Terahertz metamaterial enhanced spectrum detection system

Abstract

The invention belongs to the technical field of terahertz detection, and discloses a terahertz metamaterial enhanced spectrum detection system. Comprising the following steps: constructing a metamaterial enhancement factor tensor matrix according to the interaction between terahertz waves and a metamaterial; performing local field enhancement and surface plasma resonance optimization on the target sample based on the metamaterial enhancement factor tensor matrix to obtain a feature enhancement spectroscopy mapping function; constructing a multi-dimensional detection feature space based on a feature enhancement spectroscopy mapping function; a spectral signal enhancement topology network is formed based on the multi-dimensional detection feature space; enhancing the topological network based on the spectral signal, and generating an enhanced spectral feature discrimination function; obtaining a high-precision terahertz spectrum detection result based on the enhanced spectrum characteristic discrimination function; according to the invention, internal correlation between spectral characteristics can be mined, and comprehensive description of sample information is enhanced, so that high-precision terahertz spectrum detection is realized.

Description

Technical Field

[0001] The present invention relates to the field of terahertz detection technology. More specifically, the present invention relates to a terahertz metamaterial-enhanced spectroscopic detection system. Background Art

[0002] With the rapid development of terahertz technology, terahertz waves (frequency range of about 0.1 THz to 10 THz) are widely used in non-destructive testing, quality control, security detection, material characterization and other fields due to their strong penetration, non-ionization, non-destructive and other advantages. At the same time, as a new type of artificial material, metamaterials can effectively regulate the absorption, transmission and reflection of waves, especially in the terahertz band, by virtue of their ability to precisely control electromagnetic responses, significantly improving the flexibility and efficiency of terahertz detection systems. However, the current terahertz detection technology still faces some challenges, including unclear interaction mechanisms between terahertz waves and materials, low efficiency of spectral data processing, insufficient system sensitivity and anti-interference ability, and high requirements for sample morphology. Therefore, there is an urgent need for an intelligent terahertz metamaterial spectroscopic detection system to solve the limitations of existing terahertz detection technologies.

[0003] The patent with the publication number CN119574497A discloses a method for detecting terahertz trace molecular fingerprint spectra, including: collecting the time-domain signal of a sample when the sample formed by compounding a sample layer and a base layer is placed in the terahertz wave radiation area; the sample layer is obtained by coating an analyte on a base layer formed based on a selected base material; based on the fast Fourier transform, converting the time-domain signal of the sample into the frequency-domain signal of the sample; obtaining the transmission spectrum according to the frequency-domain signal of the sample and the reference frequency-domain signal obtained by testing air, and extracting the absorption fingerprint signal of the analyte from the transmission spectrum; based on the absorption fingerprint signals of different contents of the analyte, establishing a standard equation between the content of the analyte and the intensity of the fingerprint signal. This detection method has low cost, short detection time and simple operation, and can realize ultra-wide spectrum detection of the analyte.

[0004] However, although the above technology realizes terahertz spectroscopic detection, it relies on traditional Fourier transform and filtering algorithms to process spectral signals, making it difficult to fully meet the requirements for extracting non-linear and non-stationary signal features in complex samples, resulting in limited detection accuracy. In addition, the analysis of spectral data still establishes a standard equation based on a linear model, lacking in-depth exploration of the potential correlations and complex structures between spectral features. Moreover, the above technology has limited processing in terms of signal-to-noise ratio improvement and anti-interference ability, and it is difficult to meet the actual application requirements in higher-precision and more complex environments.

[0005] In view of this, the present invention proposes a terahertz metamaterial-enhanced spectroscopic detection system to solve the above problems. Summary of the Invention

[0006] To overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: A terahertz metamaterial-enhanced spectroscopic detection system, comprising: A matrix construction module, configured to construct a microscopic electromagnetic response model according to the interaction between terahertz waves and the metamaterial, perform resonance characteristic analysis on the microscopic electromagnetic response model, and construct a metamaterial enhancement factor tensor matrix; A function mapping module, configured to perform local field enhancement and surface plasmon resonance optimization on a target sample based on the metamaterial enhancement factor tensor matrix, and obtain a characteristic enhanced spectroscopy mapping function; A space construction module, configured to perform non-linear mode decomposition on the characteristic enhanced spectroscopy mapping function by using Hilbert-Huang transform, extract an eigen-spectrum component set, and construct a multi-dimensional detection feature space; A topology formation module, configured to perform topological invariant calculation and quantum correlation degree analysis on the multi-dimensional detection feature space to form a spectral signal enhancement topological network; A function generation module, configured to perform multi-scale entropy measure evaluation and signal-to-noise ratio adaptive optimization based on the spectral signal enhancement topological network, and generate an enhanced spectral feature discrimination function; An intelligent detection module, configured to perform uncertainty quantification and sensitivity analysis on the enhanced spectral feature discrimination function by combining variational Bayesian inference and sensitivity analysis methods to obtain a high-precision terahertz spectroscopy detection result.

[0007] Further, the method for constructing the metamaterial enhancement factor tensor matrix includes: Preset simulation parameters, and based on the preset simulation parameters, use electromagnetic simulation software to simulate the interaction between terahertz waves and the metamaterial to obtain electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data; based on the simulation parameters, electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data, construct a corresponding microscopic electromagnetic response model; According to the transmission coefficient data and the reflection coefficient data, obtain the transmission coefficient and the reflection coefficient corresponding to each frequency; respectively construct a transmission coefficient curve and a reflection coefficient curve according to the transmission coefficient and the reflection coefficient corresponding to each frequency; mark the frequencies corresponding to the minimum transmission coefficient in the transmission coefficient curve as candidate resonance frequencies; mark the frequencies corresponding to the maximum reflection coefficient in the reflection coefficient curve as candidate resonance frequencies; square the modulus lengths of the transmission coefficient and the reflection coefficient corresponding to each frequency in sequence, and then add them respectively to obtain the energy sum corresponding to each frequency; subtract the energy sum corresponding to each frequency from one respectively to obtain the absorption rate corresponding to each frequency; construct an absorption rate curve according to the absorption rate corresponding to each frequency; mark the frequencies corresponding to the maximum absorption rate in the absorption rate curve as candidate resonance frequencies; Obtain the electromagnetic field distribution data corresponding to each candidate resonance frequency, and calculate the local enhancement factor corresponding to each candidate resonance frequency. The local enhancement factor includes the local electric field enhancement factor and the local magnetic field enhancement factor; compare each local electric field enhancement factor with a preset factor threshold respectively, and mark the candidate resonance frequencies with the local electric field enhancement factor greater than or equal to the factor threshold as resonance frequencies, while the candidate resonance frequencies with the local electric field enhancement factor less than the factor threshold are not marked; construct an enhancement factor tensor matrix corresponding to each resonance frequency according to the local enhancement factor corresponding to each resonance frequency; merge the enhancement factor tensor matrices corresponding to each resonance frequency to construct a metamaterial enhancement factor tensor matrix.

[0008] Further, the method for obtaining the characteristic enhancement spectroscopy mapping function includes: Obtain the basic optical parameters of the target sample. The basic optical parameters include the complex dielectric function, Raman scattering cross-section, molecular dipole moment, and surface molecular density; obtain the incident electric field vector corresponding to each resonance frequency from the simulation parameters, multiply each incident electric field vector by the corresponding enhancement factor tensor matrix to obtain the local field of the sample corresponding to each resonance frequency; divide the square of the modulus length of each sample local field by the square of the modulus length of the corresponding incident electric field vector respectively to obtain the local field enhancement factor of each resonance frequency; mark the resonance frequency with the largest local electric field enhancement factor in the metamaterial enhancement factor tensor matrix as the dominant frequency; obtain the sample characteristic frequency from the complex dielectric function, and calculate the surface plasmon resonance frequency based on the dominant frequency and the sample characteristic frequency; Obtain the incident frequency from the simulation parameters and the Raman shift from the Raman scattering cross-section, and subtract the Raman shift from the incident frequency to obtain the scattering frequency; calculate the local field enhancement factors corresponding to the scattering frequency and the incident frequency respectively using the interpolation algorithm according to the local field enhancement factor of each resonance frequency; obtain the radiative decay rate and the non-radiative decay rate, and add the radiative decay rate and the non-radiative decay rate to obtain the total decay rate; divide the radiative decay rate by the total decay rate to obtain the quantum efficiency factor; multiply the quantum efficiency factor, the local field enhancement factor of the scattering frequency, the local field enhancement factor of the incident frequency, and the Raman scattering cross-section in sequence to obtain the characteristic enhancement spectroscopy mapping function.

[0009] Further, the method for calculating the surface plasmon resonance frequency includes: Square the modulus corresponding to the molecular dipole moment to obtain the molecular square; square the modulus corresponding to the sample local field at the dominant frequency to obtain the sample square; multiply the molecular square, the sample square, and the surface molecular density in sequence, divide by the reduced Planck constant, and then take the square root to obtain the coupling strength; according to the complex dielectric function, fit using the Lorentz model at the dominant frequency to obtain the loss factor; multiply the loss factor by the imaginary unit, add the dominant frequency, and subtract the sample characteristic frequency to obtain the complex detuning factor; square the coupling strength, divide by the complex detuning factor to obtain the frequency shift; add the frequency shift to the dominant frequency to obtain the surface plasmon resonance frequency.

[0010] Further, the method for extracting the set of eigen-spectral components includes: Decompose the feature-enhanced spectroscopy mapping function into eigen-spectral components, and perform Hilbert transform on each eigen-spectral component to obtain the amplitude spectrum and the instantaneous frequency spectrum of each eigen-spectral component, where is an integer greater than 0; calculate the signal-to-noise ratio and contribution degree of each eigen-spectral component in sequence, and compare them with the corresponding component thresholds in the preset threshold set respectively; if both the signal-to-noise ratio and the contribution degree are greater than or equal to the corresponding component thresholds, mark the corresponding eigen-spectral component as an excellent spectral component; the threshold set includes the component thresholds corresponding to the signal-to-noise ratio and the contribution degree; if there is a signal-to-noise ratio or a contribution degree less than the corresponding component threshold, do not mark the corresponding eigen-spectral component; construct a set of eigen-spectral components according to all the excellent spectral components; The method for constructing the multi-dimensional detection feature space includes: For each excellent spectral component in the set of eigen-spectral components, regard the corresponding amplitude spectrum and instantaneous frequency spectrum as a set of components, and the set of components corresponds to the excellent spectral component one by one; input each set of components into the trained feature extraction model respectively to extract spectral feature data, and the feature extraction model is a deep neural network model; construct a multi-dimensional detection feature space according to the spectral feature data corresponding to each excellent spectral component.

[0011] Further, the step of decomposing the feature-enhanced spectroscopy mapping function into eigen-spectral components includes: Step S101: Mark the feature-enhanced spectroscopy mapping function as the mapping function, and obtain all the maximum points and minimum points in the mapping function; Step S102: Construct the upper envelope line using the cubic spline interpolation method according to all the maximum points; Step S103: Construct the lower envelope line using the cubic spline interpolation method according to all the minimum points; Step S104: Take the average value of the upper envelope line and the lower envelope line to obtain the mean curve; Step S105: Subtract the mean curve from the mapping function to obtain candidate spectral components, and determine whether to retain the candidate spectral components; if not retained, use the candidate spectral components as the mapping function and return to Step S101; if retained, subtract the candidate spectral components from the mapping function to obtain residual components, and proceed to Step S106; Step S106: Determine whether the residual components are monotonic functions; if not, use the residual components as the mapping function and return to Step S101; if so, use all the retained candidate spectral components as eigen-spectral components; In the said Step S105, the method for determining whether to retain the candidate spectral components includes: Count the total number of maximum points and minimum points in the candidate spectral components and label it as the number of extreme points; count the number of zero-crossing points in the candidate spectral components and label it as the number of zero points, where the zero-crossing point is the intersection of the candidate spectral components and the zero axis; subtract the number of zero points from the number of extreme points to obtain the difference in quantity; if the difference in quantity is greater than 1, do not retain the candidate spectral components; if the difference in quantity is less than or equal to 1, retain the candidate spectral components; In the said Step S106, the method for determining whether the residual components are monotonic functions includes: Count the total number of maximum points and minimum points in the residual components and label it as the number of residuals; compare the number of residuals with a preset monotonic threshold, if the number of residuals is greater than or equal to the monotonic threshold, the residual components are monotonic functions; if the number of residuals is less than the monotonic threshold, the residual components are not monotonic functions.

[0012] Further, the method for forming the spectral signal enhancement topological network includes: Perform normalization processing on the spectral feature data corresponding to each excellent spectral component in the multi-dimensional detection feature space; project the multi-dimensional detection feature space onto a three-dimensional topological space using a dimensionality reduction algorithm, and the feature points in the three-dimensional topological space correspond one-to-one with the excellent spectral components in the multi-dimensional detection feature space; use a clustering algorithm to cluster the feature points in the three-dimensional topological space to obtain sub-topological spaces, where is an integer greater than 1; in each sub-topological space, form a triangular unit with three adjacent feature points and calculate the Berry curvature of each triangular unit; sequentially add up the Berry curvatures of all triangular units corresponding to each sub-topological space, and then divide by Map each excellent spectral component to a quantum state, and calculate the quantum correlation degree between every two quantum states; preset a correlation threshold, which is preset by those skilled in the art according to the actual situation; compare each quantum correlation degree with the preset correlation threshold respectively; if the quantum correlation degree is greater than or equal to the correlation threshold, take the corresponding two quantum states as a group of quantum sets; if the quantum correlation degree is less than the correlation threshold, do not take the corresponding two quantum states as a group of quantum sets; take each excellent spectral component as a node, take the topological invariant and the normalized spectral feature data of each node as the corresponding node attributes, establish a correlation edge between the nodes corresponding to each group of quantum sets, and the weight of the correlation edge is the quantum correlation degree of the corresponding quantum set; form a spectral signal enhanced topological network according to all the nodes, node attributes and correlation edges.

[0013] Further, the calculation method of the Berry curvature is as follows: randomly label the three feature points in the triangular unit as the first point, the second point and the third point in sequence; calculate the cross product of the second point and the third point to obtain a cross product point; calculate the dot product of the cross product point and the first point to obtain a dot product value; calculate the moduli of the first point, the second point and the third point in sequence and multiply them in sequence to obtain the total modulus; divide the dot product value by the total modulus, and then calculate the Berry curvature through the arctangent function; The calculation method of the Euler characteristic number is as follows: sequentially count the number of vertices, the number of edges and the number of faces in the triangulation network. The number of vertices is the number of feature points, the number of edges is the number of edges between all feature points, and the number of faces is the number of triangular units; subtract the number of edges from the number of vertices and then add the number of faces to obtain the Euler characteristic number; The calculation method of the quantum correlation degree is as follows: calculate the inner product between two quantum states and label it as the similarity; calculate the inner product of each quantum state itself and label it as the norm; multiply the two norms and then take the square root to obtain the norm product; divide the similarity by the norm product to obtain the quantum correlation degree.

[0014] Further, the method for generating the enhanced spectral feature discrimination function includes: Construct an attribute sequence according to the node attributes of each node in the spectral signal enhanced topological network; set a scale value, and based on the scale value, coarsen the attribute sequence into granulation sequences, where Multiply each associated edge by the mean of the node weights of the corresponding two nodes to obtain adjusted edges; preset an edge threshold, compare each adjusted edge with the edge threshold, retain the adjusted edges with values greater than or equal to the edge threshold in the spectral signal enhanced topological network, and delete the adjusted edges with values less than the edge threshold from the spectral signal enhanced topological network; map the spectral signal enhanced topological network into the feature space, construct a feature vector, and perform regularization processing on the feature vector to obtain a regularized processing vector; the feature vector includes all node attributes, multi-scale entropy values, signal-to-noise ratios, and adjusted edges in the spectral signal enhanced topological network; based on the regularized processing vector, construct a feature judgment model using the random forest algorithm, and convert the constructed feature judgment model into an enhanced spectral feature discrimination function.

[0015] Further, the method for obtaining the high-precision terahertz spectrum detection result includes: Use variational Bayesian inference to perform uncertainty quantification on the enhanced spectral feature discrimination function to obtain a confidence interval; use a sensitivity analysis method to perform sensitivity analysis on the enhanced spectral feature discrimination function to obtain a first-order sensitivity index and a total sensitivity index; based on the confidence interval, the first-order sensitivity index and the total sensitivity index, use an optimization algorithm to optimize the model hyperparameters of the feature judgment model; the optimization objective is to minimize the width of the confidence interval while minimizing the prediction error of the model; according to the optimized feature judgment model, obtain an optimized enhanced spectral feature discrimination function and label it as an optimized discrimination function; according to the output result of the optimized discrimination function, obtain the high-precision terahertz spectrum detection result.

[0016] The technical effects and advantages of a terahertz metamaterial enhanced spectral detection system according to the present invention: By constructing a metamaterial enhancement factor tensor matrix, model and analyze the interaction between terahertz waves and the metamaterial, realize the precise optimization of the local field enhancement and surface plasmon resonance of the target sample, and significantly enhance the detection sensitivity of the spectral signal; use Hilbert-Huang transform for nonlinear modal decomposition, extract the eigen-spectral components representing different physical characteristics from the feature-enhanced spectral function, and construct a multi-dimensional detection feature space, which can effectively extract the nonlinear and non-stationary feature signals in complex samples and improve the description ability of the complex structure of the spectral signal; based on the spectral signal enhanced topological network, through topological invariant analysis and quantum correlation degree calculation, explore the internal correlation between spectral features and enhance the comprehensive characterization of sample information; integrate advanced technologies such as multi-scale entropy measure, signal-to-noise ratio adaption, and variational Bayesian inference, specifically enhance the robustness and anti-interference ability of the spectral signal, generate an enhanced spectral feature discrimination function with high robustness and self-adaptability, and realize high-precision terahertz spectrum detection; can effectively meet the actual application requirements in complex environments and high-precision requirement scenarios, and contribute to promoting the wide application of terahertz technology in various fields. Brief Description of the Drawings

[0017] Figure 1 FIG. is a schematic diagram of a terahertz metamaterial-enhanced spectroscopic detection system according to Embodiment 1 of the present invention; Figure 2 FIG. is a flowchart of a terahertz metamaterial-enhanced spectroscopic detection system according to Embodiment 1 of the present invention. Detailed Description of the Embodiments

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1 Please refer to Figure 1 and Figure 2 As shown, a terahertz metamaterial-enhanced spectroscopic detection system described in this embodiment includes a matrix construction module, a function mapping module, a space construction module, a topology formation module, a function generation module, and an intelligent detection module; each module is connected by wired and / or wireless means to achieve data transmission between modules.

[0020] The matrix construction module is used to construct a microscopic electromagnetic response model based on the interaction between terahertz waves and metamaterials, analyze the resonance characteristics of the microscopic electromagnetic response model, and construct a metamaterial enhancement factor tensor matrix.

[0021] The method for constructing the metamaterial enhancement factor tensor matrix includes: Preset simulation parameters, which are pre-set by those skilled in the art according to actual research needs; the simulation parameters include but are not limited to material parameters (such as dielectric constant, permeability, conductivity, etc.), excitation source parameters (such as incident wave type, direction, frequency, etc.), boundary condition parameters (such as periodic boundary, PML, metal boundary, etc.); based on the preset simulation parameters, use electromagnetic simulation software (such as CST Microwave Studio, COMSOL Multiphysics, Lumerical FDTD, etc.) to simulate the interaction between terahertz waves and metamaterials (that is, the electromagnetic field distribution and response characteristics inside and around the metamaterial structure under terahertz wave electromagnetic excitation), and obtain electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data; based on the simulation parameters, electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data, construct a corresponding microscopic electromagnetic response model to describe the local electromagnetic response behavior of the metamaterial unit under terahertz wave excitation; According to the transmission coefficient data and the reflection coefficient data, obtain the transmission coefficient and the reflection coefficient corresponding to each frequency; according to the transmission coefficient and the reflection coefficient corresponding to each frequency, construct a transmission coefficient curve and a reflection coefficient curve respectively; mark the frequencies corresponding to the minimum values of the transmission coefficient in the transmission coefficient curve as candidate resonance frequencies; mark the frequencies corresponding to the maximum values of the reflection coefficient in the reflection coefficient curve as candidate resonance frequencies; square the magnitudes of the transmission coefficient and the reflection coefficient corresponding to each frequency in sequence, and then add them respectively to obtain the energy sum corresponding to each frequency; subtract the energy sum corresponding to each frequency from one to obtain the absorption rate corresponding to each frequency; construct an absorption rate curve according to the absorption rate corresponding to each frequency; mark the frequencies corresponding to the maximum values of the absorption rate in the absorption rate curve as candidate resonance frequencies; Obtain the electromagnetic field distribution data corresponding to each candidate resonance frequency, and calculate the local enhancement factor corresponding to each candidate resonance frequency. The local enhancement factor includes a local electric field enhancement factor and a local magnetic field enhancement factor. It should be noted that the calculation methods of the local electric field enhancement factor and the local magnetic field enhancement factor are both existing technologies, and the specific calculation process will not be elaborated here; compare each local electric field enhancement factor with a preset factor threshold respectively, mark the candidate resonance frequencies with the local electric field enhancement factor greater than or equal to the factor threshold as resonance frequencies, and do not mark the candidate resonance frequencies with the local electric field enhancement factor less than the factor threshold; the factor threshold is preset by those skilled in the art according to the actual situation; construct an enhancement factor tensor matrix corresponding to each resonance frequency according to the local enhancement factor corresponding to each resonance frequency; merge the enhancement factor tensor matrices corresponding to each resonance frequency to construct a metamaterial enhancement factor tensor matrix.

[0022] It should be noted that since both the incident field direction and the local field response direction include , and three directions, there are nine kinds of local electric field enhancement factors and nine kinds of local magnetic field enhancement factors in each enhancement factor tensor matrix, that is the local enhancement factor in the direction generated by the incident in the direction, the local enhancement factor in the direction generated by the incident in the direction, the local enhancement factor in the direction generated by the incident in the direction, the local enhancement factor in the direction generated by the incident in the direction, the local enhancement factor in the Directional local enhancement factor Generated by directional incidence Directional local enhancement factor Generated by directional incidence Directional local enhancement factor

[0023] The function mapping module is used to perform local field enhancement and surface plasmon resonance optimization on the target sample based on the metamaterial enhancement factor tensor matrix, and obtain the characteristic enhanced spectroscopy mapping function

[0024] The target sample is a sample for terahertz metamaterial spectroscopy detection The method for obtaining the characteristic enhanced spectroscopy mapping function includes Obtain the basic optical parameters of the target sample. The basic optical parameters include complex dielectric function, Raman scattering cross-section, molecular dipole moment, and surface molecular density. The complex dielectric function describes the response characteristics of the target sample under the action of electromagnetic fields of different frequencies. The Raman scattering cross-section describes the ability of molecules to generate Raman scattering under specific frequency excitation. The molecular dipole moment describes the asymmetry of the charge distribution in the molecule. The surface molecular density describes the number of target sample molecules per unit area of the metamaterial. The basic optical parameters are obtained by those skilled in the art through experimental measurement or querying scientific literature Obtain the incident electric field vector corresponding to each resonance frequency from the simulation parameters, multiply each incident electric field vector by the corresponding enhancement factor tensor matrix to obtain the local field of the sample corresponding to each resonance frequency. Divide the square of the modulus length of each sample local field by the square of the modulus length of the corresponding incident electric field vector to obtain the local field enhancement factor of each resonance frequency. Mark the resonance frequency with the largest local electric field enhancement factor in the metamaterial enhancement factor tensor matrix as the dominant frequency. Obtain the sample characteristic frequency from the complex dielectric function, and calculate the surface plasmon resonance frequency based on the dominant frequency and the sample characteristic frequency to achieve the surface plasmon resonance optimization of the target sample Obtain the incident frequency from the simulation parameters, obtain the Raman shift from the Raman scattering cross-section, and subtract the Raman shift from the incident frequency to obtain the scattering frequency. According to the local field enhancement factor of each resonance frequency, use interpolation algorithms (such as Hermite interpolation method, Lorentz fitting interpolation method, etc.) to calculate the local field enhancement factors corresponding to the scattering frequency and the incident frequency respectively. Obtain the radiative decay rate and the non-radiative decay rate, add the radiative decay rate and the non-radiative decay rate to obtain the total decay rate. Divide the radiative decay rate by the total decay rate to obtain the quantum efficiency factor. Multiply the quantum efficiency factor, the local field enhancement factor of the scattering frequency, the local field enhancement factor of the incident frequency, and the Raman scattering cross-section in sequence to obtain the characteristic enhanced spectroscopy mapping function. The radiative decay rate and the non-radiative decay rate are obtained by those skilled in the art through electromagnetic simulation of the metamaterial using the finite element or FDTD method

[0025] The method for calculating the surface plasmon resonance frequency includes: Square the modulus length corresponding to the molecular dipole moment to obtain the molecular square; square the modulus length corresponding to the local field of the dominant frequency sample to obtain the sample square; multiply the molecular square, the sample square, and the surface molecular density in sequence, divide by the reduced Planck constant, and then take the square root to obtain the coupling strength; the value of the reduced Planck constant is approximately ; according to the complex dielectric function, use the Lorentz model to fit at the dominant frequency to obtain the loss factor; the Lorentz model is a prior art, and the specific fitting process will not be elaborated here; multiply the loss factor by the imaginary unit, add the dominant frequency, and then subtract the sample characteristic frequency to obtain the complex detuning factor; square the coupling strength and then divide by the complex detuning factor to obtain the frequency shift; add the frequency shift to the dominant frequency to obtain the surface plasmon resonance frequency.

[0026] The spatial construction module is used to perform non-linear modal decomposition on the feature-enhanced spectroscopy mapping function by using the Hilbert-Huang transform, extract the set of intrinsic spectral components, and construct a multi-dimensional detection feature space.

[0027] The method for extracting the set of intrinsic spectral components includes: Decompose the feature-enhanced spectroscopy mapping function into intrinsic spectral components, and perform Hilbert transform on each intrinsic spectral component to obtain the amplitude spectrum and instantaneous frequency spectrum of each intrinsic spectral component, is an integer greater than 0; it should be noted that the Hilbert transform is a prior art, and the specific process will not be elaborated here; calculate the signal-to-noise ratio and contribution degree of each intrinsic spectral component in sequence, and compare them with the corresponding component thresholds in the preset threshold set respectively; if both the signal-to-noise ratio and the contribution degree are greater than or equal to the corresponding component thresholds, mark the corresponding intrinsic spectral component as an excellent spectral component; the threshold set includes the component thresholds corresponding to the signal-to-noise ratio and the contribution degree, and the threshold set is preset by those skilled in the art according to the actual situation; if there is a signal-to-noise ratio or contribution degree less than the corresponding component threshold, do not mark the corresponding intrinsic spectral component; construct the set of intrinsic spectral components according to all the excellent spectral components; among them, each excellent spectral component corresponds to a typical response mode, such as a sharp plasmon-enhanced peak, the characteristic vibration mode of the sample, the influence of low-frequency resonance coupling, etc.

[0028] The method for calculating the signal-to-noise ratio and contribution degree of each intrinsic spectral component includes: The signal part and the noise part in each eigen - spectral component are separated by low - pass filtering. Low - pass filtering is a prior art, and the specific method will not be elaborated here. Calculate the mean value of the signal part corresponding to each eigen - spectral component and label it as the signal mean. Calculate the standard deviation of the noise part corresponding to each eigen - spectral component and label it as the noise standard deviation. Divide the signal mean of each eigen - spectral component by the corresponding noise standard deviation to obtain the signal - to - noise ratio of each eigen - spectral component. Perform Hilbert transform on the feature - enhanced spectroscopy mapping function to obtain the amplitude spectrum and label it as the overall amplitude spectrum. Integrate the amplitude spectrum of each eigen - spectral component and the overall amplitude spectrum in the frequency domain in turn to obtain the eigen - amplitude and the overall amplitude respectively. Divide the eigen - amplitude of each eigen - spectral component by the overall amplitude to obtain the contribution degree of each eigen - spectral component.

[0029] The steps of decomposing the feature - enhanced spectroscopy mapping function into eigen - spectral components include: Step S101: Label the feature - enhanced spectroscopy mapping function as the mapping function, and obtain all the maximum points and minimum points in the mapping function. Step S102: According to all the maximum points, construct the upper envelope line by using the cubic spline interpolation method. The cubic spline interpolation method is a prior art, and the specific process will not be elaborated here. Step S103: According to all the minimum points, construct the lower envelope line by using the cubic spline interpolation method. Step S104: Take the average of the upper envelope line and the lower envelope line to obtain the mean curve. Step S105: Subtract the mean curve from the mapping function to obtain the candidate spectral component, and determine whether to retain the candidate spectral component. If not retained, use the candidate spectral component as the mapping function and return to Step S101. If retained, subtract the candidate spectral component from the mapping function to obtain the residual component, and enter Step S106. Step S106: Determine whether the residual component is a monotonic function. If not, use the residual component as the mapping function and return to Step S101. If so, take all the retained candidate spectral components as eigen - spectral components.

[0030] In the above - mentioned Step S105, the method for determining whether to retain the candidate spectral component includes: Count the total number of maximum points and minimum points in the candidate spectral component and label it as the number of extreme points. Count the number of zero - crossing points in the candidate spectral component and label it as the number of zero points. The zero - crossing point is the intersection point of the candidate spectral component and the zero axis. Subtract the number of zero points from the number of extreme points to obtain the number difference. If the number difference is greater than 1, do not retain the candidate spectral component. If the number difference is less than or equal to 1, retain the candidate spectral component.

[0031] In the above step S106, the method for determining whether the residual component is a monotonic function includes: Count the total number of maximum points and minimum points in the residual component and mark it as the residual quantity; compare the residual quantity with a preset monotonic threshold. If the residual quantity is greater than or equal to the monotonic threshold, the residual component is a monotonic function; if the residual quantity is less than the monotonic threshold, the residual component is not a monotonic function; the monotonic threshold is preset by those skilled in the art according to the actual situation.

[0032] The method for constructing a multi-dimensional detection feature space includes: Centralize the eigen-spectrum components. The amplitude spectrum and instantaneous frequency spectrum corresponding to each excellent spectrum component are used as a set of component sets, and the component sets correspond one-to-one with the excellent spectrum components; input each set of component sets into a trained feature extraction model respectively to extract spectral feature data; the spectral feature data includes peak position, peak width, peak intensity, energy integral, etc.; the feature extraction model is a deep neural network model, and the deep neural network model is a prior art, and the specific training process will not be elaborated here; construct a multi-dimensional detection feature space according to the spectral feature data corresponding to each excellent spectrum component.

[0033] A topology formation module is used to calculate the topological invariant and analyze the quantum correlation degree of the multi-dimensional detection feature space to form a spectral signal enhanced topological network; The method for forming a spectral signal enhanced topological network includes: Perform normalization processing (such as minimum-maximum normalization, Z-score standardization, etc.) on the spectral feature data corresponding to each excellent spectrum component in the multi-dimensional detection feature space to eliminate the influence of dimension and make the features of different dimensions have a unified scale; adopt a dimensionality reduction algorithm (such as principal component analysis method, t-distributed stochastic neighbor embedding method, etc.) to project the multi-dimensional detection feature space onto a three-dimensional topological space, and the feature points in the three-dimensional topological space correspond one-to-one with the excellent spectrum components in the multi-dimensional detection feature space; adopt a clustering algorithm (such as K-means clustering, DBSCAN, etc.) to cluster the feature points in the three-dimensional topological space to obtain sub-topological spaces, where is an integer greater than 1; in each sub-topological space, form a triangle unit by three adjacent feature points and calculate the Berry curvature of each triangle unit; sum up the Berry curvatures of all triangle units corresponding to each sub-topological space in turn, and then divide by to obtain the first Chern number of each sub-topological space; adopt the Delaunay triangulation algorithm to construct the triangulation network of each sub-topological space and calculate the Euler characteristic number of each sub-topological space based on the triangulation network; use the Euler characteristic number and the first Chern number as the topological invariants of each sub-topological space; Map each excellent spectral component to a quantum state, and calculate the quantum correlation degree between every two quantum states; preset a correlation threshold, which is preset by those skilled in the art according to the actual situation; compare each quantum correlation degree with the preset correlation threshold respectively; if the quantum correlation degree is greater than or equal to the correlation threshold, then take the corresponding two quantum states as a group of quantum sets; if the quantum correlation degree is less than the correlation threshold, then do not take the corresponding two quantum states as a group of quantum sets; take each excellent spectral component as a node, take the topological invariant and the normalized spectral feature data of each node as the corresponding node attributes, establish a connection edge between the nodes corresponding to each group of quantum sets, and the weight of the connection edge is the quantum correlation degree of the corresponding quantum set; form a spectral signal enhanced topological network according to all the nodes, node attributes and connection edges; it should be noted that both the Delaunay triangulation algorithm and the method of mapping spectral components to quantum states are prior arts, and the specific processes will not be elaborated here.

[0034] The calculation method of Berry curvature is as follows: randomly label the three feature points in the triangular element, and label them as the first point, the second point and the third point in turn; calculate the cross product of the second point and the third point to obtain the cross product point; calculate the dot product of the cross product point and the first point to obtain the dot product value; calculate the modulus lengths of the first point, the second point and the third point in turn, and multiply them in turn to obtain the total modulus length; divide the dot product value by the total modulus length, and then calculate the Berry curvature through the arctangent function.

[0035] The calculation method of Euler characteristic is as follows: count the number of vertices, the number of edges and the number of faces in the triangulation network in turn. The number of vertices is the number of feature points, the number of edges is the number of edges between all feature points, and the number of faces is the number of triangular elements; subtract the number of edges from the number of vertices and then add the number of faces to obtain the Euler characteristic.

[0036] The calculation method of quantum correlation degree is as follows: calculate the inner product between two quantum states and label it as similarity; calculate the inner product of each quantum state itself respectively and label it as norm; multiply the two norms and then take the square root to obtain the norm product; divide the similarity by the norm product to obtain the quantum correlation degree.

[0037] It should be noted that the calculation methods of cross product, dot product and inner product are all prior arts, and the specific calculation processes will not be elaborated here.

[0038] A function generation module, which is used to perform multi-scale entropy measure evaluation and signal-to-noise ratio adaptive optimization based on the spectral signal enhanced topological network, and generate an enhanced spectral feature discrimination function.

[0039] The method for generating an enhanced spectral feature discrimination function includes: Enhance the node attributes of each node in the spectral signal-enhanced topological network to construct an attribute sequence, that is, each item in the attribute sequence corresponds to a set of node attributes; set a scale value, which is preset by those skilled in the art according to the actual situation; based on the scale value, coarsen the attribute sequence into granulation sequences, where is an integer greater than 1; calculate the multivariate sample entropy of each granulation sequence in turn, multiply it by the weight coefficient in the preset first weight set respectively, and then add them up in turn to obtain the multi-scale entropy value; it should be noted that the calculation method of multivariate sample entropy is prior art, and the specific calculation process will not be elaborated here; calculate the signal-to-noise ratio of each node, multiply the multi-scale entropy, the signal-to-noise ratio of each node, and the topological invariant by the weight coefficients in the preset second weight set respectively, and then add them up in turn to obtain the node weight of each node; it should be understood that the calculation process of the signal-to-noise ratio of each node is the same as that of the signal-to-noise ratio of the above-mentioned eigen-spectral components, and both the first weight set and the second weight set are preset by those skilled in the art according to the actual situation; Multiply each associated edge by the mean of the node weights of the corresponding two nodes to obtain an adjusted edge; preset an edge threshold, which is preset by those skilled in the art according to the actual situation; compare each adjusted edge with the edge threshold, retain the adjusted edges with values greater than or equal to the edge threshold in the spectral signal-enhanced topological network, and delete the adjusted edges with values less than the edge threshold from the spectral signal-enhanced topological network; map the spectral signal-enhanced topological network into the feature space, construct a feature vector, and perform regularization processing on the feature vector to obtain a regularized vector; the feature vector includes all node attributes, multi-scale entropy values, signal-to-noise ratios, and adjusted edges in the spectral signal-enhanced topological network, and the regularization processing is prior art, and the specific process will not be elaborated here; based on the regularized vector, use the random forest algorithm to construct a feature judgment model, and convert the constructed feature judgment model into an enhanced spectral feature discrimination function; among them, the random forest algorithm is prior art, and the specific process will not be elaborated here; The output result of the enhanced spectral feature discrimination function is set by those skilled in the art according to the specific application scenario, including classification tasks (such as identifying different drug components, differentiating different types of plastics, distinguishing organic matter types, etc.), regression tasks (such as predicting the concentration value of a certain compound in a target sample, estimating the purity percentage of a target sample, measuring specific physical parameters of a target sample, etc.), anomaly detection (such as detecting food contamination, identifying drug adulteration, etc.).

[0040] An intelligent detection module, which is used to combine variational Bayesian inference and sensitivity analysis methods to perform uncertainty quantification and sensitivity analysis on the enhanced spectral feature discrimination function to obtain high-precision terahertz spectral detection results.

[0041] Methods for obtaining high-precision terahertz spectral detection results include: Using variational Bayesian inference to quantify the uncertainty of the enhanced spectral feature discrimination function and obtain a confidence interval. A confidence interval is a statistical concept that represents the range of uncertainty in the model prediction results. More specifically, a confidence interval is an interval estimate indicating that there is a certain probability (usually 95%) that the true value falls within this interval. Variational Bayesian inference is a prior art, and the specific process will not be elaborated here. Using sensitivity analysis methods (such as the Sobol method, variance decomposition method, etc.) to perform sensitivity analysis on the enhanced spectral feature discrimination function and obtain the first-order sensitivity index and the total sensitivity index. The first-order sensitivity index represents the contribution of a single variable acting independently to the output variance of the enhanced spectral feature discrimination function. The total sensitivity index represents the overall contribution of a single variable and its interaction with all other variables to the output variance of the enhanced spectral feature discrimination function. The output variance is the statistical variance of the output results generated by the enhanced spectral feature discrimination function, which measures the degree of dispersion of the output results. Based on the confidence interval, the first-order sensitivity index, and the total sensitivity index, use an optimization algorithm (such as the simulated annealing algorithm, genetic algorithm, clonal selection algorithm, etc.) to optimize the model hyperparameters (such as the number of trees, maximum depth, minimum number of samples in leaf nodes, etc.) of the feature judgment model. The optimization goal is to minimize the width of the confidence interval while minimizing the prediction error of the model. According to the optimized feature judgment model, obtain the optimized enhanced spectral feature discrimination function and label it as the optimized discrimination function. According to the output results of the optimized discrimination function, obtain high-precision terahertz spectral detection results.

[0042] In this embodiment, by constructing a metamaterial enhancement factor tensor matrix, the interaction between terahertz waves and the metamaterial is modeled and analyzed to achieve precise optimization of the local field enhancement and surface plasmon resonance of the target sample, significantly enhancing the detection sensitivity of the spectral signal. Using the Hilbert-Huang transform for nonlinear mode decomposition, extract the eigen-spectral components representing different physical characteristics from the feature-enhanced spectral function to construct a multi-dimensional detection feature space, which can effectively extract the nonlinear and non-stationary feature signals in complex samples and improve the description ability of the complex structure of the spectral signal. Based on the spectral signal enhancement topological network, through topological invariant analysis and quantum correlation degree calculation, explore the internal correlation between spectral features and enhance the comprehensive characterization of sample information. Integrate advanced technologies such as multi-scale entropy measure, signal-to-noise ratio adaption, and variational Bayesian inference to enhance the robustness and anti-interference ability of the spectral signal, generate an enhanced spectral feature discrimination function with high robustness and self-adaptability, and achieve high-precision terahertz spectral detection. It can effectively meet the actual application requirements in complex environments and high-precision scenarios, and contribute to the wide application of terahertz technology in various fields.

[0043] Embodiment 2 The present application also provides an electronic device. The electronic device may include one or more processors and one or more memories. Among them, computer-readable code is stored in the memory, and when the computer-readable code is run by one or more processors, it can execute a terahertz metamaterial enhanced spectroscopic detection system as described above.

[0044] The method or system according to the embodiment of the present application can also be implemented by means of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to a network, input / output, a hard disk, etc. The storage device in the electronic device, such as ROM or a hard disk, can store a terahertz metamaterial enhanced spectroscopic detection system provided by the present application. Further, the electronic device may further include a user interface. Of course, the architecture shown in the present application is only exemplary, and when implementing different devices, one or more components shown in the electronic device of the present application may be omitted according to actual needs.

[0045] Embodiment 3 One embodiment of the present application discloses a computer-readable storage medium. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are run by a processor, they can execute a terahertz metamaterial enhanced spectroscopic detection system according to the embodiment of the present application described with reference to the above drawings. The storage medium includes but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory, etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0046] In addition, according to the embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the present application provides a non-transitory machine-readable storage medium, and the non-transitory machine-readable storage medium stores machine-readable instructions, and the machine-readable instructions can be run by a processor to execute instructions corresponding to the method steps provided by the present application, for example: a terahertz metamaterial enhanced spectroscopic detection system. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0047] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0048] It should be noted that, in this document, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the said element.

[0049] In the description of the present invention, it should be understood that the terms "first", "second", etc. are only used for descriptive distinction and cannot be construed as indicating or implying relative importance.

[0050] In the description of the present invention, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0051] In the description of the present invention, the meaning of "several" is one or more, and the meaning of "a large number of" is two or more.

[0052] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials or characteristics described in connection with the said embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0053] For the formulas in this specification, the dimension is removed and only the numerical value is calculated. The formula is obtained by collecting a large amount of data and performing software simulation to get a formula closest to the actual situation. The preset parameters and threshold selection in the formula are set by those skilled in the art according to the actual situation.

[0054] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

Claims

1. A terahertz metamaterial-enhanced spectral detection system, characterized in that, Including: A matrix construction module, which is used to construct a microscopic electromagnetic response model according to the interaction between terahertz waves and metamaterials, analyze the resonance characteristics of the microscopic electromagnetic response model, and construct a metamaterial enhancement factor tensor matrix; A function mapping module, which is used to optimize the local field enhancement and surface plasmon resonance of a target sample based on the metamaterial enhancement factor tensor matrix, and obtain a characteristic enhancement spectroscopy mapping function; A space construction module, which is used to perform nonlinear mode decomposition on the characteristic enhancement spectroscopy mapping function by using the Hilbert-Huang transform, extract an eigen-spectrum component set, and construct a multi-dimensional detection feature space; A topology formation module, which is used to calculate the topological invariant and analyze the quantum correlation degree of the multi-dimensional detection feature space to form a spectral signal enhancement topological network; A function generation module, which is used to perform multi-scale entropy measure evaluation and signal-to-noise ratio adaptive optimization based on the spectral signal enhancement topological network, and generate an enhanced spectral feature discrimination function; An intelligent detection module, which is used to combine the variational Bayesian inference and sensitivity analysis methods to perform uncertainty quantification and sensitivity analysis on the enhanced spectral feature discrimination function, and obtain a high-precision terahertz spectrum detection result.

2. The terahertz metamaterial enhanced spectroscopic detection system according to claim 1, characterized in that, The method for constructing the metamaterial enhancement factor tensor matrix includes: Presetting simulation parameters, and based on the preset simulation parameters, using electromagnetic simulation software to simulate the interaction between terahertz waves and metamaterials to obtain electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data; constructing a corresponding microscopic electromagnetic response model based on the simulation parameters, electromagnetic field distribution data, transmission coefficient data, and reflection coefficient data; According to the transmission coefficient data and the reflection coefficient data, obtaining the transmission coefficient and the reflection coefficient corresponding to each frequency; respectively constructing a transmission coefficient curve and a reflection coefficient curve according to the transmission coefficient and the reflection coefficient corresponding to each frequency; marking the frequencies corresponding to the minimum transmission coefficient in the transmission coefficient curve as candidate resonance frequencies; marking the frequencies corresponding to the maximum reflection coefficient in the reflection coefficient curve as candidate resonance frequencies; squaring the modulus lengths of the transmission coefficient and the reflection coefficient corresponding to each frequency in sequence, and then adding them respectively to obtain the energy sum corresponding to each frequency; subtracting one from the energy sum corresponding to each frequency to obtain the absorption rate corresponding to each frequency; constructing an absorption rate curve according to the absorption rate corresponding to each frequency; marking the frequencies corresponding to the maximum absorption rate in the absorption rate curve as candidate resonance frequencies; Obtain the electromagnetic field distribution data corresponding to each candidate resonance frequency, and calculate the local enhancement factor corresponding to each candidate resonance frequency. The local enhancement factor includes the local electric field enhancement factor and the local magnetic field enhancement factor; compare each local electric field enhancement factor with a preset factor threshold respectively, mark the candidate resonance frequencies whose local electric field enhancement factors are greater than or equal to the factor threshold as resonance frequencies, and do not mark the candidate resonance frequencies whose local electric field enhancement factors are less than the factor threshold; construct an enhancement factor tensor matrix corresponding to each resonance frequency according to the local enhancement factor corresponding to each resonance frequency; merge the enhancement factor tensor matrices corresponding to each resonance frequency to construct a metamaterial enhancement factor tensor matrix.

3. A terahertz metamaterial enhanced spectral detection system according to claim 2, characterized in that, The method for obtaining the characteristic enhancement spectroscopy mapping function includes: Obtain the basic optical parameters of the target sample. The basic optical parameters include the complex dielectric function, Raman scattering cross-section, molecular dipole moment, and surface molecular density; obtain the incident electric field vector corresponding to each resonance frequency from the simulation parameters, multiply each incident electric field vector by the corresponding enhancement factor tensor matrix to obtain the local field of the sample corresponding to each resonance frequency; divide the square of the modulus length of each sample local field by the square of the modulus length of the corresponding incident electric field vector to obtain the local field enhancement factor of each resonance frequency; mark the resonance frequency with the largest local electric field enhancement factor in the metamaterial enhancement factor tensor matrix as the dominant frequency; obtain the sample characteristic frequency from the complex dielectric function, and calculate the surface plasmon resonance frequency based on the dominant frequency and the sample characteristic frequency; Obtain the incident frequency from the simulation parameters and the Raman shift from the Raman scattering cross-section, and subtract the Raman shift from the incident frequency to obtain the scattering frequency; calculate the local field enhancement factors corresponding to the scattering frequency and the incident frequency respectively using the interpolation algorithm according to the local field enhancement factor of each resonance frequency; obtain the radiative decay rate and the non-radiative decay rate, add the radiative decay rate and the non-radiative decay rate to obtain the total decay rate; divide the radiative decay rate by the total decay rate to obtain the quantum efficiency factor; multiply the quantum efficiency factor, the local field enhancement factor of the scattering frequency, the local field enhancement factor of the incident frequency, and the Raman scattering cross-section in sequence to obtain the characteristic enhancement spectroscopy mapping function.

4. A terahertz metamaterial enhanced spectral detection system according to claim 3, characterized in that, The method for calculating the surface plasmon resonance frequency includes: Square the modulus length corresponding to the molecular dipole moment to obtain the molecular square; square the modulus length of the local field of the sample corresponding to the dominant frequency to obtain the sample square; multiply the molecular square, the sample square, and the surface molecular density in sequence, divide by the reduced Planck constant, and then take the square root to obtain the coupling strength; perform a Lorentz model fitting at the dominant frequency according to the complex dielectric function to obtain the loss factor; multiply the loss factor by the imaginary unit, add the dominant frequency, and then subtract the sample characteristic frequency to obtain the complex detuning factor; square the coupling strength and divide by the complex detuning factor to obtain the frequency shift; add the frequency shift to the dominant frequency to obtain the surface plasmon resonance frequency.

5. A terahertz metamaterial enhanced spectroscopic detection system according to claim 4, characterized in that, The method for extracting the set of eigen-spectral components includes: Decompose the feature-enhanced spectroscopy mapping function into intrinsic spectral components, and perform Hilbert transform on each intrinsic spectral component to obtain the amplitude spectrum and instantaneous frequency spectrum of each intrinsic spectral component, where \(n\) is an integer greater than 0; calculate the signal-to-noise ratio and contribution degree of each intrinsic spectral component in turn, and compare them with the corresponding component thresholds in the preset threshold set respectively; if both the signal-to-noise ratio and the contribution degree are greater than or equal to the corresponding component thresholds, mark the corresponding intrinsic spectral component as an excellent spectral component; the threshold set includes the component thresholds corresponding to the signal-to-noise ratio and the contribution degree; if there is a signal-to-noise ratio or contribution degree less than the corresponding component threshold, do not mark the corresponding intrinsic spectral component; construct an intrinsic spectral component set according to all the excellent spectral components; The method for constructing the multi-dimensional detection feature space includes: Concentrate the intrinsic spectral components. The amplitude spectrum and the instantaneous frequency spectrum corresponding to each excellent spectral component are used as a set of component sets, and the component sets correspond one by one to the excellent spectral components. Input each set of component sets into the trained feature extraction model respectively to extract spectral feature data. The feature extraction model is a deep neural network model. Construct a multi-dimensional detection feature space according to the spectral feature data corresponding to each excellent spectral component.

6. The terahertz metamaterial enhanced spectral detection system according to claim 5, characterized in that The step of decomposing the feature-enhanced spectroscopy mapping function into intrinsic spectral components includes: Step S101: Mark the feature-enhanced spectroscopy mapping function as the mapping function, and obtain all the maximum points and minimum points in the mapping function. Step S102: Construct the upper envelope line by using the cubic spline interpolation method according to all the maximum points. Step S103: Construct the lower envelope line by using the cubic spline interpolation method according to all the minimum points. Step S104: Take the average value of the upper envelope line and the lower envelope line to obtain the mean curve. Step S105: Subtract the mean curve from the mapping function to obtain the candidate spectral component, and determine whether to retain the candidate spectral component. If not retained, use the candidate spectral component as the mapping function and return to Step S101. If retained, subtract the candidate spectral component from the mapping function to obtain the residual component, and enter Step S106. Step S106: Determine whether the residual component is a monotonic function; if not, use the residual component as the mapping function and return to Step S101; if so, use all the remaining candidate spectral components as the eigen spectral components; In the said Step S105, the method for determining whether to retain the candidate spectral component includes: Count the total number of maximum points and minimum points in the candidate spectral component and mark it as the number of extreme points. Count the number of zero-crossing points in the candidate spectral component and mark it as the number of zero points. The zero-crossing point is the intersection point of the candidate spectral component and the zero axis. Subtract the number of zero points from the number of extreme points to obtain the difference in quantity. If the difference in quantity is greater than 1, do not retain the candidate spectral component. If the difference in quantity is less than or equal to 1, retain the candidate spectral component. In the said Step S106, the method for determining whether the residual component is a monotonic function includes: Count the total number of maximum points and minimum points in the residual component and mark it as the number of residuals. Compare the number of residuals with the preset monotonic threshold. If the number of residuals is greater than or equal to the monotonic threshold, the residual component is a monotonic function. If the number of residuals is less than the monotonic threshold, the residual component is not a monotonic function.

7. The terahertz metamaterial enhanced spectral detection system according to claim 6, wherein The method for forming the spectral signal enhancement topological network includes: In the multi-dimensional detection feature space, the spectral feature data corresponding to each excellent spectral component are normalized; a dimensionality reduction algorithm is used to project the multi-dimensional detection feature space onto a three-dimensional topological space, and the feature points in the three-dimensional topological space correspond one-to-one with the excellent spectral components of the multi-dimensional detection feature space; a clustering algorithm is used to cluster the feature points in the three-dimensional topological space to obtain sub-topological spaces, where is an integer greater than 1; in each sub-topological space, three adjacent feature points are formed into a triangular unit, and the Berry curvature of each triangular unit is calculated; the Berry curvatures of all triangular units corresponding to each sub-topological space are added in sequence and then divided by to obtain the first Chern number of each sub-topological space; the Delaunay triangulation algorithm is used to construct the triangulation network of each sub-topological space, and the Euler characteristic number of each sub-topological space is calculated based on the triangulation network; the Euler characteristic number and the first Chern number are used as the topological invariants of each sub-topological space; Map each excellent spectral component to a quantum state, and calculate the quantum correlation degree between every two quantum states. Preset the correlation threshold, which is preset by those skilled in the art according to the actual situation. Compare each quantum correlation degree with the preset correlation threshold respectively. If the quantum correlation degree is greater than or equal to the correlation threshold, use the corresponding two quantum states as a set of quantum sets. If the quantum correlation degree is less than the correlation threshold, do not use the corresponding two quantum states as a set of quantum sets. Take each excellent spectral component as a node, take the topological invariant and the normalized spectral feature data of each node as the corresponding node attributes, establish an associated edge between the nodes corresponding to each set of quantum sets, and the weight of the associated edge is the quantum correlation degree of the corresponding quantum set. Form a spectral signal enhancement topological network according to all the nodes, node attributes and associated edges.

8. A terahertz metamaterial enhanced spectroscopic detection system according to claim 7, characterized in that, The calculation method of the Berry curvature is as follows: randomly label the three feature points in the triangular element, successively label them as the first point, the second point, and the third point; calculate the cross product of the second point and the third point to obtain the cross product point; calculate the dot product of the cross product point and the first point to obtain the dot product value; successively calculate the magnitudes of the first point, the second point, and the third point, and multiply them successively to obtain the total magnitude; divide the dot product value by the total magnitude, and then calculate the Berry curvature through the arctangent function; The calculation method of the Euler characteristic number is as follows: successively count the number of vertices, the number of edges, and the number of faces in the triangulation network. The number of vertices is the number of feature points, the number of edges is the number of edges between all feature points, and the number of faces is the number of triangular elements; subtract the number of edges from the number of vertices and then add the number of faces to obtain the Euler characteristic number; The calculation method of the quantum correlation degree is as follows: calculate the inner product between two quantum states and label it as the similarity; calculate the inner product of each quantum state itself and label it as the norm; multiply the two norms and then take the square root to obtain the norm product; divide the similarity by the norm product to obtain the quantum correlation degree.

9. A terahertz metamaterial enhanced spectral detection system according to claim 8, characterized in that, The method for generating the enhanced spectral feature discrimination function includes: Enhance the node attributes of each node in the spectral signal enhanced topological network to construct an attribute sequence; set a scale value, and based on the scale value, coarsen the attribute sequence into granulation sequences, where is an integer greater than 1; calculate the multi-variable sample entropy of each granulation sequence in turn, multiply it by the weight coefficient in the preset first weight set respectively, and then add them up in turn to obtain the multi-scale entropy value; calculate the signal-to-noise ratio of each node, multiply the multi-scale entropy, the signal-to-noise ratio of each node, and the topological invariant by the weight coefficient in the preset second weight set respectively, and then add them up in turn to obtain the node weight of each node; multiply each associated edge by the mean of the node weights of the corresponding two nodes to obtain the adjusted edge; preset an edge threshold, compare each adjusted edge with the edge threshold, retain the adjusted edges with values greater than or equal to the edge threshold in the spectral signal enhancement topological network, and delete the adjusted edges with values less than the edge threshold from the spectral signal enhancement topological network; map the spectral signal enhancement topological network to the feature space, construct a feature vector, and perform regularization processing on the feature vector to obtain the regularized processing vector; the feature vector includes all node attributes, multi-scale entropy values, signal-to-noise ratios, and adjusted edges in the spectral signal enhancement topological network; based on the regularized processing vector, use the random forest algorithm to construct a feature judgment model, and convert the constructed feature judgment model into an enhanced spectral feature discrimination function.

10. A terahertz metamaterial enhanced spectral detection system according to claim 9, characterized in that, The method for obtaining the high-precision terahertz spectrum detection result includes: using variational Bayesian inference to perform uncertainty quantification on the enhanced spectral feature discrimination function to obtain the confidence interval; using the sensitivity analysis method to perform sensitivity analysis on the enhanced spectral feature discrimination function to obtain the first-order sensitivity index and the total sensitivity index; based on the confidence interval, the first-order sensitivity index, and the total sensitivity index, use the optimization algorithm to optimize the model hyperparameters of the feature judgment model; the optimization goal is to minimize the width of the confidence interval while minimizing the prediction error of the model; according to the optimized feature judgment model, obtain the optimized enhanced spectral feature discrimination function and label it as the optimized discrimination function; according to the output result of the optimized discrimination function, obtain the high-precision terahertz spectrum detection result.

Images