Quantitative tracing method and system for internal and external pollution of chemical oxygen demand in lake water

Abstract

The present invention provides a method and system for quantitatively tracing the chemical oxygen demand (COD) of lake water bodies and internal and external pollution, relating to the technical field of quantitative pollution tracing. The method comprises: performing three-dimensional fluorescence spectral analysis on water samples at different points in the lake to obtain fluorescence excitation / emission matrix spectrum (EEMS) data; performing parallel factor analysis on the EEMS data to identify and classify the fluorescent components of dissolved organic matter (DOM) and determine the type characteristics of each component; using ridge regression analysis to establish a standardized multiple linear regression equation based on the maximum fluorescence intensity of the fluorescent components and the corresponding COD values, and calculating the percentage contribution of each fluorescent component to the COD of the lake water body; analyzing the impact path of potential variables on the migration and transformation of DOM, and quantifying the direct and indirect contributions of different pollution sources to each fluorescent component through path coefficients; and calculating the total contribution of rivers entering the lake, sediment release, and phytoplankton to the COD of the lake water body.

Description

Technical Field

[0001] The present invention relates to the technical field of quantitative pollution tracing, and in particular to a method and system for quantitative tracing the source of endogenous pollution of chemical oxygen demand (COD) in lake water. Background Art

[0002] Chemical oxygen demand (COD) is an important water quality indicator for surface water such as lakes and rivers. The higher the COD value, the more serious the organic pollution of the water body, which may lead to the depletion of dissolved oxygen and damage the aquatic ecosystem. As concentrations rise, macrophytes and phytoplankton in lakes fix more carbon from the atmosphere. , which is converted into dissolved organic matter in the lake. There is currently a lack of reliable technical means to identify the quantitative contribution of exogenous input and endogenous generation to the sources of COD pollution in the lake.

[0003] Patent application document CN114813585A discloses a remote sensing monitoring method and system for water quality parameters of brackish lakes, comprising: collecting water spectra and corresponding water samples from multiple regions of the lake to be monitored to obtain multiple water spectra and multiple water samples; analyzing each water sample to determine the water quality parameters of the corresponding region; performing correlation analysis between each water spectrum and the water quality parameters of each region, and performing a linear linear fit to determine multiple water quality parameter monitoring models; collecting hyperspectral images of the lake to be monitored and extracting spectral information of each water pixel; for any water pixel, determining multiple initial water quality parameters of the water pixel based on the spectral information of the water pixel and the water quality parameter monitoring models; and determining the water quality parameters of the water pixel based on the average value of the multiple initial water quality parameters of the water pixel, thereby enabling rapid detection of the water quality parameters of the lake. However, this patent cannot completely solve the existing technical problems and cannot meet the requirements of the present invention. Summary of the Invention

[0004] In view of the defects in the prior art, the purpose of the present invention is to provide a method and system for quantitatively tracing the source of internal and external pollution of the chemical oxygen demand of lake water.

[0005] The method for quantitatively tracing the source of internal and external pollution of chemical oxygen demand in lake water provided by the present invention comprises:

[0006] Step S1: Acquire water quality data, sediment data, and pollution load data of rivers entering the lake, wherein the water quality data includes chemical oxygen demand (COD), water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, soluble total nitrogen, orthophosphate, algal cell density, and chlorophyll a; the sediment data includes total nitrogen, total phosphorus, and organic matter content; and the river data includes total nitrogen, total phosphorus, and COD load entering the lake per unit time;

[0007] Step S2: Perform three-dimensional fluorescence spectral analysis on water samples at different points in the lake to obtain fluorescence excitation / emission matrix spectrum EEMS data;

[0008] Step S3: Performing Parallel Factor Analysis (PARAFAC) on the EEMS data to identify and classify the fluorescent components of dissolved organic matter (DOM). to The characteristics of each component type were determined by comparing with the Open Fluor organic fluorescence spectral library;

[0009] Step S4: Based on the fluorescent component to The maximum fluorescence intensity and the corresponding COD value were compared, and the standardized multiple linear regression equation was established using ridge regression analysis to calculate the contribution percentage of each fluorescent component to the COD of the lake water.

[0010] Step S5: Partial least squares path model (PLS-PM) is used to analyze the impact path of potential variables on DOM migration and transformation. The potential variables include phytoplankton, nutrient status, sediment, rivers entering the lake, and environmental variables. The direct and indirect contributions of different pollution sources to each fluorescent component are quantified by path coefficients.

[0011] Step S6: Calculate the total contribution of rivers entering the lake, sediment release, and phytoplankton to the COD of the lake water by combining the contribution percentage of each fluorescent component to the COD of the lake water and the path coefficients of different pollution sources.

[0012] Preferably, the parameters of the three-dimensional fluorescence spectrum analysis in step S2 include:

[0013] The excitation wavelength scanning range is 200nm~400nm, and the emission wavelength scanning range is 300nm~600nm;

[0014] The data intervals are 2nm for excitation and 1nm for emission, and the bandwidth of excitation and emission light is 5nm;

[0015] Ultrapure water was used for blank correction every 10 water samples measured, and the scanning speed was 6000 nm / min.

[0016] Preferably, the implementation of PARAFAC analysis in step S3 includes:

[0017] Data processing was performed using Matlab software and drEEM toolbox;

[0018] The preprocessing steps include subtracting Rayleigh scattering, Raman scattering, correcting for internal filter effects, normalizing Raman units, and removing noise areas;

[0019] The number of fluorescent components was determined by core consistency check and split verification, and 40 iterations were performed using non-negative constraints. The convergence criterion was set to .

[0020] Preferably, the specific method of the ridge regression analysis in step S4 is:

[0021] With COD as the dependent variable, fluorescent components to As the independent variable, the ridge parameter is introduced to solve the multicollinearity problem; firstly, the data is standardized to eliminate the dimension difference, and the formula is:

[0022]

[0023]

[0024] in, is the intensity of the jth fluorescence component of the i-th sample, 、 are the mean and standard deviation of the jth component respectively; is the COD value of the i-th sample, 、 is the mean and standard deviation of COD;

[0025] The standardized COD value As the dependent variable, the standardized fluorescence component As the independent variable, establish a ridge regression model:

[0026]

[0027] in, is the error term;

[0028] The regression coefficients are solved by minimizing the following objective function :

[0029]

[0030] in, is the ridge parameter, which is used to control the regularization strength; m is the number of samples, and n is the number of fluorescent components;

[0031] Use cross-validation to determine the optimal λ, divide the data set into a training set and a validation set, traverse the candidate values of λ, and calculate the mean square error of the validation set:

[0032]

[0033] in, Indicates traversing all samples in the validation set; Indicates the number of samples in the validation set;

[0034] Select the λ that minimizes the MSE as the optimal parameter;

[0035] The standardized regression coefficients Restore the coefficient to the original dimension , the formula is:

[0036]

[0037] The final standardized ridge regression equation is:

[0038]

[0039] Among them, the intercept term ;

[0040] Contribution percentage of each fluorescent component to COD Calculated by the following formula:

[0041]

[0042] in, is the regression coefficient of the kth fluorescent component, It represents the absolute value of the regression coefficient of the jth component, reflecting the relative weight of its impact on COD.

[0043] Preferably, the external model indicators of the PLS-PM model in step S5 include:

[0044] The reflectance index of DOM is the maximum fluorescence intensity of each fluorescent component;

[0045] The reflectance indicators of phytoplankton are algae density and chlorophyll a concentration;

[0046] Indicators reflecting nutritional status are total nitrogen, total soluble nitrogen, total phosphorus, and orthophosphate;

[0047] The reflective indicators of environmental variables are pH, water temperature, and dissolved oxygen;

[0048] The reflective indicators of sediment are organic matter, total nitrogen and total phosphorus content;

[0049] The reflective indicators of rivers entering the lake are total nitrogen, total phosphorus and COD loads.

[0050] Preferably, the significance of the path coefficient in step S5 is verified by 1000 bootstrap resamplings, and the model is determined to be valid when the goodness of fit GOF index is greater than 0.7.

[0051] Preferably, the total contribution of pollution sources in step S6 is calculated as follows:

[0052] For each fluorescent component , the pollution source contribution is the COD contribution value of the component multiplied by the path coefficient of the corresponding pollution source;

[0053] The total contribution is the sum of the product of the direct path coefficient and the indirect path coefficient.

[0054] The system for quantitatively tracing the source of internal and external pollution of chemical oxygen demand in lake water provided by the present invention comprises:

[0055] Data acquisition module, used to obtain monitoring data on lake water quality, sediment and rivers entering the lake;

[0056] Spectral analysis module, used to process three-dimensional fluorescence spectral data and perform PARAFAC analysis;

[0057] Regression modeling module, used to construct the ridge regression equation and calculate the COD contribution percentage of each fluorescent component;

[0058] Path analysis module, used to run the PLS-PM model and quantify the pollution source path coefficients;

[0059] The contribution calculation module is used to integrate the regression and path analysis results and output the total contribution and proportion of internal and external pollution to COD.

[0060] Preferably, the spectrum analysis module further comprises:

[0061] Scattering correction unit, used to subtract Rayleigh scattering and Raman scattering;

[0062] Noise filtering unit, used to remove abnormal spectral regions;

[0063] The component identification unit is used to match fluorescent component types using the Open Fluor spectral library.

[0064] Preferably, a visualization module is also included for displaying the contribution of pollution sources in a spatiotemporal dynamic chart and generating a governance strategy optimization report.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] (1) The present invention is based on conventional water quality monitoring data such as COD, nitrogen, phosphorus, phytoplankton, chlorophyll a, and nitrogen, phosphorus, and organic matter in sediments, as well as three-dimensional fluorescence spectrum data. The data is fast and easy to obtain, does not require expensive analytical techniques such as isotope methods and high-resolution mass spectrometry, does not increase the workload, and has excellent universality;

[0067] (2) The present invention establishes the path contribution coefficients of various organic matter approximate components of COD and different pollution sources, which can analyze the changes in pollution sources at monthly, annual and longer time and space scales, and can better guide local lake protection and management work. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] 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:

[0069] Figure 1 This is the pathway analysis of PLS-PM on dissolved organic matter in lakes;

[0070] Figure 2a to Figure 2d These are the four fluorescent components of organic matter in Erhai Lake, namely tryptophan-like component, high-molecular humus-like component, low-molecular humus-like component and humus-like component. DETAILED DESCRIPTION

[0071] The present invention will be described in detail below with reference to specific embodiments. The following examples 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 changes 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.

[0072] Example 1

[0073] The present invention provides a method for quantitatively tracing the source of COD pollution from internal and external sources in lake water, comprising the following steps:

[0074] Step 1: Obtain water quality and sediment data at different points in the lake. Lake water quality data include COD, water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, soluble total nitrogen, orthophosphate, algal cell density and chlorophyll a. Sediment data include total nitrogen, total phosphorus and organic matter. Data on rivers entering the lake are the load data of nitrogen, phosphorus and COD input into the lake per unit time.

[0075] Step 2: Perform three-dimensional fluorescence analysis on water samples at different locations to obtain fluorescence excitation-emission matrix spectroscopy (EEMS) data.

[0076] Specific process of three-dimensional fluorescence analysis:

[0077] Three-dimensional fluorescence spectra of water samples were measured using a fluorescence spectrophotometer and quartz fluorescence cuvettes. The excitation wavelength scan range was 200-400 nm with a 2 nm data interval, and the emission wavelength scan range was 300-600 nm with a 1 nm data interval. The excitation and emission bandwidths were set to 5 nm, and the spectral scan rate was 6000 nm / min. Ultrapure water was measured as a blank control for every 10 water samples.

[0078] Step 3: Perform parallel factor analysis (PARAFAC) on the spectral data to obtain 、 、…… The components were identified and compared with the Open Fluor organic fluorescence spectral library (http: / / www.openfluor.org) to clarify the component type characteristics.

[0079] The specific process of parallel factor analysis:

[0080] PARAFAC analysis was performed using Matlab software and the drEEM toolbox to identify the main components of organic matter. Before the PARAFAC analysis, the sample data set was preprocessed, namely, the data set was corrected using a pure water blank matrix, Rayleigh and Raman scattering were deducted and interpolated using the Delaunay triangulation method, the internal filtering effect was corrected using ultraviolet absorption spectrum data, and the Raman (RU) unit was normalized, and the spectral region where noise may exist was deducted. The least squares model of 2 to 7 components of the processed data set was tested for 40 iterations. All models were restricted by non-negative conditions, and the convergence criterion was set at Potential model outliers were evaluated and removed based on the residuals and leverage of each sample. The appropriate number of components was determined by observing the spectral shape of the organic fluorophores and the core consistency test. Finally, the PARAFAC results were validated using a split test and 10 iterations of randomly initialized models.

[0081] Step 4: It was previously generally believed that three-dimensional fluorescence spectroscopy could analyze dissolved organic matter (DOM) components, but it was difficult to quantify them. This protocol used ridge regression analysis to analyze the maximum fluorescence intensity of the lake's fluorescent components and their corresponding COD values to solve the multicollinearity problem of the correlation between the fluorescent components. The standardized ridge regression equation was then obtained, i.e., COD and 、 、…… The multiple linear regression equation of the components is calculated according to the equation. percentage.

[0082] Multiple linear regression and ridge regression analysis process:

[0083] With COD as the dependent variable, fluorescent components to As the independent variable, the ridge parameter is introduced to solve the multicollinearity problem; firstly, the data is standardized to eliminate the dimension difference, and the formula is:

[0084]

[0085]

[0086] in, is the intensity of the jth fluorescence component of the i-th sample, 、 are the mean and standard deviation of the jth component respectively; is the COD value of the i-th sample, 、 is the mean and standard deviation of COD;

[0087] The standardized COD value As the dependent variable, the standardized fluorescence component As the independent variable, establish a ridge regression model:

[0088]

[0089] in, is the error term;

[0090] The regression coefficients are solved by minimizing the following objective function :

[0091]

[0092] in, is the ridge parameter, which is used to control the regularization strength; m is the number of samples, and n is the number of fluorescent components;

[0093] Use cross-validation to determine the optimal λ, divide the data set into a training set and a validation set, traverse the candidate values of λ, and calculate the mean square error of the validation set:

[0094]

[0095] in, Indicates traversing all samples in the validation set; Indicates the number of samples in the validation set;

[0096] Select the λ that minimizes the MSE as the optimal parameter;

[0097] The standardized regression coefficients Restore the coefficient to the original dimension , the formula is:

[0098]

[0099] The final standardized ridge regression equation is:

[0100]

[0101] Among them, the intercept term ;

[0102] Contribution percentage of each fluorescent component to COD Calculated by the following formula:

[0103]

[0104] in, is the regression coefficient of the kth fluorescent component, It represents the absolute value of the regression coefficient of the jth component, reflecting the relative weight of its impact on COD.

[0105] Multicollinearity refers to the distortion or difficulty in accurately estimating the model due to precise or highly correlated relationships between explanatory variables in a linear regression model. The presence of multicollinearity can lead to a certain degree of error in the multiple linear regression model, making it difficult to interpret. There are various methods to address multicollinearity, including excluding the variables causing collinearity, differencing, and reducing the variance of parameter estimates.

[0106] In this solution, we use ridge regression, a method designed to reduce the variance of parameter estimates, to address the problem of multicollinearity. Ridge regression is a biased estimation regression method specifically designed for collinear data analysis. Essentially, it is a modified least squares estimation method. By abandoning the unbiased nature of least squares, it achieves a more realistic and reliable regression coefficient at the expense of some information loss and reduced precision. It also provides a better fit for pathological data than least squares.

[0107] Step 5: To explain the influencing mechanism, partial least squares structural equation modeling (PLS-PM) was used to estimate the complex relationships of latent variables. PLS-PM analysis was implemented using the "plspm" package in R 4.0.3. Phytoplankton, trophic status, sediments, inflowing rivers, and environmental variables have been shown to influence the migration and transformation of DOM. The path diagram of this model is shown in the following figure. Figure 1As shown in Figure 2 . The full-path model consists of two submodels: an internal model and an external model. The internal model shows the relationships between latent variables. The external model describes the relationships between latent variables and reflection indicators. The maximum fluorescence intensity (Fmax) of different components is used as the reflection indicator for DOM. Algal density and chlorophyll a concentration are selected as reflection indicators for phytoplankton. Total nitrogen, soluble total nitrogen, total phosphorus, and orthophosphate are used as reflection indicators for nitrogen and phosphorus nutrients. Reflection indicators for environmental variables include pH, water temperature, and dissolved oxygen. Organic matter, total nitrogen, and total phosphorus content are used as reflection indicators for sediment. Regarding the impact of river pollution inputs on freshwater systems, total nitrogen, total phosphorus, and COD loads are selected as reflection indicators for river inflows into lakes. DOM is directly or indirectly affected by all latent variables. Phytoplankton is primarily directly affected by trophic status and environmental variables, while the trophic status of lakes is also influenced by inflowing rivers, sediments, and environmental variables. The goodness of fit (GOF) index was used to assess the quality of the model, and a value greater than 0.7 was considered an appropriate reflection of the corresponding latent variable. The significance of the path coefficient was determined by 1000 bootstrap resamplings.

[0108] Use alone The path coefficients of 10 paths, Path 1-10, were obtained through model calculation ( - Path 1, Path 2, ..., Path 10). Similarly, use 、 、…… To obtain the path coefficient ( -path 1, path 2, ..., path 10).

[0109] Step 6: For river input sources, the contribution paths to DOM include direct path 1 and indirect path 2 (which are then converted into paths 8 and 9); for sediment sources, the contribution paths to DOM include direct path 3 and indirect path 4 (which are then converted into paths 8 and 9); for phytoplankton sources, the contribution path to DOM is only direct path 10. According to the PLS-PM path coefficient results, 、 、…… For the path coefficient results of river input, sediment and phytoplankton, the direct contribution is expressed as the direct path coefficient, and the indirect contribution is expressed as the product of the path coefficients. The total contribution is expressed as the direct effect path coefficient plus the indirect effect path coefficient.

[0110] Step 7: Use the multivariate linear regression equation calculated in step 4 to calculate the COD of lake water. Percentage, step 6 The total contribution path coefficients of different sources can be used to calculate the actual COD contribution of river input, sediment and phytoplankton.

[0111] Example 2: Quantitative traceability of COD in Erhai Lake.

[0112] According to the qualitative and quantitative analysis method of the three-dimensional fluorescence spectrum components of Erhai Lake in this study, the COD characteristics and sources of Erhai Lake at a specific time were analyzed, and the fluorescence components were analyzed. Tryptophan-like components, High molecular weight humic components, Low molecular weight humus components and Humus-like components, such as Figure 2a to Figure 2d .

[0113] The fluorescent components that constitute COD were calculated using the multiple linear regression and ridge regression quantitative analysis methods. 、 、 and The corresponding COD values are 11.66 mg / L, 0.75 mg / L, 1.60 mg / L and 0.43 mg / L respectively.

[0114] According to the partial least squares path model analysis, Table 1 shows the path coefficients of different sources for fluorescent components, and Table 2 shows the COD contributed by different sources. For example, in order to obtain the The COD contribution of the components needs to be known. The contribution ratio of the components is expressed as the path coefficient in the least squares path model. Component COD= Component contribution COD × rivers entering the lake The contribution of the component = 11.66 × 0.0175 = 0.20405 mg / L. The calculation of sediment and phytoplankton is the same as above, and the calculation is based on this. 、 and Partial content.

[0115] The specific process is:

[0116] Step 1: Data foundation;

[0117] Input: measured data from 30 sampling points in Erhai Lake;

[0118] Independent variables: maximum fluorescence intensity of the four fluorescent components (Fmax);

[0119] (Tryptophan-like): average intensity 1200;

[0120] (High molecular weight humus): average strength 85;

[0121] (low molecular humus): average strength 150;

[0122] (humus-like): average strength 50;

[0123] Dependent variable: measured COD value at the corresponding point (range: 8-15 mg / L);

[0124] Step 2: Data standardization;

[0125] Calculate the mean (μ) and standard deviation (σ) of each variable:

[0126] For example :μ=1200,σ=300 → intensity 1500 is normalized to (1500-1200) / 300=1.0;

[0127] COD: μ = 10.5 mg / L, σ = 1.8 → The measured value of 12.5 mg / L is normalized to (12.5-10.5) / 1.8≈1.11;

[0128] Step 3: Ridge parameter (λ) optimization;

[0129] The humic components were found ( / / ) were strongly correlated (correlation coefficient>0.8);

[0130] The optimal λ=0.3 is determined by 10-fold cross validation:

[0131]

[0132] Step 4: Solve for the regression coefficient;

[0133] The ridge regression objective function is minimized to solve:

[0134] Standardized coefficient: =0.82, =0.05, =0.12, =0.03;

[0135] Restore the original dimensions:

[0136]

[0137] For example : ;

[0138] Step 5: Calculate the intercept term;

[0139]

[0140] Final ridge regression equation

[0141]

[0142] Table 1 Path coefficients of various sources for DOM components

[0143]

[0144] Table 2 COD contribution of each source to DOM components

[0145]

[0146] Therefore, the COD in Erhai Lake comes from algae at about 45.52%, the COD from exogenous input is about 2.37%, and the COD released from sediment is about 9.95%.

[0147] Those skilled in the art will appreciate that, in addition to implementing the system, device, and various modules provided by the present invention in purely computer-readable program code, it is entirely possible to implement the same program in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, embedded microcontrollers, and the like by logically programming the method steps. Therefore, the system, device, and various modules provided by the present invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; the modules for implementing various functions can also be considered both software programs for implementing the method and structures within the hardware component.

[0148] 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 changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. A method for quantitatively tracing the source of chemical oxygen demand (COD) pollution in lake water, characterized in that: include: Step S1: Acquire water quality data, sediment data, and pollution load data of rivers entering the lake, wherein the water quality data includes chemical oxygen demand (COD), water temperature, pH, dissolved oxygen, total nitrogen, total phosphorus, soluble total nitrogen, orthophosphate, algal cell density, and chlorophyll a; the sediment data includes total nitrogen, total phosphorus, and organic matter content; and the river data includes total nitrogen, total phosphorus, and COD load entering the lake per unit time; Step S2: Perform three-dimensional fluorescence spectral analysis on water samples at different points in the lake to obtain fluorescence excitation / emission matrix spectrum EEMS data; Step S3: Performing Parallel Factor Analysis (PARAFAC) on the EEMS data to identify and classify the fluorescent components of dissolved organic matter (DOM). to The characteristics of each component type were determined by comparing with the Open Fluor organic fluorescence spectral library; Step S4: Based on the fluorescent component to The maximum fluorescence intensity and the corresponding COD value were compared, and the standardized multiple linear regression equation was established using ridge regression analysis to calculate the contribution percentage of each fluorescent component to the COD of the lake water. Step S5: Partial least squares path model (PLS-PM) is used to analyze the impact path of potential variables on DOM migration and transformation. The potential variables include phytoplankton, nutrient status, sediment, rivers entering the lake, and environmental variables. The direct and indirect contributions of different pollution sources to each fluorescent component are quantified by path coefficients. Step S6: Calculate the total contribution of rivers entering the lake, sediment release, and phytoplankton to the COD of the lake water by combining the contribution percentage of each fluorescent component to the COD of the lake water and the path coefficients of different pollution sources; The specific method of the ridge regression analysis in step S4 is: With COD as the dependent variable, fluorescent components to As the independent variable, the ridge parameter is introduced to solve the multicollinearity problem; firstly, the data is standardized to eliminate the dimension difference, and the formula is: in, is the intensity of the jth fluorescence component of the i-th sample, 、 are the mean and standard deviation of the jth component respectively; is the COD value of the i-th sample, 、 is the mean and standard deviation of COD; The standardized COD value As the dependent variable, the standardized fluorescence component As the independent variable, establish a ridge regression model: in, is the error term; The regression coefficients are solved by minimizing the following objective function : in, is the ridge parameter, which is used to control the regularization strength; m is the number of samples, and n is the number of fluorescent components; Use cross-validation to determine the optimal λ, divide the data set into a training set and a validation set, traverse the candidate values of λ, and calculate the mean square error of the validation set: in, Indicates traversing all samples in the validation set; Indicates the number of samples in the validation set; Select the λ that minimizes the MSE as the optimal parameter; The standardized regression coefficients Restore the coefficient to the original dimension , the formula is: The final standardized ridge regression equation is: Among them, the intercept term ; Contribution percentage of each fluorescent component to COD Calculated by the following formula: in, is the regression coefficient of the kth fluorescent component, It represents the absolute value of the regression coefficient of the jth component, reflecting the relative weight of its impact on COD.

2. The method for quantitatively tracing the source of chemical oxygen demand (COD) pollution in lake water according to claim 1, characterized in that: The parameters of the three-dimensional fluorescence spectrum analysis in step S2 include: The excitation wavelength scanning range is 200nm~400nm, and the emission wavelength scanning range is 300nm~600nm; The data intervals are 2nm for excitation and 1nm for emission, and the bandwidth of excitation and emission light is 5nm; Ultrapure water was used for blank correction every 10 water samples measured, and the scanning speed was 6000 nm / min.

3. The method for quantitatively tracing the source of chemical oxygen demand and endogenous pollution in lake water according to claim 1, characterized in that: The implementation of PARAFAC analysis in step S3 includes: Data processing was performed using Matlab software and drEEM toolbox; The preprocessing steps include subtracting Rayleigh scattering, Raman scattering, correcting for internal filter effects, normalizing Raman units, and removing noise areas; The number of fluorescent components was determined by core consistency check and split verification, and 40 iterations were performed using non-negative constraints. The convergence criterion was set to .

4. The method for quantitatively tracing the source of chemical oxygen demand (COD) pollution in lake water according to claim 1, characterized in that: The external model indicators of the PLS-PM model in step S5 include: The reflectance index of DOM is the maximum fluorescence intensity of each fluorescent component; The reflectance indicators of phytoplankton are algae density and chlorophyll a concentration; Indicators reflecting nutritional status are total nitrogen, total soluble nitrogen, total phosphorus, and orthophosphate; The reflective indicators of environmental variables are pH, water temperature, and dissolved oxygen; The reflective indicators of sediment are organic matter, total nitrogen and total phosphorus content; The reflective indicators of rivers entering the lake are total nitrogen, total phosphorus and COD loads.

5. The method for quantitatively tracing the source of chemical oxygen demand (COD) pollution in lake water according to claim 1, characterized in that: The significance of the path coefficient in step S5 was verified by 1000 bootstrap resamplings, and the model was determined to be valid when the goodness of fit (GOF) index was greater than 0.

7.

6. The method for quantitatively tracing the source of chemical oxygen demand (COD) pollution in lake water according to claim 1, characterized in that: The calculation method of the total contribution of pollution sources in step S6 is: For each fluorescent component , the pollution source contribution is the COD contribution value of the component multiplied by the path coefficient of the corresponding pollution source; The total contribution is the sum of the product of the direct path coefficient and the indirect path coefficient.

7. A quantitative tracing system for chemical oxygen demand and internal and external pollution in lake water, characterized by: The method for quantitatively tracing the source of chemical oxygen demand and internal and external pollution of lake water bodies according to any one of claims 1 to 6 comprises: Data acquisition module, used to obtain monitoring data on lake water quality, sediment and rivers entering the lake; Spectral analysis module, used to process three-dimensional fluorescence spectral data and perform PARAFAC analysis; Regression modeling module, used to construct the ridge regression equation and calculate the COD contribution percentage of each fluorescent component; Path analysis module, used to run the PLS-PM model and quantify the pollution source path coefficients; The contribution calculation module is used to integrate the regression and path analysis results and output the total contribution and proportion of internal and external pollution to COD.

8. The system for quantitative tracing of chemical oxygen demand and internal and external pollution of lake water according to claim 7 is characterized in that: The spectrum analysis module further includes: Scattering correction unit, used to subtract Rayleigh scattering and Raman scattering; Noise filtering unit, used to remove abnormal spectral regions; The component identification unit is used to match fluorescent component types using the Open Fluor spectral library.

9. The system for quantitative tracing of chemical oxygen demand and internal and external pollution of lake water according to claim 7, characterized in that: It also includes a visualization module for displaying the contribution of pollution sources in spatiotemporal dynamic charts and generating governance strategy optimization reports.

Images