axis was defined by measurement, and a total of Each time point; sample measurement frequency dimension , which is the wavenumber range of the Raman spectrum, total 1 pixel; The sample can be a meat sample or a liquid extract, and a coating can be applied to the sample surface. A grid was created, and data for the z-layer was collected along the depth direction, resulting in a total of [number] points in the 3D space. As the spatial dimension of the sample ; In each spatial location (That is, time series data is collected for each three-dimensional spatial point to obtain the total time window) ,in For single exposure time, Number of consecutive data collections ( (point in time) The sampling interval; Raman spectral wavenumber range is Covering the main characteristic peak areas of ractopamine ; Step 2, for the results obtained in Step 1 Spatial heterogeneity was quantified by analyzing the peak intensities of the Raman spectra at each spatial point to obtain the coefficient of variation. ,pass Determine the distribution of each peak in the Raman spectrum at each spatial point; Preferably, in step 2, in order to more accurately identify the spatial distribution patterns of each peak in the Raman spectrum, the Hurst exponent can be further introduced. Perform auxiliary judgment. For the results obtained in step 1... A spatial intensity sequence is constructed by extracting the peak intensity at a specific wavenumber from a given spatial point. The Hurst exponent was calculated using rescaling range analysis (R / S analysis) for lengths of... Calculate the range of a subinterval. with standard deviation The ratio; based on the scaling law relation Through double log-log regression The Hurst exponent is obtained by solving for the slope. .

[0009] ; Step 3: Perform autocorrelation function calculations for each time point of a single spatial point. Calculate and extract fluctuation memory time Normalized to ACF(0) = 1, where Due to time lag; Step 4: Subtract the strong fluorescence background in the Raman spectrum using the asymmetric weighted least squares method to obtain the purified Raman signal. ;in For Raman spectra, This is the estimated fluorescence background value after n iterations; Step 5: Analyze the Raman peak shape of the clenbuterol characteristic signal obtained in Step 4 from the purified Raman signal. Use the Voigt function to fit and extract frequency domain feature parameters; Step 6: Fit the peak shape obtained in Step 5 using the decay kinetics model of the free state and clustered state of clenbuterol molecules, and establish the clustering degree of the clustered state. With clenbuterol concentration Relationship equation; Step 7: Use the trajectory arc length s to uniquely parameterize the clenbuterol concentration. ; Step 8: Extract multi-domain features by combining the results obtained in steps 2, 3, and 7, and construct feature vectors. ; Step 9, Time-Frequency Consistency Spatial representativeness and time drift Perform a weighted synthesis and score the overall quality. A comprehensive score and uncertainty assessment of the clenbuterol concentration are performed, and a final concentration report is output.

[0010] Based on the above scheme, the coefficient of variation CV in step 2 is calculated by the following formulas: for Raman spectral peak intensity measured at a spatial point ,have: Peak intensity mean ; Standard deviation ; coefficient of variation .

[0011] Highly uniform distribution; Moderately non-uniform; Strong clustering or avoidance.

[0012] Based on the above scheme, the autocorrelation function mentioned in step 3 for: For a time series of a single spatial point ,have ; in, Indicates time average; Due to time lag, For the strength variance; The fluctuation memory time With clenbuterol concentration Establish the following power-law scaling relationship: ; in: Reference concentration The characteristic fluctuation time; This is the dynamic scaling exponent.

[0013] Based on the above scheme, the algorithm formula for the asymmetric weighted least squares method in step 4 is as follows: in: For pixel indexes in the Raman spectrum; for During the nth iteration Weight values ​​for each pixel; For the wave number The intensity of the original Raman spectrum at the location was measured. This is the estimated value of the fluorescence background baseline at this wavenumber. Here, represents the smoothness regularization coefficient; the second derivative term in the formula ensures the smoothness of the fluorescence baseline; the weight update rule is as follows: like ,but ,otherwise , The above subscript i This represents the pixel channel index (or discrete wavenumber point index) of the spectrometer detector. For example, if the spectral range contains 1024 pixels, then... Each i Corresponding to a specific Raman displacement wavenumber ν i .

[0014] Specifically: Raman spectra can be decomposed into: ;

[0015] in Raman signal Fluorescent background, It is noise.

[0016] To address the strong fluorescence interference in meat matrices, the improved ALS iterative algorithm steps are as follows: a) Initialization: Set the fluorescence background estimate to... ( The superscript 0 in the middle represents 0 iterations, i.e., the initial value). b) The nth iteration: Fitting: Use a 5th-order polynomial or cubic spline for fitting. Obtain a smooth curve ;

[0017] Weight Update: Constructing an Asymmetric Weight Matrix ; ; ;

[0018] p = 0.001-0.01 (Assigning extremely low weights to the Raman peak region so that it does not participate in fluorescence fitting) c) Optimize according to the asymmetric weighted least squares algorithm formula above.

[0019] d) Convergence criterion: or number of iterations ; e) Extract pure Raman spectra: ;

[0020] Based on the above scheme, the fitting using the Voigt function in step 5 is shown in the following equation: ;

[0021] The objective function is: ;

[0022] in, Peak amplitude (related to concentration and scattering cross section); The center of the peak ( ); Lorentzian full width at half maximum (FWHM), in units This reflects uniform broadening (coherent attenuation). The full width at half maximum (FWHM) of the Gaussian peak is expressed in units of... This reflects non-uniform broadening (differences in molecular microenvironment). Let be the standard deviation of the i-th data point.

[0023] Based on the above scheme, the clustering degree mentioned in step 6 The equation relating the concentration of ractopamine (C) to the following is: ;

[0024] in This represents the total concentration of clenbuterol molecules in both free and clustered states. For the fluctuation memory time described in step 3 The corrected equilibrium constant for the clustering reaction; the corrected formula is shown below: ;

[0025] in, This is the reference equilibrium constant under standard solution conditions; This represents the environmental coupling coefficient.

[0026] Low concentration limit : (Linear response region) High concentration limit : (Saturation zone) Specific explanation: Clenbuterol molecules exist in two aggregation states in meat matrix: 1) Free state: The state of monomer molecules or fully solvated by biological macromolecules (proteins, water), which are strongly disturbed by solvent fluctuations and have rapid coherence decay; 2) Clustered state: Dimers / polymers formed through hydrogen bonds or π-π stacking, with strong intermolecular coupling extending the coherent lifetime.

[0027] The total coherence function can be expressed as the superposition of the two-state contributions: ;

[0028] Parameter definition: : Free state amplitude (population), range [0,1]; Clustered state amplitude, range [0,1]; : Free state rapid decay rate ( Typical value 0.5-2 ; It is the dephasing time of free-state molecules, also known as the coherence lifetime; : Slow decay rate of clustered state ( Typical values ​​are 0.05-0.3. ; denoted as the decoherence time of the clustered molecules.

[0029] Normalization constraints: (Total population is conserved, and the data is treated as dimensionless, meaning that all molecules are either in a free state or in a clustered state, and the sum of the two is 100%) Time scale separation: ;

[0030] Physical requirement: The coherent lifetime of the clustered state must be longer than that of the free state, typically by a ratio of... .

[0031] Based on the above scheme, the above-mentioned method uses the trajectory arc length s as a unique parameter to quantify the concentration of clenbuterol. As shown in the following formula: For concentration sequence Corresponding parameter points ,in, Defined as the total number of standard concentration gradients. For the index in the above concentration sequence, for the first... Concentration points Its trajectory arc length The calculation formula is as follows: Initial conditions: .

[0032] in: The decay rate is a characteristic. This is another dimension of spectroscopic characteristics (such as peak width or peak intensity); These are the normalized eigenvalues; These are the normalized eigenvalues; To balance the weighting coefficients of the two feature dimensions.

[0033] concentration Considered as "particles" moving in a multidimensional parameter space, their trajectories form a one-dimensional manifold (curve). The concentration is uniquely parameterized by the trajectory arc length s: , where s is the starting point The cumulative path length along the trajectory.

[0034] Based on the above scheme, step 8 specifically involves: The feature vector is constructed as follows: .

[0035] Based on the above scheme, step 9 specifically involves: The overall quality score expression is: ; in Highest weight (0.5): Time-frequency consistency is the core of model effectiveness; Secondly (0.3): Spatial representativeness directly affects the accuracy of concentration; Lower (0.2): Time drift is less common, serving as an auxiliary verification.

[0036] The final concentration report will be in the following format: ; in The nominal concentration is obtained by mapping the trajectory arc length s in step 7; Include factor; This represents the combined standard uncertainty. For repeated measures standard deviation, The standard deviation is introduced by the spatial variation coefficient. The standard deviation of the model fitting residuals.

[0037] The criteria for identifying "abnormal samples" include a comprehensive threshold determination based on the following three dimensions: 1. Spatial anomalies: Coefficient of variation (CV) > 0.5 (indicating extreme heterogeneity or aggregation); 2. Frequency domain anomaly: Voigt fit of the objective function value (in The objective function value defined above. (This is a preset chi-square threshold; this situation indicates severe peak distortion and possible interference from impurity peaks). 3. Dynamic anomaly: autocorrelation function No exponential decay or oscillations were observed (indicating photobleaching or physical movement of the sample during the measurement process).

[0038] When any indicator exceeds the threshold, the system marks the sample as "abnormal" and triggers the low-quality scoring mechanism in step 9.

[0039] The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering described in this invention has the following beneficial effects: Based on theoretical models of molecular dynamics and statistical physics, this method inverts frequency domain spectral information to time domain coherence functions, establishes a dual-pool dynamic model of clenbuterol molecules in "free state-clustered state", and reveals for the first time the intrinsic relationship between concentration and molecular aggregation state and coherence decay; it can automatically identify abnormal samples in the detection of clenbuterol and provide a quality scoring mechanism.

[0040] This method significantly improves sensitivity and measurement range compared to traditional Raman spectroscopy: the detection limit is reduced from 10-50 μg / kg to 0.42 μg / kg, an improvement of 20-100 times, meeting national standards (<0.5 μg / kg); the linear range is extended from 1-2 orders of magnitude to 3-4 orders of magnitude (0.5-500 μg / kg), covering the entire spectrum from trace amounts to high residues. The method achieves a repeatability RSD of <5%, meeting or even exceeding the performance of LC-MS / MS, while reducing detection time to less than 10 minutes, providing a new technical approach for rapid on-site food safety testing. This method can also be extended to the detection of other small molecules, without relying on the specific chemical properties of the substances. Attached Figure Description

[0041] The present invention includes the following figures: Figure 1 This is a technical roadmap of the method described in this invention (data acquisition → time-domain inversion → dual-state fitting → manifold mapping → multi-domain fusion → concentration output). Figure 2 This is a schematic diagram of a dual-cell decay model (fast decay component + slow decay component = total coherence function). Figure 3 Concentration trajectory phase diagram in time-frequency joint characteristic space ( (parameterization of arc length in the vs plane). Figure 4 Flowchart of Hurst exponent adaptive correction (H value determination → strategy selection → concentration calculation); Figure 5 For multi-domain weight adaptive curves ( (The pattern of change with concentration) Figure 6 Comparison of detection performance for trace amounts of clenbuterol (0.5 μg / kg). Detailed Implementation

[0042] The present invention will be further described in detail below with reference to the accompanying drawings.

[0043] To address the aforementioned technical bottlenecks, this invention proposes: To address fluorescence interference, an optimized asymmetric weighted least squares (ALS) method was used for fluorescence baseline subtraction, assigning high weights to the Raman peak region and low weights to the fluorescence flat region, and automatically separating the two types of signals iteratively; at the same time, a near-infrared laser (785 nm) was used to reduce fluorescence excitation efficiency.

[0044] To address the non-uniform distribution of clenbuterol in samples, after initial determination using the coefficient of variation, the Hurst exponent from fractal statistics theory can be further introduced to automatically identify the spatial distribution pattern of the samples (clustered / uniform / avoidance). Based on the H value, an adaptive concentration estimation strategy (peak weighted / arithmetic mean / median robust estimation) can be selected to ensure the representativeness of the estimated values.

[0045] To address the insufficient detection sensitivity, a two-state decay dynamics model was established to extract the population ratio of the clustered states. (Clustering degree) is a core parameter, which is highly sensitive to concentration changes in the low-concentration region. This method overcomes the sensitivity bottleneck of traditional single-parameter methods; at the same time, it constructs a manifold trajectory arc length mapping method to transform multidimensional parameter coupling into a one-dimensional monotonic function, eliminating multiple solutions and expanding the linear range.

[0046] To address the lack of time-domain information, the coherence function G(t) is reconstructed through frequency-time Fourier inversion, and a double exponential decay model is fitted to extract the fast and slow decay rates. These parameters, including amplitude ratio, directly reflect molecular aggregation dynamics; simultaneously, analyzing the time series autocorrelation function extracts fluctuation memory time. Establish a power-law relationship with concentration. For domain information fusion, time-domain (attenuation rate), frequency-domain (peak intensity, peak width), and spatial-domain (Hurst exponent, coefficient of variation) parameters are incorporated into a unified Bayesian framework. The weights of each domain are adaptively adjusted according to the concentration range to achieve synergistic enhancement of multiple information sources. Finally, the reliability of the results is ensured through triple consistency verification (time-frequency consistency χ² test, spatial clustering R-index, and time stationarity ADF test).

[0047] Specifically: A method for intelligent mapping of ractopamine concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering, characterized by the following steps: Step 1: Collect spatiotemporal multimodal Raman spectral data of the sample containing clenbuterol to be tested, and construct a four-dimensional Raman spectral data tensor. The sample spatial dimension The sample was determined by marking points in the three-dimensional space, resulting in a total of K spatial points; the sample measurement time dimension... The time points were defined by measuring the duration axis, totaling T time points; the sample measurement frequency dimension, The range of Raman spectral wavenumbers is defined, with a total of P pixels. The sample can be a meat sample or a liquid extract. Cut the meat sample to be tested into thin slices with a thickness of 2-5 mm, place them on a quartz glass slide, and gently press with a coverslip to smooth the surface, avoiding air bubbles and uneven reflection. For liquid extract samples, take 100 μL and place it in a quartz cuvette.

[0048] Set on the sample surface A grid was created, and data for the z-layer was collected along the depth direction, resulting in a total of [number] points in the 3D space. As the spatial dimension of the sample ; In each spatial location (That is, time series data is collected for each three-dimensional spatial point to obtain the total time window) ,in For single exposure time, Number of consecutive data collections ( (point in time) The sampling interval; A confocal Raman spectrometer was used, and the key parameters were set as follows: Excitation source: 785 nm, near-infrared semiconductor laser Raman spectral wavenumber range is Covering the main characteristic peak areas of ractopamine ( ); In summary, the sample spatial dimension described in step 1 ,common Spatial points; sample measurement time dimension ,common Each time point; sample measurement frequency dimension The wavenumber range of the Raman spectrum. ,common 1 pixel; Step 2: Quantify the spatial heterogeneity of the Raman spectrum peaks at the K spatial points obtained in Step 1 to obtain the coefficient of variation (CV). Use the CV to determine the distribution of each peak in the Raman spectrum at each spatial point. The coefficient of variation CV is calculated using the following formulas: For the Raman spectral peak intensities measured at K spatial points ,have: Peak intensity mean ;

[0049] Standard deviation ;

[0050] coefficient of variation .

[0051] CV < 0.1: Highly uniform distribution; CV = 0.2-0.5: Moderate non-uniformity; CV>0.5: Strong clustering or avoidance.

[0052] The Hurst exponent can be further introduced. Perform auxiliary judgment. For the results obtained in step 1... A spatial intensity sequence is constructed by extracting the peak intensity at a specific wavenumber from a given spatial point. The Hurst exponent was calculated using rescaling range analysis (R / S analysis) for lengths of... Calculate the range of a subinterval. with standard deviation The ratio; based on the scaling law relation Through double log-log regression The Hurst exponent is obtained by solving for the slope. .

[0053] ;

[0054] Step 3: Perform autocorrelation function calculations for each time point of a single spatial point. Calculate and normalize to ACF(0)=1, then extract the fluctuation memory time. Due to time lag; The autocorrelation function for: For a time series of a single spatial point ,have ;

[0055] in, Indicates time average; Due to time lag, The strength variance.

[0056] The fluctuation memory time The following power-law scaling relationship is established between the concentration of ractopamine C and the concentration of clenbuterol: ;

[0057] in: Reference concentration The characteristic fluctuation time; The kinetic scaling exponent (typically 0.3 < 0.3) <0.8, reflecting the degree to which the solvent cage effect restricts molecular diffusion.

[0058] Step 4: Subtract the strong fluorescence background in the Raman spectrum using the asymmetric weighted least squares method to obtain the purified Raman signal. ;in For Raman spectra, This is the estimated fluorescence background value after n iterations; The algorithm formula for the asymmetric weighted least squares method is as follows: ; in: is the smoothness regularization coefficient, where the second derivative term ensures the smoothness of the fluorescence baseline.

[0059] Specifically: Raman spectra can be decomposed into: ;

[0060] in Raman signal Fluorescent background, It is noise.

[0061] To address the strong fluorescence interference in meat matrices, the improved ALS iterative algorithm steps are as follows: a) Initialization: Set the fluorescence background estimate to... ( The superscript 0 in the middle represents 0 iterations, i.e., the initial value). b) The nth iteration: Fitting: Use a 5th-order polynomial or cubic spline for fitting. Obtain a smooth curve ;

[0062] Weight Update: Constructing an Asymmetric Weight Matrix ; ; ;

[0063] p = 0.001-0.01 (Assigning extremely low weights to the Raman peak region so that it does not participate in fluorescence fitting) c) Optimize according to the asymmetric weighted least squares algorithm formula above.

[0064] d) Convergence criterion: or number of iterations ; e) Extract pure Raman spectra:

[0065] Step 5: Analyze the Raman peak shape of the clenbuterol characteristic signal obtained in Step 4 from the purified Raman signal. Use the Voigt function for fitting; Taking clenbuterol hydrochloride, a lean meat enhancer, as an example, its main Raman characteristic peaks are located at: : C=C stretching vibration of benzene ring (strong peak); CN stretching vibration (medium-intensity peak); C-Cl stretching vibration (medium peak).

[0066] This invention mainly analyzes 1597 The peak has less interference and a high signal-to-noise ratio.

[0067] The fitting using the Voigt function is shown in the following equation:

[0068] The objective function is: ;

[0069] in, Peak amplitude (related to concentration and scattering cross section); The center of the peak ( ); Lorentzian full width at half maximum (FWHM), in units This reflects uniform broadening (coherent attenuation). The full width at half maximum (FWHM) of the Gaussian peak is expressed in units of... This reflects non-uniform broadening (differences in molecular microenvironment). Let be the standard deviation of the i-th data point.

[0070] Step 6, as follows Figure 2 As shown, the fitted peak shape obtained in step 5 is fitted using a decay kinetic model of the free state and clustered state of clenbuterol molecules to establish the clustering degree of the clustered state. With clenbuterol concentration Relationship equation; The degree of clustering With clenbuterol concentration The relational equation is: ; in This represents the total concentration of clenbuterol molecules in both free and clustered states. is the equilibrium constant for the clustering reaction.

[0071] However, in complex biological matrices, the equilibrium constant of clustering reactions... It is not a fixed constant, but a variable affected by the viscosity of the local microenvironment. This invention utilizes the fluctuation memory time extracted in step 3. For the above equations Dynamic correction is performed, and the correction formula is as follows: ;

[0072] in, This is the reference equilibrium constant under standard solution conditions; This represents the currently measured fluctuation memory time; Environmental coupling coefficient (empirical value taken as 0.5-1.0); This is the corrected local equilibrium constant.

[0073] The corrected equation for step 6 becomes: ;

[0074] In the low concentration range ( The decrease in the signal-to-noise ratio of Raman spectroscopy leads to direct fitting. The error increases, but the time-domain fluctuation characteristics at this time... It remains significant and sensitive to concentration changes. Through the above correction mechanism, time-domain information is utilized ( Constraint frequency domain model parameters ( This achieves enhanced detection in the time domain assisted by the frequency domain at low concentrations, thus ensuring... Figure 6 The high sensitivity of trace detection is shown.

[0075] Low concentration limit : (Linear response region) High concentration limit : (Saturation zone) Specific explanation: Clenbuterol molecules exist in two aggregation states in meat matrix ("free state-clustered state" dual-pool hypothesis): 1) Free state: The state of monomer molecules or fully solvated by biological macromolecules (proteins, water), which are strongly disturbed by solvent fluctuations and have rapid coherence decay; 2) Clustered state: Dimers / polymers formed through hydrogen bonds or π-π stacking, with strong intermolecular coupling extending the coherent lifetime.

[0076] The total coherence function can be expressed as the superposition of the two-state contributions: ; Parameter definition: : Free state amplitude (population), range [0,1]; Clustered state amplitude, range [0,1]; : Free state rapid decay rate ( Typical value 0.5-2 ; : Slow decay rate of clustered state ( Typical values ​​are 0.05-0.3. .

[0077] Normalization constraints: (Total population is conserved, dimensionless treatment) Time scale separation:

[0078] Physical requirement: The coherent lifetime of the clustered state must be longer than that of the free state, typically by a ratio of... .

[0079] Assuming the clustering reaction involves dimer formation: (Can be extended to multi-product) According to the law of mass action, the equilibrium constant is: ; Derivation of the cluster aggregation degree expression: Let the total concentration monomer concentration

[0080] Then we can obtain: ; Step 7, as follows Figure 3 As shown, the concentration of ractopamine C is uniquely parameterized using the trajectory arc length s, as shown in the following formula: For concentration sequence Corresponding parameter points ,in, Defined as the total number of standard concentration gradients. For the index in the above concentration sequence, for the first... Concentration points Its trajectory arc length The calculation formula is as follows: Initial conditions: .

[0081] Specifically, if N parameters are used simultaneously Establish concentration The idea of ​​geometricization of multiple regression: concentration Considered as "particles" moving in a multidimensional parameter space, their trajectories form a one-dimensional manifold (curve). The concentration is uniquely parameterized by the trajectory arc length s: , where s is the starting point ( ) The cumulative path length along the trajectory.

[0082] Step 8: Extract multi-domain features by combining the results obtained in steps 2, 3, and 7, and construct feature vectors. .

[0083] Step 9, Time-Frequency Consistency Spatial representativeness and time drift A weighted average is performed, and the concentration of clenbuterol is comprehensively scored according to the following formula: ; in Highest weight (0.5): Time-frequency consistency is the core of model effectiveness; Secondly (0.3): Spatial representativeness directly affects the accuracy of concentration; Lower (0.2): Time drift is less common, serving as an auxiliary verification.

[0084] The final concentration report will be in the following format: ; in The nominal concentration is obtained by mapping the trajectory arc length s in step 7; Include factor; This represents the combined standard uncertainty. For repeated measures standard deviation, The standard deviation is introduced by the spatial variation coefficient. The standard deviation of the model fitting residuals.

[0085] To verify the effectiveness of the method described in this invention, the following embodiments and comparative examples were designed.

[0086] Example 1: Blank pork tissue was selected, and clenbuterol hydrochloride standard solution was added to prepare a solution with a concentration of [missing value]. Spiked samples (close to the national standard detection limit). Using the method described in this invention, the spatial peak intensity variation coefficient is first calculated. The sample was identified as being in a moderately non-uniform distribution state; subsequently, asymmetric weighted least squares (ALS) was used to subtract the fluorescence background, improving the signal-to-noise ratio from 2.1 to 15.6; then, a two-state decay kinetic model was employed to further analyze the fluorescence background. Feature peaks are fitted to extract clustering degree. and fluctuation memory time And mapped to the arc length of the manifold trajectory. The final model output predicts the concentration as follows: The relative error is only 6%.

[0087] Comparative Example 1 uses the traditional Raman detection method (single-point peak intensity linear regression method) to detect the same The sample was tested. Because traditional methods cannot effectively handle strong fluorescence interference and uneven spatial distribution at low concentrations, the measured peak intensity signal was extremely weak and unstable. The concentration was calculated as follows: The relative error was as high as 76%, and the result was lower than the detection limit of traditional methods, so it was misjudged as "negative".

[0088] The comparative results show that the present invention significantly improves the accuracy and sensitivity of detection in the trace detection range by integrating spatiotemporal features.

[0089] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A method for intelligent mapping of clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering, characterized in that, Includes the following steps: Step 1: Collect spatiotemporal multimodal Raman spectral data of the sample containing clenbuterol to be tested, and construct a four-dimensional Raman spectral data tensor. ; The sample space dimension The total number of points occupied by the sample in the three-dimensional space was determined by marking points. Spatial points; sample measurement time dimension The duration axis was defined by measurement, and a total of Each time point; sample measurement frequency dimension , which is the wavenumber range of the Raman spectrum, total 1 pixel; Step 2, for the results obtained in Step 1 Spatial heterogeneity was quantified by analyzing the peak intensities of the Raman spectra at each spatial point to obtain the coefficient of variation. ,pass Determine the distribution of each peak in the Raman spectrum at each spatial point; Step 3: Perform autocorrelation function calculations for each time point of a single spatial point. Calculate and extract fluctuation memory time Due to time lag; Step 4: Subtract the strong fluorescence background in the Raman spectrum using the asymmetric weighted least squares method to obtain the purified Raman signal. ;in For Raman spectra, This is the estimated fluorescence background value after n iterations; Step 5: Analyze the Raman peak shape of the clenbuterol characteristic signal obtained in Step 4 from the purified Raman signal. Use the Voigt function to fit and extract frequency domain feature parameters; Step 6: Fit the peak shape obtained in Step 5 using the decay kinetics model of the free state and clustered state of clenbuterol molecules, and establish the clustering degree of the clustered state. With clenbuterol concentration Relationship equation; Step 7, use trajectory arc length Unique parameterized ractopamine concentration ; Step 8: Extract multi-domain features by combining the results obtained in steps 2, 3, and 7, and construct feature vectors. ; Step 9, Time-Frequency Consistency Spatial representativeness and time drift Perform a weighted synthesis and score the overall quality. A comprehensive score and uncertainty assessment of the clenbuterol concentration are performed, and a final concentration report is output.

2. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that: In step 2, the Hurst exponent is introduced. To further determine the distribution pattern of each peak in the Raman spectrum, the formula is as follows: For the result obtained in step 1 A spatial intensity sequence is constructed by extracting the peak intensity at a specific wavenumber from a given spatial point. The Hurst exponent is calculated using rescaling range analysis for lengths of... Calculate the range of a subinterval. with standard deviation The ratio; Based on the scaling law relation Through double log-log regression The Hurst exponent is obtained by solving for the slope. .

3. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, The coefficient of variation (CV) mentioned in step 2 is calculated by the following formulas: for Raman spectral peak intensity measured at a spatial point ,have: Peak intensity mean ; Standard deviation ; coefficient of variation .

4. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, The autocorrelation function mentioned in step 3 for: For a time series of a single spatial point ,have ; in, Indicates time average; Due to time lag, For the strength variance; The fluctuation memory time With clenbuterol concentration Establish the following power-law scaling relationship: ; in: Reference concentration The characteristic fluctuation time; This is the dynamic scaling exponent.

5. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, The algorithm formula for the asymmetric weighted least squares method described in step 4 is as follows: ; in: For pixel indexes in the Raman spectrum; for During the nth iteration Weight values ​​for each pixel; For the wave number The intensity of the original Raman spectrum at the location was measured. This is the estimated value of the fluorescence background baseline at this wavenumber. Here, represents the smoothness regularization coefficient; the second derivative term in the formula ensures the smoothness of the fluorescence baseline; the weight update rule is as follows: like ,but ,otherwise .

6. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, The fitting using the Voigt function described in step 5 is shown in the following equation: ； The objective function is: ; in, For peak amplitude, The center of the peak For the full width of half the Lorentzian peak, The full width at half maximum (FWHM) of the Gaussian peak. For the first The standard deviation of each data point; The variable is the integral variable, representing the frequency detuning.

7. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, The clustering degree mentioned in step 6 With clenbuterol concentration The relational equation is: ； in This represents the total concentration of clenbuterol molecules in both free and clustered states. For the fluctuation memory time described in step 3 The corrected equilibrium constant for the clustering reaction; the corrected formula is shown below: ； in, This is the reference equilibrium constant under standard solution conditions; This represents the environmental coupling coefficient.

8. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 1, characterized in that, Step 7 describes using the trajectory arc length s to uniquely parameterize the clenbuterol concentration. As shown in the following formula: For concentration sequence Corresponding parameter points ,in, Defined as the total number of standard concentration gradients. For the index in the above concentration sequence, for the first... Concentration points Its trajectory arc length The calculation formula is as follows: Initial conditions: ; in: The decay rate is a characteristic. Peak width or peak intensity; These are the normalized eigenvalues; These are the normalized eigenvalues; To balance the weighting coefficients of the two feature dimensions.

9. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 8, characterized in that, Step 8 specifically includes: The feature vector is constructed as follows 。 10. The intelligent mapping method for clenbuterol concentration based on the spatiotemporal dynamic decay characteristics of Raman scattering as described in claim 8, characterized in that, Step 9 specifically includes: The overall quality score expression is: ; The final concentration report will be in the following format: ; In the above formulas: The nominal concentration is obtained by mapping the trajectory arc length s in step 7; Include factor; This represents the combined standard uncertainty. For repeated measures standard deviation, The standard deviation is introduced by the spatial variation coefficient. The standard deviation of the model fitting residuals.

Images