Miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots

Abstract

The invention relates to the technical field of spectrometers, and discloses a miniaturized in-situ three-dimensional fluorescence spectrum detection device based on an LED array and quantum dots, and the device comprises a multi-wavelength LED light source module, a sample pool, a quantum dot spectrometer and a data processing control unit which are annularly arranged. During working, the control unit sequentially lightens LED excitation samples with different wavelengths, and generated fluorescence is coded into a specific light intensity pattern by the quantum dot optical filter array and is captured by the CMOS detector; and the data processing unit reconstructs a high-resolution emission spectrum through a deep learning model, and finally synthesizes a three-dimensional fluorescence spectrum. The device adopts a fully-sealed IP68 protective shell, has the advantages of small size, low power consumption, strong interference resistance, good environmental adaptability and the like, can realize in-situ, rapid and high-precision fluorescence detection of a field water body, and is suitable for the fields of environmental monitoring, biomedicine and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of spectrometer technology, in particular to a miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots. BACKGROUND

[0002] Three-dimensional fluorescence spectrum technology has important application value in the fields of environmental monitoring, biomedicine, food safety, etc. due to its high specificity, non-destructive detection, and no need for chemical reagents. In environmental detection, this technology can be used to analyze water humic acid, petroleum pollutants, and algal metabolites; in the field of biomedicine, it can quickly identify protein conformation changes, drug molecule interactions, etc. Compared with traditional chromatography or mass spectrometry methods, three-dimensional fluorescence spectrum can provide more abundant molecular fingerprint information and has the potential for real-time monitoring. However, the size, power consumption, and environmental adaptability of existing equipment limit its application in field detection and long-term in-situ monitoring.

[0003] Currently, mainstream three-dimensional fluorescence spectrometers usually use a xenon lamp combined with a double monochromator (excitation / emission) and a photomultiplier tube architecture, which has the problems of large size, high power consumption, and mechanical gratings susceptible to vibration interference. To meet the needs of environmental monitoring, some simplified devices use LED excitation and optical filter spectrometer solutions, but the low spectral resolution (>10 nm) and severe excitation light scattering interference due to the limited bandwidth of the optical filter. In addition, fixed LED arrays are prone to introduce uneven excitation errors due to spatial arrangement differences, while mechanical rotary LED switching schemes rely on motor drives, increasing power consumption and failure risk. These factors hinder the miniaturization, low power consumption, and field deployment capabilities of the equipment, and a new optical architecture is needed to balance performance and portability. SUMMARY

[0004] The purpose of the present application is to provide a miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots, which solves the above technical problems.

[0005] The miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots comprises an LED light source module, a sample cell, a focusing lens two, a quantum dot spectrometer, and a data processing and control unit. The LED light source module comprises 12 LEDs with different center wavelengths arranged in a ring, and each LED is equipped with a condenser lens one in front to make the excitation light converge and uniformly irradiate the water sample to be tested in the center of the sample cell. The quantum dot spectrometer is arranged in a direction perpendicular to the excitation light path of the LED light source module to receive the fluorescence signal generated after the sample is excited; the quantum dot spectrometer comprises a quantum dot filter array and a CMOS detector. The data processing and control unit is electrically connected with the LED light source module and the quantum dot spectrum detection module, is used for controlling the opening and closing of different wavelength LEDs in the LED array in turn, receiving the light intensity matrix data collected by the CMOS detector, restoring the emission spectrum through a spectrum reconstruction algorithm, and combining to generate a three-dimensional fluorescence spectrum.

[0006] As a further description of the technical scheme of the present application, the quantum dot filter array is composed of colloidal quantum dots of different sizes and materials, and the quantum dots have adjustable light absorption characteristics and can cover the ultraviolet-visible-near infrared band. The whole device is packaged in a fully sealed stainless steel shell, and the protection level reaches IP68.

[0007] As a further description of the technical scheme of the present application, the working process of the device includes: Step S1, device initialization and self-checking: power supply for the device, the data processing and control unit is started, and the communication and function self-checking of the key components such as the CMOS detector in the LED light source module and the quantum dot spectrometer are performed; Step S2, setting the first excitation wavelength: the data processing and control unit determines the first excitation wavelength to be used according to the preset or user input detection program ; The data processing and control unit sends instructions to the LED driving circuit to turn on the LED lamp, and turn off all other wavelength LEDs; Step S3, single wavelength excitation and fluorescence generation: The light emitted by the LED lamp is uniformly irradiated to the center area of the sample cell after being converged by the condenser lens in front of it, the excitation light is absorbed by the measured substance in the sample cell, the electrons in the substance jump to the excited state, and then emit lower energy and longer wavelength light, i.e. fluorescence, when returning to the ground state. The fluorescence contains all the emission characteristics of the substance under 625nm excitation; Step S4, capture and encoding of fluorescence signal; Step S5, processing and temporary storage of single excitation data; Step S6, cycle switching, complete full spectrum scanning: the data processing and control unit turns off the current LED, and turns on the LED of the next wavelength in the predetermined order; For each new excitation wavelength, steps S3-S5 are strictly repeated, and each repetition will generate an emission spectrum corresponding to the new excitation wavelength, until all the preset N excitation wavelengths are scanned, at this time, N emission spectrum data are sequentially stored in the storage unit; Step S7, three-dimensional spectrum synthesis and output.

[0008] As a further description of the technical scheme of the present application, before any excitation, the device first controls the CMOS detector to collect one or more frames of dark current data and environmental background scattered light data, which will be used as a reference for subsequent processing of all fluorescence signals to subtract the noise floor.

[0009] As a further description of the technical scheme of the present application, the specific working process of step S4 includes: the generated fluorescence scatters in all directions, and in the direction perpendicular to the excitation light path, the focusing lens collects this part of the fluorescence efficiently and converges it into a light beam; The converged fluorescence light beam is incident vertically into the entrance of the quantum dot spectrometer, first irradiating on the quantum dot filter array, and the composite fluorescence signal passes through all the units at the same time, each unit attenuates the light of different wavelengths to different degrees, thereby encoding the continuous fluorescence spectrum into a specific two-dimensional light intensity distribution pattern which is spatially discrete; The two-dimensional light intensity distribution pattern is captured by the CMOS detector behind it, and each pixel of the CMOS detector records the light intensity at the corresponding position, thereby converting the light pattern into a digital matrix, i.e. a light intensity matrix .

[0010] As a further description of the technical scheme of the present application, the working process of step S5 includes: The data processing and control unit reads the light intensity matrix from the CMOS detector through the data bus ; The data processing and control unit calls the built-in algorithm to pre-process the original matrix and subtract the dark current and background: ; wherein, is the original light intensity, is the dark current data, is the scattered light background, is the calibrated light intensity; The pre-processed light intensity matrix is input into the pre-trained deep learning spectral reconstruction model, which has learned the mapping relationship from the light intensity matrix of this specific encoding method to the real emission spectrum, and the spectral reconstruction model quickly operates to output a continuous, high-resolution emission spectrum , i.e. the curve of fluorescence intensity changing with emission wavelength when the excitation wavelength is ; This piece of emission spectrum data is stored in the storage unit together with its corresponding excitation wavelength label.

[0011] As a further description of the technical scheme of the application, the training spectrum reconstruction model process adopts a xenon lamp combined with a monochromator or an LED lamp as a monochromatic light source, fixes the equipment and the light source module, sequentially turns on monochromatic light sources of different wavelengths, the CMOS acquires a transmission light intensity matrix after the monochromatic light is filtered by the quantum dot filter matrix, and the transmission light intensity matrix is used as input layer data of the model after being deducted by dark current data and environmental scattering light background, and simultaneously, a spectrometer is used to collect monochromatic light spectrum data as output layer data of the model. The spectrum reconstruction model data set optimization collects real spectra of standard fluorescent samples or target detection substances, and the detection process is to prepare a standard fluorescent sample solution, add the solution into a cuvette, irradiate from the front by a monochromatic light source, and simultaneously collect a transmission light intensity matrix of generated fluorescence on the left side using the system, and collect spectrum data on the other side using a spectrometer. The spectrum reconstruction model data enhancement uses different concentrations or mixed solutions to increase the complexity and characteristics of the spectrum data and increase the size of the data set.

[0012] As a further description of the technical scheme of the application, the working process of the step S7 includes: The data processing and control unit combines all N pieces of emission spectrum data into an N*K data matrix in order of excitation wavelengths, wherein N is the number of excitation wavelength points, and K is the number of wavelength points of each emission spectrum, and the matrix is a discrete three-dimensional fluorescence spectrum EEM; The discrete three-dimensional fluorescence spectrum EEM is subjected to interpolation processing to generate a smooth and continuous three-dimensional fluorescence spectrum surface graph; The three-dimensional fluorescence spectrum surface graph is transmitted to an upper computer, a cloud platform or a mobile terminal through a wired or wireless interface of the device, and is used for further species identification, concentration quantification or pollution assessment.

[0013] The application has the following beneficial effects: 1. Low-power LED excitation light source array The patent adopts a programmable control LED ring array to replace a traditional xenon lamp and a mechanical monochromator, realizes multi-wavelength excitation, avoids stability problems caused by a mechanical rotating structure, and enables the system to still operate stably in vibration or a complex environment. The LED array is arranged in a ring shape, ensures that each region of a sample receives uniform excitation light, and eliminates measurement errors caused by position differences of light sources. Compared with a xenon lamp, the LED light source has the advantages of small size, low power consumption and long service life, and does not need additional heat dissipation design, is particularly suitable for long-term continuous monitoring applications, and greatly reduces maintenance costs.

[0014] 2. High-precision fluorescence spectrum detection unit The patent adopts a new detection architecture of colloid quantum dot filter array working with CMOS detector. Compared with the traditional scheme (such as CCD spectrometer) based on grating or narrow-band filter, the problem of mutual restriction of "light flux-resolution" exists. The present scheme is based on wavelength multiplexing principle, and through the wide spectrum absorption characteristics and accurate energy level regulation of quantum dot material, the synchronous optimization of high light flux and high resolution is realized. This design not only significantly improves the capture ability of weak fluorescent signal, but also effectively suppresses the influence of environmental scattered light due to its unique anti-interference characteristics. The quantum dot array adopts modular design, and the number of filter units can be flexibly configured. Through the adaptive spectral reconstruction algorithm, the system fault tolerance is maintained while the high spectral resolution is maintained. In addition, the architecture eliminates the need for precise gratings and mechanical control components, reducing the manufacturing cost to 30% of the traditional CCD spectrometer, and has economic advantages.

[0015] 3. Integrated solid structure design The detection device adopts a fully sealed stainless steel shell with IP68 level waterproof and dustproof performance, which can adapt to extreme environments, realize true in-situ detection, and avoid errors caused by sample transportation and pretreatment. The integrated design eliminates the complex optical path adjustment steps, significantly improves the efficiency of on-site work, and ensures the reliability and consistency of long-term monitoring data. BRIEF DESCRIPTION OF DRAWINGS

[0016] The present application will be further described below in conjunction with the drawings.

[0017] Figure 1 is a part of the flowchart of the miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots provided by the present application; Figure 2 is a part of the structure diagram of the miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots provided by the present application.

[0018] In the figure, 1 is an LED light source module; 2 is a condenser lens one; 3 is a sample cell; 4 is a condenser lens two; 5 is a quantum dot spectrometer; 6 is a data processing and control unit. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0020] Please refer to Figure 1As shown, the miniaturized in-situ three-dimensional fluorescence spectrum detection device based on LED array and quantum dots comprises an LED light source module 1, a sample cell 3, a focusing lens two 4, a quantum dot spectrometer 5 and a data processing and control unit 6. The LED light source module 1 comprises 12 LEDs with different central wavelengths arranged in a ring shape, and each LED is provided with a condenser lens one 2 in front to make the excitation light converge and uniformly irradiate on the water sample to be detected in the center of the sample cell 3. The quantum dot spectrometer 5 is arranged in a direction perpendicular to the excitation light path of the LED light source module 1, and is used to receive the fluorescence signal generated after the sample is excited; the quantum dot spectrometer 5 comprises a quantum dot filter array and a CMOS detector. The data processing and control unit 6 is electrically connected with the LED light source module and the quantum dot spectrum detection module, is used to control the sequential opening and closing of different wavelength LEDs in the LED array, receives the light intensity matrix data collected by the CMOS detector, restores the emission spectrum through a spectrum reconstruction algorithm, and generates a three-dimensional fluorescence spectrum.

[0021] As a further description of the technical scheme of the application, the quantum dot filter array is composed of colloidal quantum dots with different sizes and materials, and the quantum dots have adjustable light absorption characteristics and can cover the ultraviolet-visible-near infrared band. The whole device is packaged in a fully sealed stainless steel shell, and the protection level reaches IP68.

[0022] As a further description of the technical scheme of the application, the working process of the device comprises: Step S1, device initialization and self-checking: power supply such as turning on the battery or external power supply, The data processing and control unit 6 is started, and the communication and function self-checking of the key components such as the CMOS detector in the LED light source module 1 and the quantum dot spectrometer 5 are performed; Step S2, setting the first excitation wavelength: the data processing and control unit 6 determines the first excitation wavelength to be used according to the preset or user input detection program ; The data processing and control unit sends instructions to the LED driving circuit to turn on the LED lamp, and turn off all other wavelength LEDs; Step S3, single-wavelength excitation and fluorescence generation: The light emitted by the LED lamp is focused by the focusing lens 2 in front of it and uniformly illuminates the central area of ​​the sample cell 3. The excitation light is absorbed by the substance to be tested in the sample cell 3, and the electrons in the substance transition to the excited state. Then, when returning to the ground state, they emit light with lower energy and longer wavelength, i.e. fluorescence. The fluorescence contains all the emission characteristics of the substance under 625nm excitation. Step S4: Acquisition and encoding of fluorescence signals; Step S5: Processing and temporary storage of single excitation data; Step S6: Cyclic switching to complete full spectrum scan: Data processing and control unit 6 shuts down the current... LEDs, and turn on the next wavelength of LEDs in a predetermined sequence; For each new excitation wavelength, strictly repeat steps S3-S5. Each repetition will generate an emission spectrum corresponding to the new excitation wavelength until all N preset excitation wavelengths have been scanned. At this point, N emission spectrum data have been saved in sequence in the storage unit. Step S7: Three-dimensional spectral synthesis and output.

[0023] Through the above technical solution, the device of this invention mainly consists of an LED light source module 1, a sample cell 3, a focusing lens 4, a quantum dot spectrometer 5, and a data processing and control unit 6. During operation, the data processing and control unit first starts the system and performs a self-test. Then, according to a preset program, it controls 12 LEDs with different center wavelengths in the LED light source module to light up sequentially. Each LED is equipped with a focusing lens, which converges the excitation light and evenly illuminates the water sample to be tested at the center of the sample cell. When LED light of a certain wavelength illuminates the sample, the fluorescent substances in the sample are excited, generating a fluorescence signal. The fluorescence signal is collected and converged by the focusing lens in a direction perpendicular to the excitation light path, and then perpendicularly incident into the quantum dot spectrometer. The core of the quantum dot spectrometer includes a quantum dot filter array and a CMOS detector: the filter array is composed of colloidal quantum dots of various sizes and materials, which can selectively attenuate fluorescence in the ultraviolet-visible-near-infrared bands, encoding the continuous fluorescence spectrum into a specific two-dimensional light intensity distribution pattern; the CMOS detector captures this pattern and converts it into a digital light intensity matrix. After reading the matrix, the data processing and control unit first subtracts dark current and environmental background noise, then uses a pre-trained deep learning spectral reconstruction model (such as a convolutional neural network or Transformer) to restore the encoded light intensity matrix into a high-resolution emission spectrum curve, and stores this curve in association with the current excitation wavelength. By sequentially cycling through all preset excitation wavelengths and completing a full-spectrum scan, the system combines multiple emission spectra in order of excitation wavelength into a discrete three-dimensional fluorescence spectrum matrix. After interpolation, a continuous three-dimensional fluorescence spectrum surface is generated, and finally output to a host computer or cloud platform via a wired or wireless interface for substance identification, concentration quantification, or pollution assessment. The entire device uses a fully sealed stainless steel casing with an IP68 protection rating, making it suitable for in-situ field monitoring. It features low power consumption, high resolution, strong anti-interference capabilities, and a compact size.

[0024] As a further description of the technical solution of the present invention, before any excitation is performed, the device first controls the CMOS detector to acquire one or more frames of dark current data and ambient background scattered light data. These data will serve as the reference for all subsequent fluorescence signal processing and will be used to subtract background noise.

[0025] As a further description of the technical solution of the present invention, the specific working process of step S4 includes: the generated fluorescence is scattered in all directions, and in the direction perpendicular to the excitation light path, the focusing lens two (4) efficiently collects this part of the fluorescence and focuses it into a beam of light; The converged fluorescence beam is incident vertically into the entrance of the quantum dot spectrometer (5), and first illuminates the quantum dot filter array. The dot filter array consists of hundreds of filter units with different transmission characteristics. The composite fluorescence signal passes through all units at the same time. Each unit attenuates light of different wavelengths to different degrees, thereby encoding the continuous fluorescence spectrum into a spatially discrete, specific two-dimensional light intensity distribution pattern. The two-dimensional light intensity distribution pattern is captured by a CMOS detector placed directly behind it. Each pixel of the CMOS detector records the light intensity at the corresponding position, thereby converting the light pattern into a digital matrix, namely the light intensity matrix. This matrix is ​​a fingerprint-like representation of the fluorescence signal at a 625nm excitation wavelength.

[0026] Using the above technical solution, when the fluorescence generated by the sample is excited, it is collected and focused by a focusing lens, and the beam is then incident perpendicularly onto the quantum dot spectrometer. It first illuminates a quantum dot filter array, which consists of hundreds of filter units with varying transmission characteristics. The composite fluorescence signal passes through all units simultaneously. Each unit, based on the size and composition differences of the quantum dot material, produces different degrees of specific attenuation for different wavelengths of light, thus encoding the originally continuous fluorescence spectral information into a spatially discrete two-dimensional light intensity distribution pattern with specific characteristics. This pattern is then captured by a CMOS detector located immediately behind the array. Each pixel of the detector records the light intensity value at its corresponding position, ultimately converting this optical distribution into a digital light intensity matrix. This completes the encoding process of the fluorescence signal from the spectral domain to the spatial domain, providing crucial encoded data input for subsequent spectral reconstruction.

[0027] As a further description of the technical solution of the present invention, the working process of step S5 includes: The data processing and control unit (6) reads the light intensity matrix from the CMOS detector via the data bus. ; The data processing and control unit invokes a built-in algorithm to preprocess the original matrix, subtracting dark current and background: ;in, Original light intensity For dark current data, Background of scattered light, The calibrated light intensity; The preprocessed light intensity matrix is ​​input into a pre-trained deep learning spectral reconstruction model. This model has learned the mapping relationship from the light intensity matrix to the true emission spectrum using this specific encoding method. The spectral reconstruction model rapidly computes and outputs a continuous, high-resolution emission spectrum. That is, when the excitation wavelength is The curve showing the change in fluorescence intensity with emission wavelength; This emission spectrum data Its corresponding excitation wavelength The tags are stored together in the storage unit.

[0028] As a further description of the technical solution of this invention, the training process of the spectral reconstruction model uses a xenon lamp in conjunction with a monochromator or an LED lamp as a monochromatic light source. The device and the light source module are fixed, and monochromatic light sources of different wavelengths are turned on sequentially. The CMOS acquires the transmitted light intensity matrix after the monochromatic light passes through a quantum dot filter matrix. The transmitted light intensity matrix, after subtracting dark current data and ambient scattered light background, is used as the input layer data of the model. Simultaneously, a spectrometer is used to collect monochromatic light spectral data, which is used as the output layer data of the model. The deep learning model can be a convolutional neural network model or a Transformer model. The spectral reconstruction model dataset is optimized to collect the true spectra of standard fluorescent samples (such as Rhodamine B, humic acid, algal extracts, and fluorescent proteins) or target substances. The detection process involves preparing a standard fluorescent sample solution, adding the solution to a cuvette, irradiating the front with a monochromatic light source (xenon lamp with a monochromator or LED lamp), collecting the transmitted light intensity matrix of the generated fluorescence on the left side using the system, and simultaneously collecting spectral data on the other side using a spectrometer. The spectral reconstruction model data enhancement uses different concentrations or mixed solutions to increase the complexity and features of the spectral data, thereby increasing the size of the dataset.

[0029] Using the above technical solution, the data processing and control unit reads the encoded raw light intensity matrix from the CMOS detector; firstly, it calls the preprocessing algorithm, based on the formula... After subtracting the pre-collected dark current data and ambient scattered light background noise, a calibrated light intensity matrix is ​​obtained. This matrix is ​​then input into a pre-trained deep learning spectral reconstruction model (such as a convolutional neural network or Transformer). This model has learned the precise mapping relationship from the specific light intensity matrix encoded by the quantum dot filter array to the real, continuous emission spectrum through a large amount of training data. The model performs fast computation and inversion on the input matrix, outputting a high-resolution continuous emission spectrum curve, i.e., the relationship between fluorescence intensity and emission wavelength at the current excitation wavelength. Finally, the reconstructed emission spectrum data and its corresponding excitation wavelength label are stored together, completing the processing and temporary storage of single excitation data.

[0030] As a further description of the technical solution of the present invention, the working process of step S7 includes: The data processing and control unit (6) combines all N emission spectrum data into an N×K data matrix according to their excitation wavelength order, where N is the number of excitation wavelength points and K is the number of wavelength points of each emission spectrum. This matrix is ​​the discrete three-dimensional fluorescence spectrum EEM. The discrete three-dimensional fluorescence spectrum EEM is interpolated to generate a smooth, continuous three-dimensional fluorescence spectrum surface. The three-dimensional fluorescence spectral surface map is transmitted to a host computer, cloud platform or mobile terminal through the device's wired or wireless interface for further species identification, concentration quantification or pollution assessment.

[0031] It should be noted that the formulas in this application are all dimensionless and numerical calculations. The formulas are obtained by software simulation based on a large amount of data and are the closest to the real situation. The thresholds, assignments and coefficients involved in this application are all empirical values ​​and are selected by those skilled in the art according to the actual situation.

[0032] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED arrays and quantum dots, characterized in that, It includes an LED light source module (1), a sample cell (3), a second focusing lens (4), a quantum dot spectrometer (5), and a data processing and control unit (6). The LED light source module (1) includes 12 LEDs with different center wavelengths arranged in a ring. Each LED is equipped with a focusing lens (2) in front of it so that the excitation light is focused and evenly irradiated onto the water sample to be tested in the center of the sample cell (3). The quantum dot spectrometer (5) is positioned in a direction perpendicular to the excitation optical path of the LED light source module (1) and is used to receive the fluorescence signal generated after the sample is excited; the quantum dot spectrometer (5) includes a quantum dot filter array and a CMOS detector. The data processing and control unit (6) is electrically connected to the LED light source module and the quantum dot spectral detection module. It is used to control the sequential opening and closing of LEDs of different wavelengths in the LED array, receive the light intensity matrix data collected by the CMOS detector, restore the emission spectrum through the spectral reconstruction algorithm, and combine them to generate a three-dimensional fluorescence spectrum.

2. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 1, characterized in that, The quantum dot filter array is composed of colloidal quantum dots of various sizes and materials. The quantum dots have tunable light absorption characteristics and can cover the ultraviolet-visible-near-infrared bands. The device is entirely encapsulated in a fully sealed stainless steel housing, achieving an IP68 protection rating.

3. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 2, characterized in that, The operation process of the device includes: Step S1, Device Initialization and Self-Test: Power on the device. The data processing and control unit (6) is started to perform communication and functional self-test on key components such as the LED light source module (1) and the CMOS detector in the quantum dot spectrometer (5); Step S2, Setting the first excitation wavelength: The data processing and control unit (6) determines the first excitation wavelength to be used according to the preset or user-input detection program. ; The data processing and control unit sends a command to the LED driver circuit to turn it on. The LED lights are turned off, and all other wavelengths of LEDs are turned off; Step S3, Single-wavelength excitation and fluorescence generation: The light emitted by the LED lamp is focused by the focusing lens (2) in front of it and uniformly illuminates the central area of ​​the sample cell (3). The excitation light is absorbed by the substance to be tested in the sample cell (3), and the electrons in the substance jump to the excited state. Then, when returning to the ground state, they emit light with lower energy and longer wavelength, i.e. fluorescence. The fluorescence contains all the emission characteristics of the substance under 625nm excitation. Step S4: Acquisition and encoding of fluorescence signals; Step S5: Processing and temporary storage of single excitation data; Step S6, Cyclic switching to complete full spectrum scan: Data processing and control unit (6) closes the current LEDs, and turn on the next wavelength of LEDs in a predetermined sequence; For each new excitation wavelength, strictly repeat steps S3-S5. Each repetition will generate an emission spectrum corresponding to the new excitation wavelength until all N preset excitation wavelengths have been scanned. At this point, N emission spectrum data have been saved in sequence in the storage unit. Step S7: Three-dimensional spectral synthesis and output.

4. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 3, characterized in that, Before any excitation, the device first controls the CMOS detector to acquire one or more frames of dark current data and ambient background scattered light data. These data will serve as the reference for all subsequent fluorescence signal processing and will be used to subtract background noise.

5. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 3, characterized in that, The specific working process of step S4 includes: the generated fluorescence is scattered in all directions, and the focusing lens 2 (4) efficiently collects this part of the fluorescence in the direction perpendicular to the excitation light path and focuses it into a beam of light; The converged fluorescence beam is incident perpendicularly into the entrance of the quantum dot spectrometer (5), and first illuminates the quantum dot filter array. The composite fluorescence signal of the dot filter array passes through all units at the same time. Each unit attenuates light of different wavelengths to different degrees, thereby encoding the continuous fluorescence spectrum into a spatially discrete, specific two-dimensional light intensity distribution pattern. The two-dimensional light intensity distribution pattern is captured by a CMOS detector placed directly behind it. Each pixel of the CMOS detector records the light intensity at the corresponding position, thereby converting the light pattern into a digital matrix, namely the light intensity matrix. .

6. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 3, characterized in that, The working process of step S5 includes: The data processing and control unit (6) reads the light intensity matrix from the CMOS detector via the data bus. ; The data processing and control unit invokes a built-in algorithm to preprocess the original matrix, subtracting dark current and background: ;in, Original light intensity, For dark current data, Background of scattered light, The calibrated light intensity; The preprocessed light intensity matrix is ​​input into a pre-trained deep learning spectral reconstruction model. This model has learned the mapping relationship from the light intensity matrix to the true emission spectrum using this specific encoding method. The spectral reconstruction model rapidly computes and outputs a continuous, high-resolution emission spectrum. That is, when the excitation wavelength is The curve showing the change in fluorescence intensity with emission wavelength; This emission spectrum data Its corresponding excitation wavelength The tags are stored together in the storage unit.

7. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 6, characterized in that, The training process of the spectral reconstruction model uses a xenon lamp in conjunction with a monochromator or an LED lamp as the monochromatic light source. The device and light source module are fixed, and different wavelengths of monochromatic light sources are sequentially turned on. The CMOS sensor acquires the transmitted light intensity matrix after the monochromatic light passes through a quantum dot filter matrix. This transmitted light intensity matrix, after subtracting dark current data and ambient scattered light background, serves as the input layer data for the model. Simultaneously, a spectrometer collects monochromatic light spectral data, which serves as the output layer data for the model. The deep learning model can be a convolutional neural network model or a Transformer model. The spectral reconstruction model dataset is optimized to collect the true spectrum of standard fluorescent samples or target substances. The detection process involves preparing a standard fluorescent sample solution, adding the solution to a cuvette, illuminating the front with a monochromatic light source, using the system on the left side to collect the transmitted light intensity matrix that generates fluorescence, and simultaneously using a spectrometer on the other side to collect spectral data. The spectral reconstruction model data enhancement uses different concentrations or mixed solutions to increase the complexity and features of the spectral data, thereby increasing the size of the dataset.

8. The miniaturized in-situ three-dimensional fluorescence spectroscopy detection device based on LED array and quantum dots according to claim 3, characterized in that, The working process of step S7 includes: The data processing and control unit (6) combines all N emission spectrum data into an N×K data matrix according to their excitation wavelength order, where N is the number of excitation wavelength points and K is the number of wavelength points of each emission spectrum. This matrix is ​​the discrete three-dimensional fluorescence spectrum EEM. The discrete three-dimensional fluorescence spectrum EEM is interpolated to generate a smooth, continuous three-dimensional fluorescence spectrum surface. The three-dimensional fluorescence spectral surface map is transmitted to a host computer, cloud platform or mobile terminal through the device's wired or wireless interface for further species identification, concentration quantification or pollution assessment.

Images