A method and system for detecting trace naphthaleneacetic acid based on terahertz spectroscopy

Abstract

This invention provides a method and system for detecting trace naphthaleneacetic acid (NAA) based on terahertz spectroscopy. The method includes determining the target parameters of a target metamaterial sensor, performing simulation based on the target parameters to obtain a target terahertz material, acquiring the NAA sample's spectrum through the target terahertz material to obtain target spectral data, training an initial model using a target feature selection method, and performing dimensionality reduction on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data, optimizing the model parameters of the initial model, and further optimizing the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model, and achieving the residual detection of trace NAA through the target detection model. This invention helps improve the accuracy of pesticide residue detection and also provides a new approach for optimizing terahertz spectral data processing methods.

Description

Technical Field

[0001] This invention belongs to the technical field of pesticide residue detection, specifically relating to a method and system for detecting trace naphthaleneacetic acid based on terahertz spectroscopy. Background Technology

[0002] 1-Naphthaleneacetic acid (NAA) is a synthetic plant growth regulator widely used in agricultural production, primarily to promote plant growth, regulate physiological metabolic processes, and increase crop yield. NAA belongs to the auxin class of substances, and its mechanism of action mainly involves mimicking the function of natural auxin (IAA), promoting plant cell elongation, inducing adventitious root formation, and to some extent delaying fruit drop. In fruit tree cultivation, NAA is often used to preserve fruit, promote rooting of cuttings, and extend the harvest period, thus helping to improve the marketability and storage resistance of crops. However, as a synthetic plant hormone, improper use or excessive dosage of NAA can easily lead to residues on the surface or in the tissues of fruits and vegetables, potentially impacting human health with long-term consumption. Although NAA has relatively low toxicity, studies have shown that it may possess potential reproductive toxicity, endocrine disruption effects, and adverse effects on liver and kidney function. Therefore, many countries and regions have imposed strict maximum residue limits (MRLs) on NAA in agricultural products to ensure food safety. For example, my country's GB 2763-2021 standard for maximum residue limits of pesticides in food stipulates that the maximum residue limit of NAA in apples is 0.1 mg / kg. Since NAA is usually present in low concentrations on the surface of fruits and vegetables, its detection faces technical challenges due to insufficient sensitivity and selectivity. Therefore, developing a simple, rapid, accurate, and highly sensitive NAA residue detection technology is of significant practical importance for ensuring food safety and public health.

[0003] Terahertz waves typically refer to electromagnetic waves with frequencies ranging from 0.1 to 10 THz, located between microwaves and infrared light. This untapped portion of the electromagnetic spectrum is often referred to as the "terahertz blank area" or the "intersection of electronics and photonics." With advancements in materials science, microelectronics, and laser devices, terahertz technology has experienced rapid development in recent years, demonstrating immense application potential in non-destructive testing, bio-imaging, security inspection, communications, and spectral analysis. Its high sensitivity, non-contact nature, and strong penetration make it a significant research direction in cutting-edge interdisciplinary technologies. Compared to visible and infrared light, terahertz waves exhibit excellent penetration into non-polar materials (such as plastics, paper, wood, ceramics, and fabrics), enabling imaging and detection of the internal structures of non-metallic or low-moisture materials. Therefore, terahertz technology possesses unique advantages in non-destructive testing, security inspection, and cultural relic preservation.

[0004] However, pesticide residues are often present in extremely small quantities. Traditional terahertz spectroscopy, when detecting trace pesticide residues, often suffers from weak signals due to the weak interaction between terahertz waves and the substance, resulting in insufficient detection sensitivity. Furthermore, terahertz spectroscopy is sensitive to moisture and other biological components, which can introduce interference signals and affect the accurate identification of trace substances. Therefore, although terahertz spectroscopy has shown good resolution in the detection of some samples, its resolving power and selectivity remain limited in the case of trace substances. Thus, for the detection of trace substances, traditional terahertz spectroscopy often needs to be combined with other techniques to improve its sensitivity and accuracy.

[0005] Metamaterials are artificial materials that achieve specific electromagnetic responses through the design and periodic construction of microscopic structural units. Their properties do not originate from the intrinsic properties of the constituent materials, but are determined by their geometry and arrangement. Compared to natural materials, metamaterials can exhibit a series of unconventional optical phenomena in specific frequency bands, such as negative refractive index, superlens effect, electromagnetic shielding, and perfect absorption, thereby enabling precise control of electromagnetic wave propagation. Especially in the terahertz band, metamaterials, due to their size and structure matching the wavelength, can effectively enhance the localization of the electric field and improve the coupling efficiency with matter, becoming one of the key means to improve the functional performance of terahertz waves. In terahertz detection, metamaterials enhance the absorption or transmission response at specific frequencies through resonant structures, which can significantly improve the system's ability to detect weak signals and its detection sensitivity. Ahmed F et al. designed a magnetic resonance sensor with a high absorption rate of 99.43% at the resonant frequency for the detection of trace pesticide residues. Chowdhury et al. designed a novel metasurface (MS) sensor with a refractive index (RI) sensitivity of 0.125 THz / RIU and quality factors of 5.262 and 4.084, respectively, for the detection of imidacloprid. Du X et al. designed a bimodal flexible metamaterial composed of a rectangular ring-rectangular planar array, with absorbances of 94.06% and 97.07%, respectively, achieving high-sensitivity detection of lambda-cyhalothrin with a detection limit of 0.01 mg / L. These studies all demonstrate the feasibility of metamaterials in pesticide residue detection.

[0006] Therefore, to address the issues of insufficient sensitivity and selectivity in existing technologies, a trace naphthaleneacetic acid detection method based on terahertz spectroscopy is proposed. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a method and system for detecting trace naphthaleneacetic acid based on terahertz spectroscopy, which solves the technical problems in the prior art.

[0008] In a first aspect, the present invention provides the following technical solution: a method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy, comprising:

[0009] The target parameters of the target metamaterial sensor are determined based on the characteristic peak frequency of naphthaleneacetic acid, and simulation is performed based on the target parameters to obtain the target terahertz material.

[0010] A naphthaleneacetic acid (NAA) sample was prepared, and the NAA sample was subjected to spectral acquisition using the target terahertz material to obtain target spectral data.

[0011] An initial model is constructed, and the initial model is trained using a target feature selection method. The target spectral data is then dimensionality-reduced based on the trained initial model to obtain dimensionality-reduced spectral data.

[0012] The model parameters of the initial model are optimized using a target parameter optimization method to obtain optimized parameters. The initial model is then optimized based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model.

[0013] The target detection model enables the detection of trace amounts of naphthaleneacetic acid residues.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: The detection framework of this invention, which integrates L-type terahertz metamaterial sensor and RFE-AOO-SVR intelligent modeling method, not only realizes highly sensitive quantitative analysis of trace NAA residues, but also verifies the feasibility and superiority of the AOO optimization algorithm in the field of spectral detection, providing a practical and scalable research path for terahertz spectroscopy technology in the field of rapid detection of pesticide residues.

[0015] Preferably, the step of determining the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and performing simulation based on the target parameters to obtain the target terahertz material includes:

[0016] Obtain the characteristic peak frequency of naphthaleneacetic acid, and determine the target peak frequency point of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid;

[0017] The quality factor and sensitivity of the target terahertz material are calculated by simulating the spectrum, and the parameters of the preset database are scanned according to the target peak frequency point, the quality factor and the sensitivity to obtain the target parameters of the target metamaterial sensor.

[0018] A single periodic structure is selected as the modeling object, and modeling and simulation are performed according to the target parameters to obtain the target terahertz material.

[0019] Preferably, the target peak frequency points include 2.19 THz and 2.70 THz.

[0020] Preferably, the target terahertz material has an L-shaped composite bimodal structure, and the target parameters include: period P1: 58 μm; length of the rectangular edge of the structure L1: 32 μm; length of the long side of the L-shaped structure L2: 22 μm; length of the short side of the L-shaped structure L3: 17 μm; spacing between the long sides of different L-shaped structures L4: 14 μm; spacing between the short sides of different L-shaped structures L5: 4 μm; width of the rectangular edge of the structure W1: 2 μm; width of the L-shaped structure W2: 4 μm. The substrate material of the target terahertz material is silicon with a refractive index of 3.335, and the surface metal layer is gold.

[0021] Preferably, the step of preparing the naphthaleneacetic acid sample and acquiring the spectral data of the naphthaleneacetic acid sample using the target terahertz material includes:

[0022] NAA standard solution was selected as the reference solution. Ultrapure deionized water was used to quantitatively dilute the reference solution using a high-precision pipette. Each sample was mixed at 3000 rpm for 3 minutes using a vortex mixer to ensure that the solute molecules were uniformly distributed in the solvent. Several sets of concentration gradient samples were transferred to brown volumetric flasks and sealed for storage. Dry air was continuously circulated into the optical cavity of the spectrometer using an air compressor beforehand. Temperature and humidity sensors were used to monitor and ensure that the relative humidity in the cavity was ≤10% and the temperature was stable within the range of 25±0.5℃. After the environment stabilized, 20 μl of different concentration gradient samples were quantitatively aspirated from low concentration to high concentration using a pipette and dropped onto the surface of the metamaterial sensor. A composite sampling mode of 5 points × 10 repeated measurements was used to obtain the target spectral data.

[0023] Preferably, the step of training the initial model using a target feature selection method and performing dimensionality reduction processing on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data includes:

[0024] The target spectral data is initialized into a feature set, which includes several feature subsets. This feature set is then input into the initial model for prediction output to obtain predicted values. :

[0025] ;

[0026] In the formula, Let the feature weights be the feature weights of the j-th feature subset. For the i-th feature of the j-th feature subset, The intercept is... The number of features in the feature subset;

[0027] The feature subsets are sequentially input into the initial model for training, and the feature weights of each feature are output in each training round. The features in the feature subset are sorted in descending order according to the absolute value of the feature weights, and the features with the lowest absolute value of the feature weights are removed to obtain a new feature subset.

[0028] The new feature subset is re-input into the initial model, and the process of feature weight output and feature removal is repeated until the dimensionality reduction requirement is met, so as to obtain dimensionality-reduced spectral data.

[0029] Preferably, the step of optimizing the model parameters of the initial model using the target parameter optimization method to obtain optimized parameters, and then optimizing the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain the target detection model includes:

[0030] The model parameters of the initial model are initialized to the target population, and each individual in the target population is initialized:

[0031] ;

[0032] In the formula, For the initialized individual, , These are the lower bound and upper bound of the variable, respectively. It is a uniform random vector;

[0033] The initial individual samples simulate the periodic shifting behavior of oat stems driven by wind, guiding each sample to swing towards the current optimal solution:

[0034] ;

[0035] In the formula, As the oscillation amplitude control factor, The oscillation frequency is... For phase perturbation, As the current globally optimal individual, , These are the individuals after the t-th and t+1-th iterations, respectively;

[0036] Add perturbation update based on the individual after the (t+1)th iteration To obtain updated individuals:

[0037] ;

[0038] In the formula, , These are two randomly selected individuals. For disturbance adjustment factor;

[0039] During the iteration process, boundary correction is performed on each updated individual and the fitness of each updated individual before and after the iteration is compared, and individuals with lower fitness are retained.

[0040] The iterative process is repeated until the iteration stopping condition is met, and the optimal individual is output to obtain the optimized parameters. Based on the optimized parameters and the dimensionality-reduced spectral data, the initial model is optimized to obtain the target detection model.

[0041] Secondly, the present invention provides the following technical solution: a trace naphthaleneacetic acid detection system based on terahertz spectroscopy detection technology, the system comprising:

[0042] The materials module is used to determine the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and to perform simulation based on the target parameters to obtain the target terahertz material.

[0043] The acquisition module is used to prepare a naphthaleneacetic acid sample and to acquire the spectral data of the naphthaleneacetic acid sample through the target terahertz material.

[0044] The dimensionality reduction module is used to construct an initial model, train the initial model using a target feature selection method, and perform dimensionality reduction processing on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data.

[0045] An optimization module is used to optimize the model parameters of the initial model based on a target parameter optimization method to obtain optimized parameters, and to optimize the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model.

[0046] The detection module is used to detect trace amounts of naphthaleneacetic acid residues using the target detection model.

[0047] Thirdly, the present invention provides the following technical solution: a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the trace naphthaleneacetic acid detection method based on terahertz spectroscopy detection technology as described above.

[0048] Fourthly, the present invention provides the following technical solution: a storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the above-described method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of a trace naphthaleneacetic acid detection method based on terahertz spectroscopy provided in Embodiment 1 of the present invention;

[0051] Figure 2 A calculation diagram of the quality factor Q value parameter provided in Embodiment 1 of the present invention;

[0052] Figure 3 The graph showing the calculation of the sensitivity S-parameter provided in Embodiment 1 of the present invention;

[0053] Figure 4 This is a diagram showing the dimensional parameters of the target terahertz material provided in Embodiment 1 of the present invention;

[0054] Figure 5 This is a three-dimensional structural diagram of the target terahertz material provided in Embodiment 1 of the present invention;

[0055] Figure 6 Hotspot distribution diagrams of the target terahertz material provided in Embodiment 1 of the present invention at 2.19 THz and 2.7 THz, respectively;

[0056] Figure 7 The prediction performance evaluation diagrams for each model provided in Embodiment 1 of the present invention are shown.

[0057] Figure 8 This is a fitting graph between the true value and the predicted value of the target detection model provided in Embodiment 1 of the present invention;

[0058] Figure 9 This is a structural block diagram of the trace naphthaleneacetic acid detection system based on terahertz spectroscopy detection technology provided in Embodiment 2 of the present invention;

[0059] Figure 10 This is a schematic diagram of the hardware structure of a computer provided for another embodiment of the present invention.

[0060] The embodiments of the present invention will be further described below with reference to the accompanying drawings. Detailed Implementation

[0061] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain embodiments of the present invention, and should not be construed as limiting the present invention.

[0062] Example 1

[0063] In Embodiment 1 of the present invention, as Figure 1 As shown, a method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy includes:

[0064] S1. Determine the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and perform simulation based on the target parameters to obtain the target terahertz material;

[0065] Step S1 includes:

[0066] S11. Obtain the characteristic peak frequency of naphthaleneacetic acid, and determine the target peak frequency point of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid.

[0067] Specifically, naphthaleneacetic acid (NAA) has characteristic peak frequencies in the terahertz band of 1.66, 1.78, 1.92, 2.04, 2.19, and 2.70 THz. To improve the sensitivity of terahertz spectroscopy in detecting NAA through resonance effects, the characteristic peak frequencies inherent in the substance are usually selected as the peak frequencies for metamaterial design (i.e., selected from 1.66, 1.78, 1.92, 2.04, 2.19, and 2.70 THz).

[0068] In addition, a phenomenon exists in metamaterial design: if the frequencies of two characteristic peaks are close together in software simulation, the actual measured peaks after fabrication may merge into one due to insufficient instrument resolution or power. Therefore, to avoid this phenomenon while ensuring the metamaterial achieves the requirement of "improved sensitivity," we selected characteristic peaks with a frequency difference greater than 0.5 THz as our design premise. In this study, we selected the 2.19 and 2.70 THz peak positions inherent in naphthaleneacetic acid for design.

[0069] By analyzing the spectrum of naphthaleneacetic acid, the characteristic peak frequencies of naphthaleneacetic acid can be obtained. By selecting two frequencies with a phase difference greater than 0.50 THz from the characteristic peak frequencies, the target peak frequency point can be obtained.

[0070] S12. Calculate the quality factor and sensitivity of the target terahertz material through simulation spectrum, and perform parameter scanning on the preset database according to the target peak frequency point, the quality factor and the sensitivity to obtain the target parameters of the target metamaterial sensor.

[0071] Specifically, the formulas for calculating the quality factor Q and sensitivity S are as follows:

[0072] ; ;

[0073] Sensitivity (S) is a key indicator that measures a structure's ability to respond to changes in external physical or chemical parameters. It is often used to describe a structure's performance in sensing changes in refractive index, concentration, or thickness. Sensitivity is typically defined as the resonant frequency shift (S / S). ) and the physical quantity that caused the change (such as the change in refractive index, The ratio of 1 / 2 GHz to 1 RIU is expressed in units of GHz / RIU. Higher sensitivity indicates a more significant response of the structure to minute perturbations. In the field of metamaterials, the Q-value (Quality Factor) is an important parameter for measuring resonant performance, defined as the resonant frequency (Q / RIU). ) and resonance peak bandwidth ( The Q value reflects the selectivity of a structure to a specific frequency and its ability to store and dissipate energy within the system. A higher Q value indicates a sharper resonance peak, higher frequency resolution, and lower system energy loss. In terahertz metamaterial devices, a high Q value typically signifies a stronger local electric field enhancement effect and higher detection sensitivity, making it particularly suitable for high-precision sensing scenarios such as weak signal sensing and trace substance identification. Therefore, optimizing structural parameters to obtain a higher Q value is one of the key design goals for improving the performance of terahertz metamaterial sensors.

[0074] Among these, the optimal values ​​for Q and S of the metamaterial are achieved when: period P1: 58 μm; length of the rectangular edge of the structure L1: 32 μm; length of the long side of the L-shaped structure L2: 22 μm; length of the short side of the L-shaped structure L3: 17 μm; spacing between the long sides of different L-shaped structures L4: 14 μm; spacing between the short sides of different L-shaped structures L5: 4 μm; width of the rectangular edge of the structure W1: 2 μm; and width of the L-shaped structure W2: 4 μm. Figure 2 As shown, the FWHM of f1 and f2 are 132GHz and 636GHz respectively. According to the formula for calculating the Q value, the Q values ​​are 15.9 and 4.29, respectively. Figure 3 As shown, when the refractive index increases from 1.2 to 2.0, f1 and f2 shift by 26.7 GHz and 80 GHz, respectively. The calculated sensitivities are 33.4 GHz / RIU and 100 GHz / RIU, respectively.

[0075] S13. Select a single periodic structure as the modeling object, and perform modeling and simulation according to the target parameters to obtain the target terahertz material;

[0076] Specifically, based on the above analysis, modeling and simulation can be performed according to the target parameters to obtain the target terahertz material;

[0077] like Figure 4 , Figure 5 As shown, the target peak frequencies include 2.19 THz and 2.70 THz. The target terahertz material has an L-shaped composite bimodal structure. The target parameters include: period P1: 58 μm; length of the rectangular edge of the structure L1: 32 μm; length of the long side of the L-shaped structure L2: 22 μm; length of the short side of the L-shaped structure L3: 17 μm; spacing between the long sides of different L-shaped structures L4: 14 μm; spacing between the short sides of different L-shaped structures L5: 4 μm; width of the rectangular edge of the structure W1: 2 μm; width of the L-shaped structure W2: 4 μm. The substrate material of the target terahertz material is silicon with a refractive index of 3.335, and the surface metal layer is gold.

[0078] according to Figure 4 It can be seen that the metamaterial sensor has two characteristic peaks in the terahertz band, namely 2.19THz and 2.70THz.

[0079] The designed metamaterial sensor was structurally modeled using Lumerical FDTD Solutions software. Considering that metamaterials are typically composed of periodically repeating structural units, their overall electromagnetic response characteristics can be approximated by simulating the unit structures, thus significantly improving simulation efficiency and computational accuracy. Therefore, this embodiment selects a single periodic structure as the modeling object for simulation analysis. Silicon with a refractive index of 3.335 was chosen as the substrate material, and gold was used as the surface metal layer to enhance its resonant response characteristics in the terahertz band.

[0080] like Figure 6 As shown, Figure 6In the figure, 'a' and 'b' represent the spatial distribution of the electric field intensity of the target terahertz material at 2.19 THz and 2.70 THz, respectively, simulated using FDTD. The electric field distributions shown reveal the electromagnetic response characteristics of the designed terahertz metamaterial at different frequencies (2.19 THz and 2.70 THz) and its potential physical mechanisms. In the 2.19 THz electric field distribution, the electric field intensity exhibits a symmetrical oscillating mode with significant field enhancement at specific locations. However, in the 2.70 THz electric field distribution, the electric field mode changes, and the location and intensity of the field enhancement region differ from the 2.19 THz result. This difference may stem from the excitation of higher-order resonant modes of the metamaterial or the structure's dispersive response to higher frequencies. Therefore, the metamaterial exhibits significant frequency selectivity, verifying its tunability and application potential in the terahertz band.

[0081] S2. Prepare a naphthaleneacetic acid sample, and perform spectral acquisition on the naphthaleneacetic acid sample using the target terahertz material to obtain target spectral data;

[0082] Step S2 includes:

[0083] NAA standard solution was selected as the reference solution. Ultrapure deionized water was used to quantitatively dilute the reference solution using a high-precision pipette. Each sample was mixed at 3000 rpm for 3 minutes using a vortex mixer to ensure that the solute molecules were uniformly distributed in the solvent. Several sets of concentration gradient samples were transferred to brown volumetric flasks and sealed for storage. Dry air was continuously circulated into the optical cavity of the spectrometer using an air compressor beforehand. Temperature and humidity sensors were used to monitor and ensure that the relative humidity in the cavity was ≤10% and the temperature was stable within the range of 25±0.5℃. After the environment stabilized, 20 μl of different concentration gradient samples were quantitatively aspirated from low concentration to high concentration using a pipette and dropped onto the surface of the metamaterial sensor. A composite sampling mode of 5 points × 10 repeated measurements was used to obtain the target spectral data.

[0084] Specifically, this study used NAA standard solution provided by the Aladdin reagent platform as the reference material. To construct a trace detection system, 15 concentration gradient samples were prepared using a stepwise dilution method according to the requirements of the national standard GB 2763-2021. Ultrapure deionized water was used to quantitatively dilute the stock solution using a high-precision pipette. Each sample was vortexed at 3000 rpm for 3 minutes to ensure uniform distribution of solute molecules in the solvent. The final series of concentration solutions were transferred to brown volumetric flasks and sealed for storage. Specific concentration distribution parameters are detailed in Table 1.

[0085] Table 1. Concentration gradient of NAA solution

[0086]

[0087] To eliminate environmental interference, dry air was continuously supplied to the spectrometer's optical cavity using an air compressor before the formal experiment. Temperature and humidity sensors were used to monitor the process, ensuring the relative humidity within the cavity was ≤10% and the temperature remained stable within 25±0.5℃. Once the experimental environment was stable, NAA solutions of different concentration gradients were quantitatively added to the metamaterial sensor surface at 20 μl increments, from low to high concentration. Due to the strong absorption of THz waves by liquid water, the metamaterial was dried in a 50℃ drying oven for 30 minutes to ensure complete evaporation and prevent interference with the experimental results. Afterward, the dried metamaterial was placed into the instrument's optical cavity, allowing it to stand for 2 minutes after each sample addition to allow the system and cavity environment to stabilize. To improve data reliability, a composite sampling mode of 5 points × 10 repeated measurements was used. A total of 800 raw spectral data points, i.e., the target spectral data, were acquired from 15 concentration samples plus the bare metamaterial without added samples.

[0088] S3. Construct an initial model, train the initial model using a target feature selection method, and perform dimensionality reduction on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data.

[0089] Step S3 includes:

[0090] S31. Initialize the target spectral data into a feature set, which includes several feature subsets. Input the feature set into the initial model for prediction output to obtain predicted values. :

[0091] ;

[0092] In the formula, Let the feature weights be the feature weights of the j-th feature subset. For the i-th feature of the j-th feature subset, The intercept is... The number of features in the feature subset;

[0093] S32. Input the feature subsets into the initial model sequentially for training, and output the feature weight of each feature in each training round. Sort the features in the feature subset in descending order according to the absolute value of the feature weights, and remove the feature with the lowest absolute value of the feature weight to obtain a new feature subset.

[0094] S33. Re-input the new feature subset into the initial model and repeat the process of feature weight output and feature removal until the dimensionality reduction requirement is met, so as to obtain dimensionality-reduced spectral data;

[0095] Specifically, the dimensionality reduction requirement here is that the condition is met once the number of features is reduced to a certain level. The algorithm described above is a model-based feature selection method. Its core idea is to evaluate the importance of each feature by training the model, assessing the importance of each variable based on the model's regression coefficients or feature weights, gradually eliminating features that contribute the least to the model, and repeating the training and evaluation process until the number of features is reduced to a preset range or the performance is optimal. In each iteration, the root mean square error of the current feature subset is obtained through cross-validation to measure the subset's support for the prediction model. The goal is also to minimize the prediction error, thereby selecting the optimal feature subset.

[0096] In each iteration, the contribution of each feature is evaluated based on the absolute value of the feature weights, and the variable with the smallest weight is removed. This effectively retains key variables that have a significant impact on prediction performance and gradually removes redundant or noisy variables. It is suitable for modeling needs of high-dimensional and small sample data, and shows good feature compression capabilities, especially in spectral data analysis where the number of variables is much greater than the number of samples.

[0097] S4. The model parameters of the initial model are optimized using the target parameter optimization method to obtain optimized parameters. The initial model is then optimized based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model.

[0098] Step S4 includes:

[0099] S41. Initialize the model parameters of the initial model to the target population, and initialize each individual in the target population:

[0100] ;

[0101] In the formula, For the initialized individual, , These are the lower bound and upper bound of the variable, respectively. It is a uniform random vector;

[0102] Specifically, the initial model here is the SVR support vector regression model.

[0103] S42. Simulate the periodic shifting behavior of oat stems driven by wind by the initial individuals and guide each individual to swing towards the current optimal solution:

[0104] ;

[0105] In the formula, As the oscillation amplitude control factor, The oscillation frequency is... For phase perturbation, As the current globally optimal individual, , These are the individuals after the t-th and t+1-th iterations, respectively;

[0106] S43. Add perturbation updates based on the individuals after the (t+1)th iteration. To obtain updated individuals:

[0107] ;

[0108] In the formula, , These are two randomly selected individuals. For disturbance adjustment factor;

[0109] Specifically, the purpose of adding perturbation updates is to enhance local search capabilities.

[0110] S44. During the iteration process, perform boundary correction on each updated individual and compare the fitness of each updated individual before and after the iteration, retaining the individual with lower fitness.

[0111] Specifically, the boundary correction here means that if an individual exceeds the boundary during the iteration process, it is constrained to the boundary value to ensure that the update of each dimension does not go out of bounds.

[0112] S45. Repeat the iteration process until the iteration stopping condition is met, output the optimal individual to obtain the optimization parameters, and optimize the initial model based on the optimization parameters and the dimensionality reduction spectral data to obtain the target detection model.

[0113] S5. The target detection model is used to detect trace amounts of naphthaleneacetic acid residues.

[0114] Specifically, after obtaining the target detection model, the sample to be tested can be obtained. The target terahertz material obtained in the above steps is used to collect the spectrum of the sample to be tested. The obtained spectral data is input into the target detection model, and the detection result can be output.

[0115] It should be noted that, to verify the effectiveness of the target detection model (RFE-AOO-SVR) of this invention, a systematic comparative analysis was conducted with SVR models optimized using Ant Colony Optimization (ACO), Simulated Annealing (SA), and Genetic Algorithm (GA). All models were trained and predicted on the same training and test sets. Evaluation metrics such as correlation coefficient (R), root mean square error (RMSE), and mean absolute error (MAE) were used to comprehensively evaluate the models' performance in terms of fitting accuracy and generalization ability, ensuring the objectivity and fairness of the comparison results. C is the penalty coefficient, and γ is the kernel function parameter (gamma). The model results are shown in Table 2. Figure 6 As shown:

[0116] Table 2 Evaluation of the Prediction Performance of Each Model

[0117]

[0118] According to Table 2 above and Figure 7 The experimental results show that the AOO-SVR model exhibits the best performance in NAA terahertz spectral quantitative detection, with the highest correlation coefficient (R) and lowest error indices (RMSE and MAE) on the training and test sets, outperforming the SVR model optimized using ant colony optimization (ACO), simulated annealing (SA), and genetic algorithm (GA). This advantage is primarily due to the AOO algorithm's stronger global search and local mining capabilities, enabling it to effectively explore the parameter space over a larger range while avoiding getting trapped in local optima, thus obtaining a better combination of hyperparameters. Furthermore, AOO's dynamic oscillation mechanism and local perturbation strategy further enhance the accuracy and stability of the optimization, allowing the model to maintain high fitting accuracy while possessing better generalization ability. Therefore, AOO-SVR demonstrates stronger practicality and robustness in spectral data modeling tasks.

[0119] Therefore, the AOO-SVR model demonstrates superior performance compared to the other three models, fully validating the feasibility and advantages of applying the AOO algorithm to SVR modeling, and proving the potential and practical value of this method in quantitative analysis of terahertz spectroscopy. Its excellent modeling capabilities not only help improve the accuracy of pesticide residue detection but also provide new ideas for optimizing terahertz spectral data processing methods.

[0120] Meanwhile, to further evaluate the sensitivity and detection capability of the target detection model in practical applications, this paper calculates the limit of detection (LOD) of the model based on data extracted from three types of features. The LOD is an important indicator of the ability of a quantitative model to identify substances in a low concentration range. It is typically used to characterize the model's response to the lowest detectable concentration of the target substance. Calculating the LOD further verifies the effectiveness and reliability of the model in trace NAA detection tasks. In this study, the LOD is calculated using the 3-standard deviation method (3σ principle), which estimates the LOD by dividing three times the standard deviation (σ) of the predicted sample value by the slope (k) of the model, thus ensuring the statistical significance and engineering applicability of the results. The calculation formula is as follows:

[0121] ;

[0122] like Figure 8 As shown, Figure 8The fitted plots of the model predictions and actual values ​​of the target detection model are shown. The LOD of the three models, calculated using the above formula, is 0.036 μg / mL. The target detection model after feature extraction exhibits a good detection limit level, indicating that this method has a certain sensitivity in the detection of low concentrations of NAA residues. RFE-AOO-SVR shows greater advantages in suppressing redundant information and improving model response performance, demonstrating its practical application potential and applicability in processing high-dimensional nonlinear data in terahertz spectroscopy.

[0123] The trace naphthaleneacetic acid (NAA) detection method based on terahertz spectroscopy provided in Embodiment 1 of this invention integrates an L-shaped terahertz metamaterial sensor with the RFE-AOO-SVR intelligent modeling method. This not only achieves highly sensitive quantitative analysis of trace NAA residues but also verifies the feasibility and superiority of the AOO optimization algorithm in the field of spectral detection. It provides a practical and scalable research path for the rapid detection of pesticide residues using terahertz spectroscopy.

[0124] Example 2

[0125] like Figure 9 As shown, in Embodiment 2 of the present invention, a trace naphthaleneacetic acid detection system based on terahertz spectroscopy is provided, the system comprising:

[0126] Material module 1 is used to determine the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and to perform simulation based on the target parameters to obtain the target terahertz material;

[0127] Acquisition module 2 is used to prepare a naphthaleneacetic acid sample and to acquire the spectral data of the naphthaleneacetic acid sample through the target terahertz material.

[0128] Dimensionality reduction module 3 is used to construct an initial model, train the initial model using a target feature selection method, and perform dimensionality reduction processing on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data;

[0129] Optimization module 4 is used to optimize the model parameters of the initial model based on the target parameter optimization method to obtain optimized parameters, and optimize the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model;

[0130] Detection module 5 is used to detect trace amounts of naphthaleneacetic acid residues using the target detection model;

[0131] The material module 1 includes:

[0132] The acquisition submodule is used to acquire the characteristic peak frequency of naphthaleneacetic acid and determine the target peak frequency point of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid.

[0133] The scanning submodule is used to calculate the quality factor and sensitivity of the target terahertz material through simulated spectrum, and to perform parameter scanning on a preset database based on the target peak frequency point, the quality factor and the sensitivity to obtain the target parameters of the target metamaterial sensor.

[0134] The simulation submodule is used to select a single periodic structure as the modeling object and perform modeling and simulation based on the target parameters to obtain the target terahertz material.

[0135] The acquisition module 2 is specifically used for:

[0136] NAA standard solution was selected as the reference solution. Ultrapure deionized water was used to quantitatively dilute the reference solution using a high-precision pipette. Each sample was mixed at 3000 rpm for 3 minutes using a vortex mixer to ensure that the solute molecules were uniformly distributed in the solvent. Several sets of concentration gradient samples were transferred to brown volumetric flasks and sealed for storage. Dry air was continuously circulated into the optical cavity of the spectrometer using an air compressor beforehand. Temperature and humidity sensors were used to monitor and ensure that the relative humidity in the cavity was ≤10% and the temperature was stable within the range of 25±0.5℃. After the environment stabilized, 20 μl of different concentration gradient samples were quantitatively aspirated from low concentration to high concentration using a pipette and dropped onto the surface of the metamaterial sensor. A composite sampling mode of 5 points × 10 repeated measurements was used to obtain the target spectral data.

[0137] The dimensionality reduction module 3 includes:

[0138] The prediction submodule is used to initialize the target spectral data into a feature set, which includes several feature subsets. The feature set is then input into the initial model for prediction output to obtain predicted values. :

[0139] ;

[0140] In the formula, Let the feature weights be the feature weights of the j-th feature subset. For the i-th feature of the j-th feature subset, The intercept is... The number of features in the feature subset;

[0141] The training submodule is used to sequentially input the feature subsets into the initial model for training, and output the feature weight of each feature in each training round. The features in the feature subset are sorted in descending order according to the absolute value of the feature weights, and the feature with the lowest absolute value of the feature weight is removed to obtain a new feature subset.

[0142] The feature removal submodule is used to re-input the new feature subset into the initial model and repeatedly execute the feature weight output and feature removal process until the dimensionality reduction requirement is met, so as to obtain dimensionality-reduced spectral data.

[0143] The optimization module 4 includes:

[0144] The initialization submodule is used to initialize the model parameters of the initial model to the target population, and to initialize each individual in the target population:

[0145] ;

[0146] In the formula, For the initialized individual, , These are the lower bound and upper bound of the variable, respectively. It is a uniform random vector;

[0147] The oscillation submodule is used to simulate the periodic offset behavior of oat stalks driven by wind and guide each individual to oscillate towards the current optimal solution after initialization.

[0148] ;

[0149] In the formula, As the oscillation amplitude control factor, The oscillation frequency is... For phase perturbation, As the current globally optimal individual, , These are the individuals after the t-th and t+1-th iterations, respectively;

[0150] The update submodule is used to add perturbation updates based on the individuals after the (t+1)th iteration. To obtain updated individuals:

[0151] ;

[0152] In the formula, , These are two randomly selected individuals. For disturbance adjustment factor;

[0153] The retention submodule is used to perform boundary correction on each updated individual during the iteration process and compare the fitness of each updated individual before and after the iteration, retaining individuals with lower fitness.

[0154] The repeat submodule is used to repeatedly execute the iteration process until the iteration stopping condition is met, output the optimal individual to obtain the optimization parameters, and optimize the initial model based on the optimization parameters and the dimensionality reduction spectral data to obtain the target detection model.

[0155] In other embodiments of the present invention, the present invention provides the following technical solution: a computer, including a memory 102, a processor 101, and a computer program stored in the memory 102 and executable on the processor 101, wherein the processor 101 executes the computer program to implement the trace naphthaleneacetic acid detection method based on terahertz spectroscopy detection technology as described above.

[0156] Specifically, the processor 101 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0157] The memory 102 may include a large-capacity memory for data or instructions. For example, and not limitingly, the memory 102 may include a hard disk drive (HDD), a floppy disk drive, a solid-state drive (SSD), flash memory, an optical disk drive, a magneto-optical disk drive, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 102 may include removable or non-removable (or fixed) media. Where appropriate, the memory 102 may be internal or external to a data processing device. In a particular embodiment, the memory 102 is non-volatile memory. In a particular embodiment, the memory 102 includes read-only memory (ROM) and random access memory (RAM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable read-only memory (PROM), an erasable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), an electrically alterable read-only memory (EAROM), or flash memory, or a combination of two or more of these. Where appropriate, the RAM can be Static Random-Access Memory (SRAM) or Dynamic Random-Access Memory (DRAM). DRAM can be Fast Page Mode Dynamic Random Access Memory (FPMDRAM), Extended Data Out Dynamic Random Access Memory (EDODRAM), Synchronous Dynamic Random-Access Memory (SDRAM), etc.

[0158] The memory 102 can be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 101.

[0159] The processor 101 reads and executes the computer program instructions stored in the memory 102 to implement the above-described method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy.

[0160] In some embodiments, the computer may further include a communication interface 103 and a bus 100. For example, Figure 10 As shown, the processor 101, memory 102, and communication interface 103 are connected through bus 100 and communicate with each other.

[0161] The communication interface 103 is used to enable communication between the various modules, devices, units, and / or equipment in the embodiments of the present invention. The communication interface 103 can also enable data communication with other components such as external devices, image / data acquisition devices, databases, external storage, and image / data processing workstations.

[0162] Bus 100 includes hardware, software, or both, that couples components of a computer device together. Bus 100 includes, but is not limited to, at least one of the following: data bus, address bus, control bus, expansion bus, and local bus. For example, and not as a limitation, bus 100 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a Micro Channel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 100 may include one or more buses. Although specific buses are described and illustrated in the embodiments of the present invention, the present invention is contemplated by any suitable bus or interconnect.

[0163] The computer can use the trace naphthaleneacetic acid detection system based on terahertz spectroscopy to execute the trace naphthaleneacetic acid detection method of the present invention, thereby realizing the detection of trace naphthaleneacetic acid based on terahertz spectroscopy.

[0164] In some further embodiments of the present invention, in conjunction with the above-described method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy, the present invention provides the following technical solution: a storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the above-described method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy.

[0165] Those skilled in the art will understand that the logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can mean any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0166] More specific examples of readable media (a non-exhaustive list) include: electrical connections (electronic devices) with one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0167] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0168] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0169] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy, characterized in that, include: The target parameters of the target metamaterial sensor are determined based on the characteristic peak frequency of naphthaleneacetic acid, and simulation is performed based on the target parameters to obtain the target terahertz material. A naphthaleneacetic acid (NAA) sample was prepared, and the NAA sample was subjected to spectral acquisition using the target terahertz material to obtain target spectral data. An initial model is constructed, and the initial model is trained using a target feature selection method. The target spectral data is then dimensionality-reduced based on the trained initial model to obtain dimensionality-reduced spectral data. The model parameters of the initial model are optimized using a target parameter optimization method to obtain optimized parameters. The initial model is then optimized based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model. The target detection model enables the detection of trace amounts of naphthaleneacetic acid residues. The target terahertz material has an L-shaped composite bimodal structure. The target parameters include: period P1: 58 μm; length of the rectangular edge of the structure L1: 32 μm; length of the long side of the L-shaped structure L2: 22 μm; length of the short side of the L-shaped structure L3: 17 μm; spacing between the long sides of different L-shaped structures L4: 14 μm; spacing between the short sides of different L-shaped structures L5: 4 μm; width of the rectangular edge of the structure W1: 2 μm; width of the L-shaped structure W2: 4 μm. The substrate material of the target terahertz material is silicon with a refractive index of 3.335, and the surface metal layer is gold. The steps of training the initial model using a target feature selection method and then performing dimensionality reduction on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data include: The target spectral data is initialized into a feature set, which includes several feature subsets. This feature set is then input into the initial model for prediction output to obtain predicted values. : ； In the formula, Let the feature weights be the feature weights of the j-th feature subset. For the i-th feature of the j-th feature subset, The intercept is... The number of features in the feature subset; The feature subsets are sequentially input into the initial model for training, and the feature weights of each feature are output in each training round. The features in the feature subset are sorted in descending order according to the absolute value of the feature weights, and the features with the lowest absolute value of the feature weights are removed to obtain a new feature subset. The new feature subset is re-input into the initial model and the process of feature weight output and feature removal is repeated until the dimensionality reduction requirement is met, so as to obtain dimensionality-reduced spectral data. The steps of optimizing the model parameters of the initial model using the target parameter optimization method to obtain optimized parameters, and optimizing the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain the target detection model include: The model parameters of the initial model are initialized to the target population, and each individual in the target population is initialized: ； In the formula, For the initialized individual, , These are the lower bound and upper bound of the variable, respectively. It is a uniform random vector; The initial individual samples simulate the periodic shifting behavior of oat stems driven by wind, guiding each sample to swing towards the current optimal solution: ； In the formula, As the oscillation amplitude control factor, The oscillation frequency is... For phase perturbation, As the current globally optimal individual, , These are the individuals after the t-th and t+1-th iterations, respectively; Add perturbation update based on the individual after the (t+1)th iteration To obtain updated individuals: ； In the formula, , These are two randomly selected individuals. For disturbance adjustment factor; During the iteration process, boundary correction is performed on each updated individual and the fitness of each updated individual before and after the iteration is compared, and individuals with lower fitness are retained. The iterative process is repeated until the iteration stopping condition is met, and the optimal individual is output to obtain the optimized parameters. Based on the optimized parameters and the dimensionality-reduced spectral data, the initial model is optimized to obtain the target detection model.

2. The method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy as described in claim 1, characterized in that, The steps of determining the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and performing simulation based on the target parameters to obtain the target terahertz material include: Obtain the characteristic peak frequency of naphthaleneacetic acid, and determine the target peak frequency point of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid; The quality factor and sensitivity of the target terahertz material are calculated by simulating the spectrum, and the parameters of the preset database are scanned according to the target peak frequency point, the quality factor and the sensitivity to obtain the target parameters of the target metamaterial sensor. A single periodic structure is selected as the modeling object, and modeling and simulation are performed according to the target parameters to obtain the target terahertz material.

3. The method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy as described in claim 2, characterized in that, The target peak frequency points include 2.19 THz and 2.70 THz.

4. The method for detecting trace naphthaleneacetic acid based on terahertz spectroscopy as described in claim 1, characterized in that, The step of preparing a naphthaleneacetic acid sample and acquiring the spectral data of the naphthaleneacetic acid sample using the target terahertz material includes: NAA standard solution was selected as the reference solution. Ultrapure deionized water was used to quantitatively dilute the reference solution using a high-precision pipette. Each sample was mixed at 3000 rpm for 3 minutes using a vortex mixer to ensure that the solute molecules were uniformly distributed in the solvent. Several sets of concentration gradient samples were transferred to brown volumetric flasks and sealed for storage. Dry air was continuously circulated into the optical cavity of the spectrometer using an air compressor beforehand. Temperature and humidity sensors were used to monitor and ensure that the relative humidity in the cavity was ≤10% and the temperature was stable within the range of 25±0.5℃. After the environment stabilized, 20 μl of different concentration gradient samples were quantitatively aspirated from low concentration to high concentration using a pipette and dropped onto the surface of the metamaterial sensor. A composite sampling mode of 5 points × 10 repeated measurements was used to obtain the target spectral data.

5. A trace naphthaleneacetic acid (NAA) detection system based on terahertz spectroscopy, wherein the system employs the trace NAA detection method based on terahertz spectroscopy as described in claim 1, characterized in that... The system includes: The materials module is used to determine the target parameters of the target metamaterial sensor based on the characteristic peak frequency of naphthaleneacetic acid, and to perform simulation based on the target parameters to obtain the target terahertz material. The acquisition module is used to prepare a naphthaleneacetic acid sample and to acquire the spectral data of the naphthaleneacetic acid sample through the target terahertz material. The dimensionality reduction module is used to construct an initial model, train the initial model using a target feature selection method, and perform dimensionality reduction processing on the target spectral data based on the trained initial model to obtain dimensionality-reduced spectral data. An optimization module is used to optimize the model parameters of the initial model based on a target parameter optimization method to obtain optimized parameters, and to optimize the initial model based on the optimized parameters and the dimensionality-reduced spectral data to obtain a target detection model. The detection module is used to detect trace amounts of naphthaleneacetic acid residues using the target detection model.

6. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the trace naphthaleneacetic acid detection method based on terahertz spectroscopy as described in any one of claims 1 to 4.

7. A storage medium, characterized in that, The storage medium stores a computer program, which, when executed by a processor, implements the trace naphthaleneacetic acid detection method based on terahertz spectroscopy as described in any one of claims 1 to 4.

Images