A high-precision universal method for quantitative analysis of multiple fluorescent substances

Abstract

The present invention provides a high-precision, universal method for quantitatively analyzing the composition of multiple fluorescent substances. The method comprises: preprocessing raw spectral data of the multiple fluorescent substances to obtain spectral datasets; segmenting the spectral datasets; establishing a quantitative analysis model for crosstalk fluorescence spectroscopy based on the segmented spectral datasets, and performing quantitative analysis of the fluorescent substance components based on the quantitative analysis model. Compared with existing three-dimensional fluorescence spectral correction methods, the crosstalk fluorescence spectroscopy analysis method of the present invention is a more complete analysis method with higher quantitative analysis accuracy. Furthermore, the method development process does not rely on three-dimensional spectra, making it universal.

Description

Technical Field

[0001] The present invention relates to the field of quantitative analysis technology, and in particular to a high-precision universal method for quantitatively analyzing components of multiple fluorescent substances. Background Art

[0002] Fluorescence quantitative analysis is a commonly used method for material analysis. Since some substances contain fluorescent chromophores, they will emit fluorescence when excited by light of a certain wavelength. The intensity of the emitted fluorescence is proportional to the intensity of the excitation light and the content of the substance (Eq.1):

[0003]

[0004] In Eq. 1, I f It refers to the fluorescence intensity emitted by the fluorescent component within the range that the photodetector can detect (called the central area). is the fluorescence quantum yield of the fluorescent substance; I0 is the excitation light intensity; ε is the molar extinction coefficient of the fluorescent substance; l is the effective optical path length, which refers to the excitation light path length within the detection range of the photodetector; and c is the concentration of the fluorescent substance. Within a certain concentration range, this theoretical analysis basis has been proven to effectively reflect the concentration of the analyte in many cases. It is widely used to analyze the content of fluorescent substances. When the fluorescent substance being analyzed is a single component, it has high sensitivity and analytical accuracy.

[0005] However, in many applications, such as biological fluids, food samples, water quality testing, and petrochemical products, the target analyte contains multiple fluorescent components. When performing fluorescence quantitative analysis on these components, the fluorescence spectrum of a particular component is inevitably interfering with other components. When excited by light of a specific wavelength, the resulting emission spectrum is actually the combined result of the interaction between light and multiple fluorescent components. This includes both aliasing of the fluorescence emissions from each component after being excited by that wavelength and absorption of fluorescence emitted by one component by another. In this case, applying Eq. 1 to quantify a particular component will result in significant deviations. In other words, assuming that the fluorescence intensity at the characteristic wavelength of the fluorescent component to be quantified is still used to quantify the content of that component, according to Eq. 1, the relationship is linear. However, the light intensity at wavelength λ may be aliased with fluorescence emitted by other fluorescent components, or some of the fluorescence at wavelength λ may be absorbed by other fluorescent or non-fluorescent components. Under these conditions, the simple linear relationship described in Eq. 1 at the characteristic fluorescence wavelength no longer holds.

[0006] We collectively refer to the aforementioned fluorescence aliasing between fluorescent components and the quenching of fluorescence due to absorption (referred to as fluorescence absorption quenching, which can include absorption of fluorescence by fluorescent or non-fluorescent components) as fluorescence spectral crosstalk. To analyze the mechanism of fluorescence spectral crosstalk, we assume that there is no significant spatial attenuation effect of the excitation light, that is, each fluorescent component satisfies Eq. 1 when analyzed individually. Therefore, fluorescence aliasing of multicomponent fluorescence spectra can be considered as a simple linear combination of the fluorescence spectra of each single fluorescent component. This can be solved by combining multiple linear combinations of Eq. 1 or using multivariate linear analysis, which is also a widely used method.

[0007] However, in previous research and practice, fluorescence spectral crosstalk caused by fluorescence absorption quenching has often been intentionally or unintentionally ignored. It has only been reported in a few studies on the correction of internal filtering in three-dimensional fluorescence spectra, and researchers generally use Eq. 2 to correct for the effects of fluorescence absorption quenching.

[0008]

[0009] Wherein, is the observed fluorescence intensity value detected by the photodetector, is the ideal fluorescence intensity value in the central area, refers to the absorbance of the excitation light passing through the sample cell, and is the absorbance of the emitted fluorescence passing through the sample cell. Applying Eq.2 to correct the inner filtering effect of the three-dimensional fluorescence spectrum has a good effect, but it is necessary to meet the condition that the central area is located at the center of the cross-section of the sample cell, and it is necessary to use a spectrophotometer to measure the absorption spectrum of the sample separately. The controlled dilution approach (CDA) is another means of correcting the influence of fluorescence absorption quenching. CDA is suitable for three-dimensional fluorescence spectroscopy, but it requires additional experimental operations to increase the number of sample measurements. Another disadvantage of CDA is that sample dilution will reduce the signal-to-noise ratio of the fluorescence signal and sometimes change the properties of the molecules in the solution, resulting in huge data errors. Summary of the Invention

[0010] In view of the defects in the prior art, the present invention aims to provide a high-precision universal method for quantitatively analyzing the components of multiple fluorescent substances.

[0011] According to one aspect of the present invention, a high-precision universal method for quantitatively analyzing the components of multiple fluorescent substances is provided, comprising:

[0012] Preprocessing the original spectral data of various fluorescent substance components to obtain spectral data sets;

[0013] segmenting the spectral dataset;

[0014] A quantitative analysis model for crosstalk fluorescence spectrum analysis is established according to the segmented spectral data set, and quantitative analysis of fluorescent substance components is performed based on the quantitative analysis model.

[0015] Furthermore, the pre-processing of the original spectral data of the multiple fluorescent substance components includes:

[0016] Obtain raw fluorescence spectrum data of various fluorescent substances;

[0017] The function value of the β function is calculated according to the original fluorescence spectrum data, and n characteristic wavelengths are selected.

[0018] Furthermore, the expression of the β function is:

[0019]

[0020] in, represents the concentration vector of the components in the sample, ε(λ ab ) represents the molar extinction coefficient of each component, represents the fluorescence emission quantum yield, λ ab is the absorption wavelength, λ em is the fluorescence emission wavelength; the absorption conversion coefficient γ(λ ab )=μ·ψ(λ ab ), ψ(λ ab ) is the fluorescent substance for wavelength λ ab The absorption conversion rate of the excitation light, μ, indicates how much of the absorbed light is used to excite fluorescence, and μ is the absorption of the excitation light I within the linear concentration range of the Lambert-Beer law. ab The correction coefficient of the linear term in the Taylor expansion of ab =I0(1-e -2.303εbc )=2.303I0εbμc has the smallest error.

[0021] Furthermore, the fluorescence spectrum data of each group of samples of the raw fluorescence spectrum data are the average value after five measurements, and the preprocessing further includes dark current deduction and SG smoothing processing.

[0022] Furthermore, the segmenting of the spectral dataset includes segmenting the spectral dataset into a training set, a validation set, and a test set.

[0023] Furthermore, 30% of the spectral data in the spectral dataset are randomly selected to form the training set, 20% of the spectral data in the spectral dataset are randomly selected to form the validation set, and 50% of the spectral data in the spectral dataset are randomly selected to form the test set.

[0024] Furthermore, the establishment of a quantitative analysis model for crosstalk fluorescence spectroscopy analysis includes:

[0025] Under the excitation light condition of the excitation light wavelength, the expression of the relationship between the excitation light wavelength and the light intensity received at the fluorescence characteristic wavelength of the fluorescent component to be measured in the water sample and the concentration of the fluorescent component to be measured is established:

[0026]

[0027] Based on the above expression, assuming that the sample contains n components A, B, C..., the spectral intensity of n wavelength positions can be selected from the fluorescence spectrum, and a quantitative model can be established based on the β function to calculate c i , (i=A,B,C…) write as c i , (i=1,2,3…), then at the rth wavelength λ r Department:

[0028]

[0029] Fitting β by gradient descent method under the minimum mean square error criterion r Function to obtain the instrument constant p ri and q ri The quantitative model is determined by the value of .

[0030] Furthermore, the training set includes at least 2n samples, and the samples represent the spectral data in the training set.

[0031] Furthermore, the quantitative analysis of the fluorescent substance components based on the quantitative analysis model includes:

[0032] When the model reaches the optimal value, the fluorescence spectrum I f (λ) Calculate the function value of the β function Then, the β values ​​of the selected n wavelength positions are r Write down the system of equations, which is:

[0033]

[0034] Solve the equations to calculate c i , according to c i Calculate the average

[0035] Furthermore, the equations are solved to calculate c i , according to c i Calculate the average include:

[0036] Write down the Jacobian matrix of the equation system:

[0037]

[0038] Randomly select an initial value of the concentration vector Iterative calculation by computer And the iterative results will converge to the solution of the equations:

[0039]

[0040]

[0041] Compared with the prior art, the present invention has at least one of the following beneficial effects:

[0042] The present invention employs a crosstalk fluorescence spectroscopy method to quantitatively analyze multiple fluorescent components. By analyzing the formation and mutual conversion mechanisms of fluorescence spectral crosstalk, an analytical model for fluorescence spectral crosstalk of multiple components under single-wavelength excitation light is established. This method, which serves as a basis for quantitative measurement, uniformly addresses the effects of both fluorescence absorption quenching and fluorescence aliasing. Compared to prior three-dimensional fluorescence spectroscopy correction methods, the present crosstalk fluorescence spectroscopy method is a more comprehensive analytical method with higher quantitative analysis accuracy. Furthermore, the method development process is independent of three-dimensional spectroscopy, making it universally applicable. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:

[0044] Figure 1 is a flow chart of a crosstalk fluorescence spectroscopy quantitative analysis method (CFSA) according to an embodiment of the present invention;

[0045] Figure 2 are the excitation spectrum and emission spectrum of the fluorescent dye used for computer simulation in the embodiments of the present invention;

[0046] Figure 3 This is a flow chart for simulating the generation of crosstalk fluorescence spectra of multiple fluorescent components using a computer simulation method in an embodiment of the present invention;

[0047] Figure 4 1 are the absorption and emission spectra of pure solutions of tryptophan, sodium fluorescein, and rhodamine B at different concentrations according to a preferred embodiment of the present invention;

[0048] Figure 5 : This is the fluorescence spectra of all samples measured in a preferred embodiment of the present invention. The fluorescence peaks with wavelengths from short to long represent tryptophan, sodium fluorescein, and rhodamine B, respectively.

[0049] Figure 6 These are the quantitative analysis results of fluorescein sodium, rhodamine B and tryptophan content using the multiple linear regression method, partial least squares method and CFSA analysis method according to a preferred embodiment of the present invention, respectively. DETAILED DESCRIPTION

[0050] The present invention will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several variations and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0051] The embodiment of the present invention provides a high-precision universal method for quantitatively analyzing the components of multiple fluorescent substances. Figure 1 , the method comprising:

[0052] S1. Preprocessing the original spectral data of multiple fluorescent substance components to obtain a spectral data set;

[0053] In the step of preprocessing the raw spectral data, the function value of the β function is calculated and the characteristic wavelength is selected according to the fluorescence distribution of the analyte, which specifically includes:

[0054] Obtain raw fluorescence spectrum data of various fluorescent substances;

[0055] The function value of the β function is calculated based on the original fluorescence spectrum data, and n characteristic wavelengths are selected.

[0056] The spectral data set contains all the measured fluorescence spectra. Based on each of these fluorescence spectra, a beta function can be calculated, and then the data used for modeling in the beta function can be selected based on the characteristic wavelength. Calculating the beta function can make the final quantitative formula concise and simplify the form. Selecting the characteristic wavelength can avoid solving a large number of coefficients when solving the model coefficients. Among them, the beta function is an expression for the relationship between the excitation light wavelength and the received light intensity at the fluorescence characteristic wavelength of the fluorescent component to be measured in the water sample and the concentration of the fluorescent component to be measured. The expression of the beta function is:

[0057]

[0058] In formula 1, represents the concentration vector of the components in the sample, ε i (λ) represents the molar extinction coefficient of each component, represents the fluorescence emission quantum yield, λ is the wavelength; the absorption conversion coefficient γ i (λ)=μ·ψ(λ ab ), ψ(λ ab ) is the fluorescent substance for wavelength λ abThe absorption conversion rate of the excitation light, μ, indicates how much of the absorbed light is used to excite fluorescence, and μ is the absorption of the excitation light I within the linear concentration range of the Lambert-Beer law. ab The correction coefficient of the linear term in the Taylor expansion of ab =I0(1-e -2.303εbc )=2.303I0εbμc has the smallest error.

[0059] The method for determining the n characteristic wavelengths is: generally, the peak wavelength of each fluorescent component is selected, and n wavelengths are selected for n components.

[0060] In the step of obtaining the raw fluorescence spectral data of a plurality of fluorescent substances, the fluorescence spectral data of each group of samples of the raw fluorescence spectral data are the average values ​​after five measurements. Of course, those skilled in the art can also make any appropriate adjustments to this. In addition, the preprocessing of the raw spectral data also includes dark current subtraction and SG smoothing. The dark current subtraction can reduce the baseline error during data processing, and the S_G smoothing can reduce the noise during measurement.

[0061] S2, segmented spectral dataset;

[0062] To avoid irrational sampling distribution of training set samples, a validation set was included in the dataset segmentation to ensure that the quantitative model determined by the training set samples generalized well to the test set. During both computer simulations and actual experiments, the dataset was randomly segmented. Segmenting the spectral dataset specifically involves dividing it into a training set, a validation set, and a test set. Specifically, 30% of the spectral data from the dataset was randomly selected to form the training set, 20% of the spectral data from the dataset was randomly selected to form the validation set, and 50% of the spectral data from the dataset was randomly selected to form the test set.

[0063] S3. Establish a quantitative analysis model for crosstalk fluorescence spectrum analysis based on the segmented spectral data set, and perform quantitative analysis of fluorescent substance components based on the quantitative analysis model.

[0064] In some specific embodiments, establishing a quantitative analysis model for crosstalk fluorescence spectroscopy analysis includes:

[0065] Under the excitation light condition of the excitation light wavelength, an expression is established to express the relationship between the excitation light wavelength and the light intensity received at the fluorescence characteristic wavelength of the fluorescent component to be measured in the water sample and the concentration of the fluorescent component to be measured, which is the β function as shown in (Formula 1):

[0066]

[0067] Based on this expression, assuming that the sample contains n components A, B, C..., the spectral intensity of n wavelength positions can be selected from the fluorescence spectrum, and a quantitative model can be established based on the β function to calculate c i , (i=A,B,C…) write as c i , (i=1,2,3…), after the wavelength is determined, the coefficient in the first term is written as p, and the coefficient in the second term is written as q, then at the rth wavelength λ r Department:

[0068]

[0069] The beta function of a sample can be expressed as n expressions such as Equation 2. The instrument constant p in the quantitative model ri and q ri They can form an n×n matrix, and the constants that need to be determined are 2n 2 Therefore, the training set needs to contain at least 2n samples, which represent the spectral data in the training set. β can be fitted by gradient descent method under the minimum mean square error criterion. r Function to obtain the instrument constant p ri and q ri The quantitative model is determined by the value of .

[0070] According to the above expression, the instrument constant p in the quantitative model is determined ri and q ri After the validation set samples have been verified to have good generalization ability, the content of each fluorescent component can be calculated by iteratively solving the equations according to the following expression. Further, quantitative analysis of the fluorescent substance composition is carried out, including:

[0071] Determine the parameter p ri and q ri Finally, when the model reaches the optimal state, that is, as the iteration proceeds, the parameters p and q that need to be fitted no longer change, and the fluorescence spectrum I f (λ) Calculate the function value of the β function Then, the β values ​​of the selected n wavelength positions are r Write the system of equations, which is:

[0072]

[0073] Solve this system of equations to calculate c i , according to c i Calculate the average

[0074] To solve the concentration vector of n components in the unknown sample, Solve the system of equations to calculate c i , according to c i Calculate the average Includes: Writing the Jacobian matrix of the equation system:

[0075]

[0076] Randomly select an initial value of the concentration vector Iterative calculation by computer And the iterative results will converge to the solution of the equations:

[0077]

[0078]

[0079] In the embodiment of the present invention, a computer simulation method is used to simulate the crosstalk fluorescence spectrum of multiple fluorescent components to verify the theoretical analysis of the crosstalk of multiple components fluorescence spectrum under single wavelength excitation light. The following steps are performed:

[0080] Three fluorescent components are selected for simulation, which we call Component A, Component B, and Component C. The simulation data of their excitation and emission spectra per unit concentration can be derived from common fluorescent dyes, such as Figure 2 They are Alexa Fluor 350, Alexa Fluor 430, and Alexa Fluor 532. Their excitation and emission spectra are shown in the figure. As can be seen, 300nm excitation light can cause all three fluorescent components to emit fluorescence. The fluorescence emissions of these three components exhibit significant overlap, with some absorption and emission bands overlapping, leading to significant fluorescence absorption quenching. This property validates the effectiveness of the crosstalk fluorescence spectroscopy quantitative analysis (CFSA) method.

[0081] The simulation process is as follows Figure 3 When generating the simulated fluorescence spectrum, the concentrations of the three components in each sample are generated by a random number generator. Combining the absorption spectra of each component, a crosstalk fluorescence spectrum containing absorption quenching factors can be simulated. The generation formula is Equation (1).

[0082] The embodiment of the present invention theoretically analyzes the formation and mutual conversion mechanism of fluorescence spectral crosstalk, establishes an analytical model for fluorescence spectral crosstalk of multi-component fluorescent substances under single wavelength excitation light, and proposes a crosstalk fluorescence spectral quantitative analysis method (Coupling Fluorescence Spectrum Analysis, CFSA). Using this as a basis for quantitative measurement, it can uniformly solve the two problems of fluorescence absorption quenching and fluorescence aliasing. For the established analytical model and the proposed method, simulation calculations and experiments verify that fluorescence aliasing and fluorescence absorption quenching cause fluorescence spectral crosstalk, and verify that the quantitative analysis accuracy of multi-component fluorescent substances using CFSA is superior to that of conventional linear methods, such as multivariate linear regression (MLR) and partial least squares (PLS). The method in this embodiment has higher quantitative analysis accuracy.

[0083] The following uses three common fluorescent substances used in the experiment: tryptophan, sodium fluorescein and rhodamine B as examples to further explain in detail the high-precision universal method for quantitatively analyzing the components of multiple fluorescent substances in the embodiment of the present invention.

[0084] S1. Preprocessing the original spectral data of multiple fluorescent substance components to obtain a spectral data set;

[0085] Three common fluorescent substances were used in the experiment. Their fluorescence emission wavelengths are ranked from short to long: tryptophan, sodium fluorescein, and rhodamine B. The absorption and emission spectra of pure solutions at different concentrations are shown in Figure 2. Figure 4 shown.

[0086] Similarly, short-wavelength fluorescence emission is often covered by the absorption band of long-wavelength fluorescence emission components, resulting in absorption quenching. Although tryptophan does not have significant fluorescence aliasing with the other two fluorescent components, there is significant fluorescence aliasing between sodium fluorescein and rhodamine B.

[0087] An ultraviolet LED light source with a central wavelength of 275 nm was used as the excitation light source for the experiment, which can stimulate the fluorescence signals of the three types of fluorescent components.

[0088] It is important to note that when configuring the concentration, the sample should not have a significant spatial attenuation effect of the excitation light during the experiment.

[0089] The experimental setup can operate in quadrature reception mode. Fluorescence spectra were measured by a spectrometer. The fluorescence spectrum for each sample group was the average of five measurements. Spectral preprocessing also included dark current subtraction and SG smoothing.

[0090] S2, segmented spectral dataset;

[0091] When configuring the three-fluorescence component samples, the reference concentration of each component is also random, and a total of 30 samples are configured.

[0092] In the subsequent data analysis process, it will be randomly divided into three data sets.

[0093] Then, you can enter Figure 1 The quantitative analysis process is shown.

[0094] S3. Based on the segmented spectral data set, a quantitative analysis model for crosstalk fluorescence spectral analysis is established to perform quantitative analysis of fluorescent substance components.

[0095] S31, under the excitation light condition of the excitation light wavelength, establish an expression for the relationship between the excitation light wavelength and the received light intensity at the fluorescence characteristic wavelength of the fluorescent component to be measured in the water sample and the concentration of the fluorescent component to be measured.

[0096] S32, based on the expression of S1, assuming that the sample contains n components A, B, C..., the spectral intensity of n wavelength positions can be selected from the fluorescence spectrum, and a quantitative model can be established based on the β function. i , (i=A,B,C…) write as c i , (i=1,2,3…), then at the rth wavelength λ r The β function of a sample can be listed as n equations (2). The instrument constant p in the quantitative model ri and q ri They can form an n×n matrix, and the constants that need to be determined are 2n 2 Therefore, the training set needs to contain at least 2n samples. β can be fitted by gradient descent under the minimum mean square error criterion. r Function to obtain the instrument constant p ri and q ri The quantitative model is determined by the value of .

[0097] S33, instrument constant p to be determined ri and q ri Then, when the quantitative model is used to quantitatively analyze the unknown multi-fluorescence component samples in the test set, the measured fluorescence spectrum I f (λ) Calculate the function value of the β function Then, the β values ​​of the selected n wavelength positions are r Write the system of equations:

[0098]

[0099] To solve the concentration vector of n components in the unknown sample, We can first write the Jacobian matrix of the equation system:

[0100]

[0101] Randomly select an initial value of the concentration vector Iterative calculation by computer And the iterative results will converge to the solution of the equations:

[0102]

[0103] Figure 5 The fluorescence spectra of all samples were measured. The fluorescence peaks with the longest wavelengths represent tryptophan, fluorescein sodium, and rhodamine B, respectively. The peak wavelengths of the three fluorescent components in the spectrum vary from sample to sample, and the measured fluorescence spectra also exhibit a significant absorption quenching effect.

[0104] Figure 6 Figure 3 shows the quantitative analysis results of fluorescein sodium, rhodamine B, and tryptophan using multiple linear regression, partial least squares, and CFSA analysis, respectively. The results demonstrate that the CFSA analysis method has excellent quantitative analysis capabilities for actual crosstalk fluorescence spectra, and the analysis of fluorescence spectral crosstalk is consistent with objective experimental results.

[0105] from Figure 6 The quantitative analysis results of Table 1 and Table 2 can be used to obtain the determination coefficient R 2 The results of the mean square error show that for the three substances, fluorescein sodium, rhodamine B and tryptophan, the crosstalk fluorescence spectroscopy quantitative analysis method has a higher determination coefficient R than the multiple linear regression method and the partial least squares method. 2 , and a smaller mean square error, which proves that the crosstalk fluorescence spectrum quantitative analysis method in the embodiment of the present invention has higher quantitative analysis accuracy.

[0106] Table 1 Determination coefficient R 2

[0107]

[0108] Table 2 Mean square error

[0109]

[0110] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various modifications or variations within the scope of the claims without affecting the essence of the present invention. The above preferred features may be used in any combination as long as they do not conflict with each other.

Claims

1. A high-precision universal method for quantitatively analyzing the components of multiple fluorescent substances, characterized in that: include: Preprocessing the original spectral data of various fluorescent substance components to obtain spectral data sets; segmenting the spectral dataset; establishing a quantitative analysis model for crosstalk fluorescence spectrum analysis based on the segmented spectral data set, and performing quantitative analysis of fluorescent substance components based on the quantitative analysis model; The preprocessing of the original spectral data of the multiple fluorescent substance components includes: Obtain raw fluorescence spectrum data of various fluorescent substances; Calculate the function value of the β function according to the original fluorescence spectrum data, , ; and select n characteristic wavelengths; The method of establishing a quantitative analysis model for crosstalk fluorescence spectroscopy analysis includes: Under the excitation light condition of the excitation light wavelength, the expression of the relationship between the excitation light wavelength and the light intensity received at the fluorescence characteristic wavelength of the fluorescent component to be measured in the water sample and the concentration of the fluorescent component to be measured is established: ； in, represents the optical path, , represents the concentration vector of the components in the sample, represents the molar extinction coefficient of each component, represents the fluorescence emission quantum yield, is the absorption wavelength, is the fluorescence emission wavelength; absorption conversion coefficient , The fluorescent substance has a wavelength of The absorption conversion rate of the excitation light, Indicates how much of the absorbed light is used to excite fluorescence. Absorbs excitation light within the linear concentration range of the Lambert-Beer law The correction coefficient of the linear term in the Taylor expansion of Make The error is minimal; Based on the above expression, assuming that the sample contains Components , then we can select The spectral intensity at each wavelength position is calculated by building a quantitative model based on the β function. writing , then in wavelength Department: ； Fitting by gradient descent method under the minimum mean square error criterion Function to obtain instrument constants and The quantitative model is determined by the value of .

2. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 1, characterized in that: The fluorescence spectrum data of each group of samples of the raw fluorescence spectrum data are the average value after five measurements, and the pre-processing further includes dark current deduction and SG smoothing processing.

3. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 1, characterized in that: The segmenting of the spectral dataset includes segmenting the spectral dataset into a training set, a validation set, and a test set.

4. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 3, characterized in that: 30% of the spectral data in the spectral dataset are randomly selected to form the training set, 20% of the spectral data in the spectral dataset are randomly selected to form the validation set, and 50% of the spectral data in the spectral dataset are randomly selected to form the test set.

5. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 3, characterized in that: The training set contains at least samples, each of which represents spectral data in the training set.

6. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 1, characterized in that: The quantitative analysis of fluorescent substance components based on the quantitative analysis model includes: When the model reaches the optimal value, the fluorescence spectrum is measured Calculate the function value of the beta function , and then by the selected wavelength position Write down the system of equations, which is: ； Solve the equations to calculate ,according to Calculate the average .

7. The high-precision universal method for quantitatively analyzing multiple fluorescent substance components according to claim 6, characterized in that: Solve the equations to calculate ,according to Calculate the average ,include: Write down the Jacobian matrix of the equation system: ； Randomly select an initial value of the concentration vector , calculated iteratively by computer , and the iterative results will converge to the solution of the equations: ， 。

Images