A spectral detection device and method based on ultraviolet-fluorescence feature level fusion

Abstract

The application provides a spectrum detection device and method based on ultraviolet-fluorescence feature level fusion, and belongs to the field of spectrum detection and analysis. The device comprises an ultraviolet light source module, a sample detection module, a fluorescence collection module, a spectrum analysis module and a data processing module. The ultraviolet light source module is used for providing stable ultraviolet excitation light; the sample detection module is used for irradiating the ultraviolet light onto a sample to be detected and collecting the excited fluorescence signal; the fluorescence collection module captures the fluorescence signal through a high-sensitivity detector; the spectrum analysis module performs spectrum separation on the collected ultraviolet light and fluorescence signal; and the data processing module adopts a feature level fusion algorithm to perform fusion analysis on the features of the ultraviolet spectrum and the fluorescence spectrum, so as to improve the detection precision. The application can simultaneously utilize the advantages of the ultraviolet spectrum and the fluorescence spectrum, comprehensively detects the physical and chemical characteristics of a sample, and is widely applicable to the fields of food safety detection, biological medicine analysis, environmental monitoring and the like.

Description

Technical Field

[0001] This invention relates to the field of spectral detection and analysis technology, and more specifically, to a spectral detection device and method based on ultraviolet-fluorescence feature-level fusion. Background Technology

[0002] Spectroscopic detection technology, with its advantages of speed, non-destructive nature, and high specificity, has been widely used in fields such as food safety, biomedicine, and environmental monitoring. Traditional spectroscopic detection methods typically rely solely on ultraviolet absorption spectroscopy or fluorescence spectroscopy for analysis, which has technical limitations in balancing detection sensitivity and specificity.

[0003] Ultraviolet (UV) absorption spectroscopy can reflect the characteristic absorption information of chromophores in a sample, and it has a good identification effect on some substances with specific functional groups. However, it is easily affected by the background of the sample matrix and its identification ability for complex sample components is insufficient. Fluorescence spectroscopy has high sensitivity and can effectively detect trace substances. However, the fluorescence signal is easily affected by environmental fluctuations, and a single fluorescence feature cannot fully characterize the physicochemical properties of the sample. When dealing with multi-component mixed samples, feature overlap is prone to occur, leading to a decrease in the accuracy of the detection results.

[0004] Meanwhile, traditional ultraviolet light sources mostly use ultraviolet lamps, which have a wide output spectral width but poor monochromaticity of excitation light, further limiting the accuracy of fluorescence spectroscopy detection. In addition, existing spectral fusion technologies mostly remain at the data-level fusion level, simply splicing together the raw data of the two spectra, failing to fully explore the correlation characteristics between ultraviolet and fluorescence spectra. The fused data has high redundancy and low feature utilization, failing to fully leverage the synergistic advantages of the two spectral technologies.

[0005] In terms of optical path control, traditional spectral detection devices suffer from insufficient laser power adjustment precision, poor optical path stability, and susceptibility to stray light interference during spectral generation and separation, which adversely affects the reliability of the final detection results and makes them unsuitable for high-precision detection requirements in complex scenarios. Therefore, we propose an improvement by developing a spectral detection device and method based on ultraviolet-fluorescence feature-level fusion. Summary of the Invention

[0006] The purpose of this invention is to address the problems raised in the existing background technology. To achieve the above-mentioned objective, this invention provides the following technical solution: a spectral detection device based on ultraviolet-fluorescence feature-level fusion, comprising an ultraviolet light source module for providing stable ultraviolet laser light; The sample detection module is used to irradiate the sample to be tested with ultraviolet laser and induce fluorescence emission; The fluorescence acquisition module is used to capture the fluorescence signal of the sample to be tested; The spectral analysis module is used to separate and analyze ultraviolet spectral signals and fluorescence spectral signals; The data processing module is used to fuse the characteristics of ultraviolet and fluorescence spectra and output the detection results.

[0007] As a preferred technical solution of the present invention, the ultraviolet light source module controls the spectral range of the output light by adjusting the temperature of the nonlinear crystal. Because the single spectral width of the output light from the device is narrow and the total spectral range is wide, the measurement results of laser-induced fluorescence emission spectrum are more accurate than those of traditional ultraviolet lamps.

[0008] The sample detection module includes an optical fiber probe for introducing ultraviolet light and exporting fluorescence signals to the fluorescence acquisition module.

[0009] As a preferred embodiment of the present invention, the fluorescence acquisition module includes: High-sensitivity photodetectors are used to capture fluorescence signals; An optical filter bank is used to separate the target fluorescence signal from the background noise.

[0010] The spectral analysis module includes: A grating-based spectrophotometer system is used to separate ultraviolet and fluorescence spectra. The signal processing unit is used to extract spectral features.

[0011] As a preferred embodiment of the present invention, the data processing module employs a feature-level fusion algorithm, including: Feature extraction algorithms are used to extract feature parameters from ultraviolet and fluorescence spectra. Feature fusion algorithm is used to weightedly fuse the feature parameters of ultraviolet and fluorescence spectra.

[0012] As a preferred technical solution of the present invention, the feature fusion algorithm adopts a machine learning model, including but not limited to support vector machine (SVM), convolutional neural network (CNN) or random forest algorithm.

[0013] As a preferred embodiment of the present invention, the device further includes a user interaction module for displaying detection results and supporting parameter settings.

[0014] As a preferred technical solution of the present invention, the device is applicable to various sample types, including but not limited to liquid, solid and powder samples.

[0015] A spectral detection method based on ultraviolet-fluorescence characteristic-level fusion, comprising the following steps: collimating a nanosecond pulsed Gaussian laser through a focusing lens, then incident it onto a power control device consisting of a half-wave plate, a Faraday rotator, a light absorber, and a polarizing beam splitter to control the total power used in subsequent optical paths; after being focused by the focusing lens, the laser is incident onto a nonlinear crystal to generate frequency-doubled light; a dichroic mirror is used to separate the fundamental and frequency-doubled light; a light absorber is used to absorb the remaining fundamental light; simultaneously, a mirror is used to twist the frequency-doubled light into subsequent optical paths; a focusing mirror, a filter, and a half-wave plate are used to focus, further separate, and adjust the polarization direction of the frequency-doubled light before incident it into an optical parametric oscillator (OPO). The OPO consists of a simple linear cavity composed of an input mirror, a nonlinear crystal, and an output mirror; a dichroic mirror is used to reflect the visible light generated by the OPO. Near-infrared beams and frequency-doubled beams are split. A first and second total internal reflection mirror are used to orient and twist the visible-near-infrared beam into the subsequent optical path. A total internal reflection mirror is used to twist the frequency-doubled beam, and a beam splitter is used to split the beam. One frequency-doubled beam is shaped by a half-wave plate and then enters a nonlinear crystal for fourth-harmonic generation. A portion of this beam passes through a dichroic mirror and is combined with the visible-near-infrared beam, then enters the nonlinear crystal for sum-frequency processing to generate a near-ultraviolet-visible beam. This beam passes through a total internal reflection mirror and a focusing mirror and is incident into the test cell to measure the ultraviolet-visible absorption spectrum. The fourth-harmonic beam directly illuminates the test cell for ultraviolet excitation and emission. It is then focused into the propagation path by the focusing mirror, split using a grating, and then illuminates the spectral analysis device. The data collected by the device is processed by a computer for spectral data acquisition, calculation, and analysis, and the data is characterized and fused before the results are output.

[0016] As a preferred technical solution of the present invention, the steps include using a spectral analysis algorithm to perform noise reduction, baseline correction and feature selection on the spectral signal.

[0017] As a preferred technical solution of the present invention, the feature fusion processing in the steps includes feature weighting, feature normalization and classification modeling.

[0018] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention innovatively employs a feature-level fusion algorithm, first extracting highly recognizable feature parameters from ultraviolet and fluorescence spectra respectively, and then completing weighted fusion through a machine learning model, breaking through the limitations of traditional single-spectrum detection. It utilizes ultraviolet absorption spectroscopy to accurately reflect the functional group characteristics of the sample, and relies on fluorescence spectroscopy to achieve highly sensitive identification of trace substances, effectively reducing interference from the sample matrix background, significantly improving the specificity and accuracy of detection, and enabling clear feature differentiation even for complex multi-component samples.

[0019] The ultraviolet light source module of this invention achieves flexible control of the output spectral range by adjusting the temperature of the nonlinear crystal, resulting in an output spectrum with the dual advantages of narrow linewidth and wide coverage. Compared to traditional ultraviolet lamps, laser-induced fluorescence emission has higher spectral resolution, effectively avoiding feature overlap problems caused by broadband excitation, and further ensuring the accuracy of fluorescence spectral detection results.

[0020] This invention achieves precise control of laser power and efficient separation and directional transmission of beams from different wavelengths by integrating a power control device and a multi-stage beam separation and shaping component. It effectively filters out stray light and interference from residual fundamental frequency light, ensuring optical path stability, reducing signal noise levels, and providing high-quality beam input for subsequent spectral detection and analysis.

[0021] This invention's device supports the detection of various sample types, including liquids, solids, and powders. Combining the simultaneous detection capabilities of UV-Vis absorption spectroscopy and UV excitation-emission fluorescence spectroscopy, it can comprehensively acquire the physicochemical properties of samples. It is suitable for various application scenarios, such as screening for prohibited additives in food safety, analyzing active ingredients in biopharmaceuticals, and identifying pollutants in environmental monitoring, meeting complex detection needs.

[0022] The accompanying user interaction module of this invention enables visualization of detection results and flexible parameter settings. Combined with automated spectral data acquisition, noise reduction correction, and feature fusion processes, it greatly simplifies the operation process, reduces errors caused by manual intervention, and improves detection efficiency and operability. Attached Figure Description Figure 1 This is a schematic diagram of the structure provided by the present invention.

[0023] The image shows: 1. Nanosecond pulsed Gaussian laser; 2. Focusing lens; 3. Half-wave plate; 4. Faraday rotator; 5. Light absorber; 6. Polarizing beam splitter; 7. Focusing lens; 8. Nonlinear crystal; 9. Dichroic mirror; 10. Light absorber; 11. Mirror; 12. Focusing mirror; 13. Filter; 14. Half-wave plate; 15. Input mirror; 16. Nonlinear crystal; 17. Output mirror; 18. Dichroic mirror; 19. Total internal reflection mirror; 20. Total internal reflection mirror; 21. Total internal reflection mirror; 22. Beam splitter; 23. Half-wave plate; 24. Nonlinear crystal; 25. Dichroic mirror; 26. Nonlinear crystal; 27. Total internal reflection mirror; 28. Focusing mirror; 29. ​​Focusing mirror; 30. Grating; 31. Spectral analysis device; 32. Computer. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are specific implementations of the present invention and are not limited to all embodiments.

[0025] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0026] It should be noted that, in the absence of conflict, the embodiments and features and technical solutions in the embodiments of the present invention can be combined with each other. It should be noted that similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0027] Example 1: A spectral detection device based on ultraviolet-fluorescence feature-level fusion, comprising: The ultraviolet light source module is used to provide stable ultraviolet laser light. The sample detection module is used to irradiate the sample to be tested with ultraviolet laser and induce fluorescence emission; The fluorescence acquisition module is used to capture the fluorescence signal of the sample to be tested; The spectral analysis module is used to separate and analyze ultraviolet spectral signals and fluorescence spectral signals; The data processing module is used to fuse the characteristics of ultraviolet and fluorescence spectra and output the detection results.

[0028] The spectral detection method of the device includes the following steps: a nanosecond pulsed Gaussian laser 1 is collimated by a focusing lens 2, and then incident on a power control device consisting of a half-wave plate 3, a Faraday rotator 4, a light absorber 5, and a polarizing beam splitter 6 to control the total power used in the subsequent optical path. After being focused by a focusing lens 8, the laser is incident on a nonlinear crystal 9 to generate frequency-doubled light. A dichroic mirror 10 is used to separate the fundamental frequency light and the frequency-doubled light. A light absorber 11 absorbs the remaining fundamental frequency light, and a reflecting mirror 12 twists the frequency-doubled light to the subsequent optical path. A focusing mirror 13, a filter 14, and a half-wave plate 15 are used to focus, further separate, and adjust the polarization direction of the frequency-doubled light before it is incident on an optical parametric oscillator. The optical parametric oscillator is a simple linear cavity consisting of an input mirror 16, a nonlinear crystal 17, and an output mirror 18. A dichroic mirror 19 is used to separate the visible-near-infrared beam and the frequency-doubled light generated by the optical parametric oscillator. The beam splitting process involves using a first total reflection mirror 20 and a second total reflection mirror 22 to orient and twist the visible-near-infrared beam to the subsequent optical path. A total reflection mirror 21 is used to twist the frequency-doubled beam, and a beam splitter 23 splits the frequency-doubled beam. One beam of frequency-doubled beam is shaped by a half-wave plate and then enters a nonlinear crystal 26 for fourth-harmonic generation. A portion of this beam passes through a dichroic mirror 24 and is combined with the visible-near-infrared beam, entering a nonlinear crystal 27 for sum-frequency processing to generate a near-ultraviolet-visible beam. This near-ultraviolet-visible beam passes through a total reflection mirror 28 and a focusing mirror 30 and is incident into the test cell to measure the ultraviolet-visible absorption spectrum. The fourth-harmonic beam directly illuminates the test cell for ultraviolet excitation and emission. It is then focused by a focusing mirror 31 onto the propagation optical path, and a grating 32 is used for beam splitting. The beam is then illuminated by a spectral analysis device 33. The data collected by the device is processed by a computer 34 for spectral data acquisition, calculation, and analysis. The computer performs feature and fusion processing on the data and outputs the results.

[0029] The ultraviolet light source module controls the spectral range of the output light by adjusting the temperature of the nonlinear crystal 17. Because the single spectral width of the device output is narrow and the total spectral range is wide, the measurement results of laser-induced fluorescence emission spectrum are more accurate than those of traditional ultraviolet lamps.

[0030] The sample detection module includes a fiber optic probe for introducing ultraviolet light and exporting fluorescence signals to the fluorescence acquisition module.

[0031] The fluorescence acquisition module includes a high-sensitivity photodetector for capturing fluorescence signals; An optical filter bank is used to separate the target fluorescence signal from the background noise.

[0032] The spectral analysis module includes: a grating spectrometer system for separating ultraviolet and fluorescence spectra; The signal processing unit is used to extract spectral features.

[0033] The data processing module employs a feature-level fusion algorithm, including a feature extraction algorithm for extracting feature parameters from ultraviolet and fluorescence spectra. Feature fusion algorithm is used to weightedly fuse the feature parameters of ultraviolet and fluorescence spectra.

[0034] Feature fusion algorithms employ machine learning models, including but not limited to support vector machines (SVM), convolutional neural networks (CNN), or random forest algorithms.

[0035] The device further includes a user interaction module for displaying detection results and supporting parameter settings.

[0036] The device is suitable for a variety of sample types, including but not limited to liquid, solid and powder samples.

[0037] A spectral detection device and method based on ultraviolet-fluorescence feature-level fusion includes steps such as noise reduction, baseline correction, and feature selection of spectral signals using spectral analysis algorithms.

[0038] The feature fusion process in the steps includes feature weighting, feature normalization, and classification modeling.

[0039] The spectral detection device and method based on ultraviolet-fluorescence feature-level fusion of the present invention can significantly improve the sensitivity, specificity and applicability of spectral detection, and meet the complex detection needs of multiple fields.

[0040] Example 2: This example addresses the need for rapid detection of trace amounts of diethyl phthalate (DEP, a typical endocrine disruptor) in drinking water. It utilizes a spectroscopic detection device based on UV-fluorescence characteristic-level fusion to achieve accurate quantitative analysis of DEP concentration. The sample was effluent from a municipal waterworks, with an actual DEP concentration range of 0.01–1.0 μg / L. The sample type was liquid, and it needed to meet the requirements of rapid on-site detection, single detection time ≤ 5 minutes, high sensitivity (detection limit ≤ 0.005 μg / L), high specificity, and resistance to interference from residual chlorine, humic acid, and other substances.

[0041] Device configuration and parameter settings, core module parameters: Ultraviolet light source module: It adopts a nanosecond pulsed Gaussian laser with a wavelength of 1064nm as the fundamental frequency light. By adjusting the temperature of the nonlinear crystal 17, it is controlled at 25±0.1℃. The output spectrum range is 200~400nm of near-ultraviolet-visible beam, the center wavelength of the fourth harmonic light is 266nm, the single spectral width is ≤0.5nm, and the total power is adjusted to 50mW by the power control device to ensure excitation efficiency while avoiding sample photolysis.

[0042] Sample detection module: A quartz fiber optic probe with a numerical aperture of 0.22 and a core diameter of 200μm is used to introduce ultraviolet light into the test cell, which has a volume of 5mL and is made of quartz. The probe also simultaneously exports the fluorescence signal generated by the sample and emits wavelengths of 400-600nm to the fluorescence acquisition module.

[0043] Fluorescence acquisition module: Equipped with a high-sensitivity photomultiplier tube (PMT) with a response time ≤1ns to capture fluorescence signals, and paired with an optical filter bank with a center wavelength of 450nm and a bandwidth of 10nm to separate target fluorescence from background noise, such as the non-specific fluorescence of humic acid in water.

[0044] Spectral analysis module: The grating spectrometer system, with a grating line density of 1200 lines / mm, separates the ultraviolet absorption spectrum (200–400 nm) from the fluorescence emission spectrum (400–600 nm). The signal processing unit sequentially performs spectral noise reduction, employing wavelet threshold noise reduction algorithm, baseline correction, polynomial baseline fitting, and feature selection, and filters characteristic peaks based on the mutual information method.

[0045] Data processing module: The feature extraction algorithm selects peak area, peak height, and half-maximum width as core feature parameters, namely the absorption peak area A1 at 266nm in the ultraviolet spectrum, the emission peak height H1 at 450nm in the fluorescence spectrum, and the half-maximum width W1. The feature fusion algorithm adopts a support vector machine (SVM) model. First, the feature parameters are normalized, with min-max normalized to the [0,1] interval. Then, the weighting coefficients are optimized through grid search, with A1 weighted at 0.4, H1 weighted at 0.5, and W1 weighted at 0.1, to construct a concentration prediction model.

[0046] User interaction module: The touch screen displays the detection process, spectral curve, feature fusion results and final DEP concentration value. It supports manual adjustment of parameters such as power and integration time, with the integration time set to 1 second.

[0047] Sample pretreatment: Take 10mL of drinking water sample, no complicated pretreatment is required, and directly inject it into the test cell to avoid introducing pollution or loss of target substances, which meets the needs of rapid on-site detection.

[0048] Implementation of testing procedures: Equipment startup and parameter calibration: Turn on the device, preheat for 10 minutes, and perform baseline calibration using standard blank samples and ultrapure water to eliminate the influence of optical path background noise.

[0049] Ultraviolet excitation and signal acquisition: The ultraviolet light source module outputs 266nm near-ultraviolet light, which is introduced into the test cell through the fiber optic probe to irradiate the sample. DEP molecules absorb ultraviolet light and are excited, emitting 450nm characteristic fluorescence. The fluorescence acquisition module's PMT captures the fluorescence signal, the optical filter bank filters out background noise, and the ultraviolet absorption spectrum signal is acquired simultaneously.

[0050] Spectral analysis and feature extraction: The grating spectrometer of the spectral analysis module separates two spectral signals. The signal processing unit removes random noise through wavelet thresholding and corrects baseline drift through polynomial fitting. Finally, it extracts three feature parameters: the area of ​​the ultraviolet absorption peak A1, the height of the fluorescence emission peak H1, and the full width at half maximum (W1).

[0051] Feature-level fusion and result output: The data processing module normalizes the three feature parameters, substitutes them into the pre-trained SVM fusion model, and calculates the predicted DEP concentration value through a weighted fusion algorithm. The user interaction module displays the spectral curve, feature parameters, and final detection results, accurate to 0.001 μg / L. The entire detection process takes 4.2 minutes.

[0052] Test results and performance verification: Accuracy verification: DEP standard samples with known concentrations of 0.05 μg / L, 0.5 μg / L, and 0.9 μg / L were tested, and the measured values ​​were 0.048 μg / L, 0.503 μg / L, and 0.897 μg / L, respectively, with a relative error of ≤0.6%, which is significantly better than traditional UV lamp detection, with a relative error of ≥3.2%.

[0053] Sensitivity verification: The detection limit of the device for DEP is 0.003 μg / L, which is lower than the preset requirement of ≤0.005 μg / L. This is attributed to the strong excitation efficiency of the narrow linewidth ultraviolet light source and the amplification effect of the feature-level fusion algorithm on weak signals.

[0054] Specificity verification: When humic acid (10 times concentration), 10 μg / L residual chlorine, and 0.5 mg / L interfering substance were added to the sample, the relative error of the detection results was 0.8%, indicating that the noise separation function of the optical filter group and the anti-interference ability of the SVM model effectively improved the detection specificity.

[0055] Repeatability verification: The same 0.3 μg / L standard sample was tested 10 times. The relative standard deviation (RSD) was 1.1%, which reflects the stability of the device.

[0056] This embodiment achieves rapid and accurate detection of trace DEP in drinking water through the coordinated operation of various modules of the device and the SVM feature-level fusion algorithm. Compared with traditional detection methods, it has three major advantages: First, it eliminates the need for complex sample pretreatment, improving detection efficiency by more than 5 times; second, the feature-level fusion algorithm integrates ultraviolet-fluorescence dual spectral information, effectively offsetting the influence of interfering substances and significantly improving specificity and accuracy; third, the device is compact in size, suitable for on-site detection scenarios, and meets the complex detection needs of environmental protection, water affairs and other fields.

[0057] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.

Claims

1. A spectroscopic detection device based on ultraviolet-fluorescence characteristic-level fusion, characterized in that, include: The ultraviolet light source module is used to provide stable ultraviolet laser light. The sample detection module is used to irradiate the sample to be tested with ultraviolet laser and induce fluorescence emission; The fluorescence acquisition module is used to capture the fluorescence signal of the sample to be tested; The spectral analysis module is used to separate and analyze ultraviolet spectral signals and fluorescence spectral signals; The data processing module is used to fuse the characteristics of ultraviolet and fluorescence spectra and output the detection results.

2. The spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The ultraviolet light source module controls the spectral range of the output light by adjusting the temperature of the nonlinear crystal (17). Because the single spectral width of the output light from the device is narrow and the total spectral range is wide, the measurement results of laser-induced fluorescence emission spectrum are more accurate than those of traditional ultraviolet lamps. The sample detection module includes an optical fiber probe for introducing ultraviolet light and exporting fluorescence signals to the fluorescence acquisition module.

3. A spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The fluorescence acquisition module includes: High-sensitivity photodetectors are used to capture fluorescence signals; An optical filter bank is used to separate the target fluorescence signal from the background noise. The spectral analysis module includes: A grating-based spectrophotometer system is used to separate ultraviolet and fluorescence spectra. The signal processing unit is used to extract spectral features.

4. A spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The data processing module employs a feature-level fusion algorithm, including: Feature extraction algorithms are used to extract feature parameters from ultraviolet and fluorescence spectra. Feature fusion algorithm is used to weightedly fuse the feature parameters of ultraviolet and fluorescence spectra.

5. A spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 4, characterized in that, The feature fusion algorithm employs machine learning models, including but not limited to support vector machines (SVM), convolutional neural networks (CNN), or random forest algorithms.

6. A spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 5, characterized in that, The device further includes a user interaction module for displaying detection results and supporting parameter settings.

7. A spectral detection device based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The device is suitable for various sample types, including but not limited to liquid, solid and powder samples.

8. A spectral detection method based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The spectral detection method of the device includes the following steps: a nanosecond pulsed Gaussian laser (1) is collimated by a focusing lens (2), and then incident on a power control device consisting of a half-wave plate (3), a Faraday rotator (4), a light absorber (5), and a polarizing beam splitter (6) to control the total power used in the subsequent optical path. After being focused by a focusing lens (8), the laser is incident on a nonlinear crystal (9) to generate frequency-doubled light. A dichroic mirror (10) is used to separate the fundamental light and the frequency-doubled light. The remaining fundamental frequency light is absorbed using an absorber (11), and the frequency-doubled light is twisted into the subsequent optical path using a mirror (12). The frequency-doubled light is focused, further separated, and its polarization direction is adjusted using a focusing mirror (13), a filter (14), and a half-wave plate (15), and then incident into an optical parametric oscillator. The optical parametric oscillator is a simple linear cavity composed of an input mirror (16), a nonlinear crystal (17), and an output mirror (18). The visible-near-infrared beam generated by the optical parametric oscillator is filtered by a dichroic mirror (19). The frequency-doubled light is split into beams. The visible-near-infrared beam is oriented and twisted to the subsequent optical path using a first total reflection mirror (20) and a second total reflection mirror (22). The frequency-doubled light is twisted using a total reflection mirror (21). The frequency-doubled light is split into beams using a beam splitter (23). One beam of frequency-doubled light is shaped by a half-wave plate and then enters a nonlinear crystal (26) for fourth-order frequency doubling. A portion of the beam is combined with the visible-near-infrared beam by a dichroic mirror (24) and then enters a nonlinear crystal (27) for frequency summation to generate near-ultraviolet-visible light. The near-ultraviolet-visible light beam is incident into the test cell through a total reflection mirror (28) and a focusing mirror (30) to measure the ultraviolet-visible absorption spectrum. The fourth harmonic light is directly irradiated into the test cell for ultraviolet excitation and emission. It is focused into the propagation optical path through the focusing mirror (31), and the beam is split using a grating (32). Then it is irradiated into the spectral analysis device (33). The data collected by the device is processed by a computer (34) for spectral data acquisition, calculation and analysis, and the data is characterized and fused and the results are output.

9. A spectroscopic detection method based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The steps include using spectral analysis algorithms to denoise the spectral signal, perform baseline correction, and select features.

10. A spectral detection method based on ultraviolet-fluorescence feature-level fusion according to claim 1, characterized in that, The feature fusion process in the steps includes feature weighting, feature normalization, and classification modeling.

Images