A source apportionment method for per- and polyfluoroalkyl substances in surface water

Abstract

The application discloses a source analysis method for perfluoro-polyfluorinated compounds in surface water, and belongs to the field of pollutant source analysis in surface water. The method integrates PMF / FCM-SVR and POI, solves the problems of noise sensitivity of existing source analysis models, nonlinear and high-dimensional data regression, and can perform non-negative constraint on data, determine the contribution rate of various pollution sources to perfluoro-polyfluorinated compounds, clarify the number of manufacturing POIs and analyze the correlation between the two, and finally realize the effect of perfluoro-polyfluorinated compound source analysis. Moreover, by including spatial information and various forms of auxiliary data in the model to expand the correlation data, the contribution weight of various perfluoro-polyfluorinated compound pollution sources is better defined, thereby realizing the comprehensiveness and integrity of water pollution information, and providing help for later environmental governance.

Description

Technical Field

[0001] The invention relates to a source apportionment method for perfluorinated and polyfluorinated compounds in surface water, belonging to the field of source apportionment of pollutants in surface water. Background Art

[0002] Perfluorocarbons (PFAs) are a class of fluorinated chemicals known for their high stability and durability. Because they are difficult to degrade in the environment, PFAs are widely used in industrial applications such as food packaging, tableware coatings, and stain-resistant furniture. In recent years, the large-scale production and use of PFAs has led to increasing concentrations in natural water bodies. Due to their biotoxicity and high stability, these compounds pose a serious threat to the ecological environment and human health.

[0003] Due to the difficulty and workload of field investigations into pollution sources, currently, the main methods for source apportionment of perfluorinated and polyfluorinated compounds (PFFCs) in the aquatic environment include the ratio method and the absolute principal component-multiple linear regression model (APCS-MLR). The ratio method uses the ratios of PFHpA / PFOA, PFOA / PFOS, and PFOS / PFOA to identify the source of PFFCs. This method is not universally applicable and requires prior knowledge of the source categories in the target area. The APCS-MLR method does not require precise information on the specific emission sources. Instead, it performs principal component analysis on preprocessed data, calculates the absolute principal component score (APCS) for each sample on a selected principal component, and uses the APCS as the independent variable in a multiple linear regression analysis of the target variable. Through regression analysis, the contribution of each principal component to the target variable is determined, and a regression equation is established. However, the APCS-MLR model requires that the data meet the assumptions of multiple linear regression, such as linearity and normal distribution. If these assumptions are not met, the model may fail. APCS-MLR is also sensitive to noise in the data, which may lead to inaccurate principal component analysis results. In addition, APCS-MLR does not impose non-negativity constraints on each component, so the final results may be difficult to interpret and the physical or chemical meanings of the principal components may be unclear. Summary of the Invention

[0004] In order to solve the problems that existing source apportionment methods for perfluorinated and polyfluorinated compounds in surface water lack consideration of field spatial pollution source information, the source apportionment results are out of touch with reality, the ability to handle outliers and noise is limited, the data volume is large and meets linear requirements, and the non-negativity of the data cannot be constrained, making it difficult to analyze specific pollution sources or pollution source types, the present invention provides a source apportionment method for perfluorinated and polyfluorinated compounds in surface water. The technical solution is as follows:

[0005] Step 1: Collect water samples from the target area and pre-treat the perfluorinated and polyfluorinated compounds to be tested;

[0006] Step 2: Analyze the pre-treated water sample to obtain the concentration data of the perfluorinated and polyfluorinated compounds to be tested;

[0007] Step 3: Screen the number of different types of POIs around the target area from the electronic map, perform linear fitting on the concentration data of the PFAS and the number of each type of POI, and obtain the correlation between human activities and PFAS concentrations;

[0008] Step 4: Further divide the POI into several specific industry categories, and perform Origin descriptive statistics and Spearman correlation analysis on the PFAS concentration data and the specific industry quantity to obtain the correlation between PFAS in the target area and specific industries;

[0009] Step 5: Inputting the perfluoro-polyfluoro compound concentration data and uncertainty into the PMF model to obtain contribution information of the perfluoro-polyfluoro compound concentration data;

[0010] Step 6: Based on the correlation between perfluoro-polyfluoro compounds and specific industries obtained in step 4, the contribution information of specific industries to each source is output as the perfluoro-polyfluoro compound source analysis result.

[0011] Optionally, the calculation formula for the uncertainty UNc of perfluoro-polyfluoro compounds in step 5 is:

[0012] When the concentration is less than or equal to MDL:

[0013]

[0014] When the concentration is greater than MDL:

[0015]

[0016] Where MDL is the method detection limit; g is the error fraction, and d represents the concentration of perfluorinated and polyfluorinated compounds.

[0017] Optionally, the basic equation of the PMF model is as follows:

[0018] A=BC+K

[0019] Wherein: the sample concentration matrix A is an n×m matrix, n is the number of samples, and m is the number of perfluorinated and polyfluorinated compound types; the source contribution rate matrix B is an n×p matrix, p is the number of perfluorinated and polyfluorinated compounds; the source component spectrum matrix C is a p×m matrix; and the residual matrix K is an n×m matrix.

[0020] Optionally, in step 2, the concentration of perfluoro-polyfluoro compounds in the water sample is analyzed using high performance liquid chromatography tandem mass spectrometry.

[0021] Optionally, step 5 includes: decomposing the perfluoro-polyfluoro compound concentration data matrix by running the PMF model 20 times, and then setting the optimal factors to 5 to obtain factor distribution results and factor contribution results, and using the error estimation method in PMF to evaluate the deviation and uncertainty of the PMF results.

[0022] The present invention also provides a method for source apportionment of perfluorinated and polyfluorinated compounds in surface water, comprising:

[0023] Step 1: Collect water samples from the target area and pre-treat the perfluorinated and polyfluorinated compounds to be tested;

[0024] Step 2: Analyze the pre-treated water sample to obtain the concentration data of the perfluorinated and polyfluorinated compounds to be tested;

[0025] Step 3: Screen the number of different types of POIs around the target area from the electronic map, perform linear fitting on the concentration data of the PFAS and the number of each type of POI, and obtain the correlation between human activities and PFAS concentrations;

[0026] Step 4: Further divide the POI into several specific industry categories, and perform Origin descriptive statistics and Spearman correlation analysis on the PFAS concentration data and the specific industry quantity to obtain the correlation between PFAS in the target area and specific industries;

[0027] Step 5: Clustering the concentration data of perfluorinated and polyfluorinated compounds using the FCM model to obtain several sources of perfluorinated and polyfluorinated compounds in the target area and the contribution rate of each source;

[0028] Step 6: using the SVR algorithm to compare the measured and predicted values ​​of perfluoro-polyfluoro compounds to evaluate the performance of the FCM model;

[0029] Step 7: Based on the correlation between perfluoro-polyfluoro compounds and specific industries obtained in step 4, the spatial correlation information of perfluoro-polyfluoro compound sources is output as the perfluoro-polyfluoro compound source analysis result.

[0030] Optionally, the objective function of the FCM model is:

[0031]

[0032] where X = {x1, ..., x n} is the input perfluorinated-polyfluorinated compound concentration dataset; D = {d1, ..., d k} represents the set of each cluster center; the element u of Uzk represents the membership of the kth sample to the zth cluster; w is the number of identified sources, and the optimal value of w is obtained by trial and error.

[0033] Optionally, the regression function of the SVR algorithm is:

[0034] f(x z )=H T β(x z )+E

[0035] The weight vector and bias vector are represented by variable coefficients H and E respectively, β(x z ) is x z Mapping function to a hyperplane.

[0036] Optionally, step 6 uses the root mean square error RMSE and R 2 It reflects the deviation between the measured value and the predicted value and the fit of the model.

[0037] Optionally, the method further comprises: standardizing and normalizing the perfluoro-polyfluoro compound concentration data before inputting into the FCM.

[0038] The beneficial effects of the present invention are:

[0039] By integrating PMF / FCM-SVR and POI, the present invention solves the problems of existing source apportionment models being sensitive to data noise, nonlinearity, and high-dimensional data regression. It can also impose non-negativity constraints on the data, determine the contribution rates of various pollution sources to perfluoro-polyfluoro compounds, clarify the number of POIs in each manufacturing industry, and analyze their correlation with perfluoro-polyfluoro compounds, ultimately achieving the effect of perfluoro-polyfluoro compound source apportionment. Moreover, by expanding the correlation data by including spatial information and various forms of auxiliary data in the model, the contribution weights of various perfluoro-polyfluoro compound pollution sources are better defined, thereby achieving the comprehensiveness and completeness of water pollution information, and providing assistance for subsequent environmental governance. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 Flowchart of the perfluoro-polyfluoro compound source apportionment method of the present invention.

[0042] Figure 2is a schematic diagram of the relevance between the perfluoro-polyfluorinated compounds and the industry provided by the embodiment of the present application.

[0043] Figure 3 is a parameter average source allocation result diagram of POI-FCM-SVR provided by the embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0045] Embodiment one:

[0046] The embodiment provides a perfluoro-polyfluorinated compound source analysis method, referring to Figure 1 , the method comprises the following steps:

[0047] Step 1: Sampling the surface water of the target area, laying monitoring points, and collecting water sample samples.

[0048] In the process of water sampling, in order to exclude the influence of objective environment on data, the relationship between time and position should be considered during sampling, for example, the embodiment totally samples 180 samples.

[0049] Step 2: Selecting the types of pollutants, determining the target substances to be analyzed, and pretreating the target substances to be measured in the collected water sample samples.

[0050] In the embodiment, the target pollutants are determined as 11 perfluoro-polyfluorinated compounds, including: perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutyl sulfonic acid (PFBS), perfluorohexyl sulfonic acid (PFHxS) and perfluorooctyl sulfonic acid (PFOS), sodium perfluorooctanesulfonyl (OBS), perfluorohexyl ethyl sulfonic acid (6:2FTSA) and hexafluoropropylene oxide dimer (HFPO-DA) for concentration detection. The blank analysis results are all less than the detection limit, and the blank and sample standard addition recovery rates are controlled in 90% to 120%; the detection limits of various substances in the sample are 0.011 to 2.063 ng / L.

[0051] The pretreatment process of the embodiment mainly comprises:

[0052] Step 21: passing 1L sample through a 0.45μm filter membrane;

[0053] Step 22: installing a solid phase extraction column on a solid phase extraction device, and sequentially adding 4mL of 0.1% ammonia / methanol solution, 4mL of methanol and 4mL of pure water, and always keeping the column head wet during the solvent adding process;

[0054] Step 23: Elute the solid phase extraction column with 4 mL of a 25 mmol / L ammonium acetate aqueous solution (pH = 4);

[0055] Step 24: After the solid phase extraction column is dried by vacuum filtration, it is sequentially eluted with 4 mL of methanol and 4 mL of 0.1% ammonia / methanol solution, and the eluate is collected in a 15 mL polypropylene centrifuge tube;

[0056] Step 25: Concentrate the collected eluent with nitrogen in a nitrogen blow dryer. Set the water bath temperature to 50°C, dilute to 1 mL with methanol, filter through a 0.22 μm filter, and place in a 1.5 mL brown injection vial. Store at 4°C.

[0057] Step 3: Analyze the test solution using high performance liquid chromatography tandem mass spectrometry to obtain the concentration data of perfluorinated and polyfluorinated compounds in the water sample.

[0058] Step 4: Filter the number of POIs of different categories around the monitoring points from the electronic map, use different POI categories to represent different human activities, and establish a correlation between human activities and PFAS.

[0059] Step 41: Obtain a key from the electronic map open platform and use the POI data capture tool "POIKit" to capture POI information from the electronic map (https: / / www.amap.com / ).

[0060] Step 42: Keyword search uses EasyPoi (version_v10.93), with a search radius of 2500m. Set keywords such as "company enterprise", "transportation facility service" and "accommodation service", etc., and search according to their corresponding POI codes.

[0061] Step 43: Filter out the number of POIs in the accommodation service category, the transportation category, the agriculture, forestry, fishery, and animal husbandry category, and the company and enterprise category, and perform linear fitting between perfluorinated and polyfluorinated compounds and the number of POIs in the above four categories to explore the relationship between PFASs and residential life, transportation, agriculture, forestry, fishery, and animal husbandry, and industry.

[0062] In this embodiment, perfluoro-polyfluoro compounds have a significant correlation with corporate POIs.

[0063] Step 44: Classify company-related POIs into 13 manufacturing industries, including textiles, chemical fiber manufacturing, and metal manufacturing. Perform Origin descriptive statistics and Spearman correlation analysis on the PFAS concentration data and the 13 POI types around the monitoring points to obtain the correlation between PFAS in the target area and various manufacturing industries.

[0064] The present embodiment uses Origin software for descriptive statistics and Spearman correlation analysis, and the concentration data of perfluoro-polyfluorinated compounds in the sample water body are summarized for statistical analysis. The Spearman correlation analysis is used to explore the significant relationship between perfluoro-polyfluorinated compounds and the number of manufacturing POIs, and the significance is represented by p value, p<0.05 represents significant correlation, and p<0.01 represents highly significant correlation, Figure 2 A basic characteristic information diagram of the correlation between perfluoro-polyfluorinated compounds and manufacturing industries provided by the present embodiment is shown in the figure.

[0065] Step 5: Input the concentration data and uncertainty of perfluoro-polyfluorinated compounds into the PMF model to obtain the contribution information of the concentration data of perfluoro-polyfluorinated compounds.

[0066] Wherein, the basic equation of PMF model is as follows:

[0067] A=BC+K

[0068] In the formula, the sample concentration matrix A is an n×m matrix, n is the number of samples, and m is the number of perfluoro-polyfluorinated compound species; the source contribution rate matrix B is an n×p matrix, p is the number of perfluoro-polyfluorinated compounds; the source component spectrum matrix C is a p×m matrix; and the residual matrix K is an n×m matrix.

[0069] The calculation formula of the uncertainty (UNc) of perfluoro-polyfluorinated compounds is as follows:

[0070] When the concentration of pollutants is less than or equal to MDL:

[0071]

[0072] When the concentration of pollutants is greater than MDL:

[0073]

[0074] In the formula, MDL is the method detection limit; g is the error score, and d represents the concentration of PFASs, ng / L.

[0075] The input data of the PMF model are the concentrations of perfluoro-polyfluoro compounds and the results obtained from the uncertainty calculation formula. The output data are factor / contribution plots, factor fingerprint plots, residual analysis, etc. The perfluoro-polyfluoro compound concentration data matrix was decomposed by running the PMF model 20 times, and then the optimal factors were set to 5 to obtain the factor distribution results and factor contribution results. At the same time, three error estimation methods in PMF were used: BS (Bootstrap), DISP (Displacement), and BS-DISP (Bootstrap-Displacement) to evaluate the bias and uncertainty of the PMF results. DISP determines the rationality of the PMF source apportionment results, and BS identifies factors with poor reproducibility. The main sources of uncertainty are identified by the DISP interval and the DISP interval in BS-DISP.

[0076] Step 6: Based on the correlation between the above POIs and the manufacturing industry, the contribution information of the manufacturing industry to each source is output as the PFPC source apportionment results. The results are shown in Table 1:

[0077] Table 1 PMF-POI source resolution output results of this embodiment

[0078]

[0079] This example solves the problem of ignoring the pollution sources around the study area and the non-negative constraints of data during source apportionment by integrating PMF and POI. It determines the contribution rate of various pollution sources to perfluoro-polyfluoro compounds, clarifies the number of POIs in each manufacturing industry, and analyzes their correlation with perfluoro-polyfluoro compounds. Ultimately, it achieves the effect of perfluoro-polyfluoro compound source apportionment, improves the accuracy of the analysis data and results, makes the results more practical, and provides assistance for subsequent environmental governance.

[0080] Example 2:

[0081] This example provides a method for source apportionment of perfluoro-polyfluoro compounds, see Figure 1 , the method comprising:

[0082] Step 1: Sampling the surface water in the target area, setting up monitoring points, and collecting water samples.

[0083] During water sampling, in order to eliminate the impact of the objective environment on the data, the relationship between time and location should be considered during sampling. For example, in this embodiment, a total of 180 samples were collected.

[0084] Step 2: Select the type of pollutant, determine the target substance to be analyzed, and pre-treat the target substance in the collected water sample.

[0085] This example identified 11 perfluoro-polyfluoro compounds as target pollutants, including perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutylsulfonic acid (PFBS), perfluorohexylsulfonic acid (PFHxS), perfluorooctanesulfonic acid (PFOS), sodium perfluorononenyloxybenzenesulfonate (OBS), perfluorohexylethylsulfonic acid (6:2FTSA), and hexafluoropropylene oxide dimer (HFPO-DA). Blank analysis results were all below the detection limit, and the recoveries for both blanks and samples were controlled between 90% and 120%. The detection limits for each substance in the samples ranged from 0.011 to 2.063 ng / L.

[0086] The pre-processing process of this embodiment mainly includes:

[0087] Step 21: Pass 1 L of sample through a 0.45 μm filter;

[0088] Step 22: Install the solid phase extraction column on the solid phase extraction device and add 4 mL of 0.1% ammonia / methanol solution, 4 mL of methanol, and 4 mL of pure water in sequence, keeping the column head moist during the addition of solvents;

[0089] Step 23: Elute the solid phase extraction column with 4 mL of a 25 mmol / L ammonium acetate aqueous solution (pH = 4);

[0090] Step 24: After the solid phase extraction column is dried by vacuum filtration, it is sequentially eluted with 4 mL of methanol and 4 mL of 0.1% ammonia / methanol solution, and the eluate is collected in a 15 mL polypropylene centrifuge tube;

[0091] Step 25: Concentrate the collected eluent with nitrogen in a nitrogen blow dryer. Set the water bath temperature to 50°C, dilute to 1 mL with methanol, filter through a 0.22 μm filter, and place in a 1.5 mL brown injection vial. Store at 4°C.

[0092] Step 3: Analyze the test solution using high performance liquid chromatography tandem mass spectrometry to obtain the concentration data of perfluorinated and polyfluorinated compounds in the water sample.

[0093] Step 4: Filter the number of POIs of different categories around the monitoring points from the electronic map, use different POI categories to represent different human activities, and establish a correlation between human activities and PFAS.

[0094] Step 41: Obtain a key from the electronic map open platform and use the POI data capture tool "POIKit" to capture POI information from the electronic map (https: / / www.amap.com / ).

[0095] Step 42: Keyword search uses EasyPoi (version_v10.93), with a search radius of 2500m. Set keywords such as "company enterprise", "transportation facility service" and "accommodation service", etc., and search according to their corresponding POI codes.

[0096] Step 43: Filter out the number of POIs in the accommodation service category, the transportation category, the agriculture, forestry, fishery, and animal husbandry category, and the company and enterprise category, and perform linear fitting between perfluorinated and polyfluorinated compounds and the number of POIs in the above four categories to explore the relationship between PFASs and residential life, transportation, agriculture, forestry, fishery, and animal husbandry, and industry.

[0097] In this embodiment, perfluoro-polyfluoro compounds have a significant correlation with corporate POIs.

[0098] Step 44: Classify company-related POIs into 13 manufacturing industries, including textiles, chemical fiber manufacturing, and metal manufacturing. Perform Origin descriptive statistics and Spearman correlation analysis on the PFAS concentration data and the 13 POI types around the monitoring points to obtain the correlation between PFAS in the target area and various manufacturing industries.

[0099] This example uses Origin software for descriptive statistics and Spearman correlation analysis, and uses the concentration data of perfluorinated and polyfluorinated compounds in sample water bodies for summary statistics; Spearman correlation analysis is used to explore the significant relationship between perfluorinated and polyfluorinated compounds and the number of POIs in each manufacturing industry. Its significance is represented by the p value, where p < 0.05 indicates a significant correlation and p < 0.01 indicates a highly significant correlation. Figure 2 This embodiment provides a schematic diagram of the basic characteristic information of the correlation between perfluoro-polyfluoro compounds and the manufacturing industry.

[0100] Step 5: Use the FCM-SVR model to process the PFAS concentration dataset and obtain source assignment results for PFAS in the study area. The FCM algorithm was built in Python 3.8 with an initial fuzziness of 2 and 100 iterations. The SVR model was built using the LIBSVM toolkit.

[0101] Step 51: The perfluoro-polyfluoro compounds are pre-standardized and normalized to eliminate the influence of the dimension, enhance the stability of the model and reduce the noise response.

[0102] Step 52: Use the FCM algorithm to cluster the perfluorinated and polyfluorinated compound dataset. The input data is the perfluorinated and polyfluorinated compound data. In this example, the number of clusters is 5, indicating that there are five sources of perfluorinated and polyfluorinated compounds in this area, and their source contribution rates are clearly defined. The results are as follows: Figure 3 shown.

[0103] The objective function of the FCM algorithm is:

[0104]

[0105] where X = {x1, ..., x n} is the input perfluorinated-polyfluorinated compound concentration dataset; D = {d1, ..., d k} represents the set of each cluster center; the element u of U zk represents the membership of the kth sample to the zth cluster; w is the number of identified sources, and the optimal value of w is obtained by trial and error.

[0106] Step 53: Use the SVR algorithm to compare the measured values ​​and predicted values ​​of perfluoro-polyfluoro compounds to evaluate the performance of the FCM model. The measured values ​​and predicted values ​​obtained by FCM are fitted and regressed using the kernel function to obtain R 2 and root mean square error (RMSE).

[0107] Before SVR, cross-validation and grid search are used to confirm the kernel function and optimize the regression parameters. The goal of SVR is to locate a hyperplane in the feature space so that the mapping of perfluoro-polyfluoro compound data to the hyperplane is as close to the true value as possible and to maximize the distance between the hyperplane and the sample points to enhance robustness and generalization ability. The regression function of SVR is:

[0108] f(x z )=H T β(x z )+E

[0109] The weight vector and bias vector are represented by variable coefficients H and E respectively, and β(x z ) is x z Mapping function to a hyperplane.

[0110] This example uses RMSE and R 2 It reflects the deviation between the measured value and the predicted value and the fit of the model.

[0111]

[0112]

[0113] in, is the measured value of perfluorinated and polyfluorinated compounds, is the predicted value of the model about perfluoro- polyfluorinated compounds, T' r is the actual observed mean, T' p is the predicted average of the perfluoro-polyfluorinated compound, and m is the total number of perfluoro-polyfluorinated compound samples.

[0114] Step 6: The correlation between perfluoro-polyfluorinated compounds and human activities, and the spatial correlation information of perfluoro-polyfluorinated compounds obtained by FCM-SVR are output as the source analysis results of perfluoro-polyfluorinated compounds, as shown in the following formula: Figure 3

[0115] The embodiment integrates FCM-SVR and POI, and solves the problems of sensitivity to data noise, non-linear and high-dimensional data regression of the existing source analysis model.

[0116] On the basis of the source analysis methods of embodiment one and embodiment two, both can be considered comprehensively, that is, integrating PMF, FCM-SVR and POI, which can solve the problems of sensitivity to data noise, non-linear and high-dimensional data regression of the existing source analysis model, and can perform non-negative constraint on the data, determine the contribution rate of various pollution sources to perfluoro-polyfluorinated compounds, clarify the number of manufacturing POIs and analyze the correlation between the two, and finally realize the effect of perfluoro-polyfluorinated compound source analysis. Moreover, by including spatial information and various forms of auxiliary data in the model to expand the correlation data, the contribution weight of various perfluoro-polyfluorinated compound pollution sources is better defined, thereby realizing the comprehensiveness and integrity of water pollution information, and providing help for later environmental governance.

[0117] Part of the steps in the embodiments of the application can be realized by software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk.

[0118] The above only describes the preferred embodiments of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.​

Claims

1. A method for source apportionment of perfluorinated and polyfluorinated compounds in surface water, characterized in that: The method comprises: Step 1: Collect water samples from the target area and pre-treat the perfluorinated and polyfluorinated compounds to be tested; Step 2: Analyze the pre-treated water sample to obtain the concentration data of the perfluorinated and polyfluorinated compounds to be tested; Step 3: Screen the number of different types of POIs around the target area from the electronic map, perform linear fitting on the concentration data of the PFAS and the number of each type of POI, and obtain the correlation between human activities and PFAS concentrations; Step 4: Further divide the POI into several specific industry categories, and perform Origin descriptive statistics and Spearman correlation analysis on the PFAS concentration data and the specific industry quantity to obtain the correlation between PFAS in the target area and specific industries; Step 5: Clustering the concentration data of perfluorinated and polyfluorinated compounds using the FCM model to obtain several sources of perfluorinated and polyfluorinated compounds in the target area and the contribution rate of each source; Step 6: Use the SVR algorithm to compare the measured values ​​and predicted values ​​of perfluoro-polyfluoro compounds to evaluate the performance of the FCM model; perform regression fitting on the measured values ​​and predicted values ​​obtained by FCM using the kernel function to obtain R 2 and root mean square error RMSE; Step 7: Based on the correlation between perfluoro-polyfluoro compounds and specific industries obtained in step 4, the spatial correlation information of perfluoro-polyfluoro compound sources is output as the perfluoro-polyfluoro compound source analysis result.

2. The method for source apportionment of perfluorinated and polyfluorinated compounds in surface water according to claim 1, characterized in that: The objective function of the FCM model is: in X ={ x 1,…, x n } is the input PFPC concentration dataset; D ={ d 1,…, d k } represents the set of each cluster center; U Elements u zk Indicates the z The samples belong to k The membership degree of each cluster; w is the number of sources identified, obtained by trial and error w The best value of .

3. The method for source apportionment of perfluorinated and polyfluorinated compounds in surface water according to claim 2, characterized in that: The regression function of the SVR algorithm is: The weight vector and bias vector are represented by variable coefficients H and E express, yes Mapping function to a hyperplane.

4. The method for source apportionment of perfluorinated and polyfluorinated compounds in surface water according to claim 3, characterized in that: Step 6 uses the root mean square error RMSE and R 2 It reflects the deviation between the measured value and the predicted value and the fit of the model.

5. The method for source apportionment of perfluorinated and polyfluorinated compounds in surface water according to claim 4, characterized in that: The method further includes: standardizing and normalizing the perfluoro-polyfluoro compound concentration data before inputting into the FCM.

Images