A system and method for tracing and analyzing typical organic pollution of a petroleum contaminated site

Abstract

The application discloses a kind of petroleum pollution site typical organic pollution tracing analysis system and method, contain a kind of NMFCMB composite receptor system.The system includes NMF module, coupling module and CMB module.The method includes the following steps: S1, data preprocessing is carried out using NMF module;S2, main component factor is extracted using NMF module;S3, non-negative constraint factor decomposition is carried out using NMF module;S4, main pollution source is analyzed using coupling module;S5, pollution source contribution rate is calculated and uncertainty analysis is carried out using CMB module;S6, iteration calculation is carried out repeatedly to step S1 to S5, until the convergence condition is satisfied.The application solves the problem of poor accuracy of source analysis results and poor stability of the number of pollution sources in the prior art.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of environmental management, and particularly relates to a typical organic pollution source tracing analysis system and method for a petroleum contaminated site. BACKGROUND

[0002] The contaminated sites generated by the petroleum industry have great ecological environmental pollution risks, which not only bring serious harm risks to ecological environmental safety, but also hinder the social and economic development of cities to some extent.

[0003] However, the contaminated sites are often complicated in hydrogeological conditions, and identifying the source risk brought by organic pollution is a basic prerequisite for designing a targeted comprehensive source tracing strategy. In order to determine the source of organic pollutants, many technologies and tools (receptor models and diffusion models) have been proposed in the past few decades. The diffusion model is a predictive model, which predicts the temporal and spatial variation of pollutants by inputting the emission data and related parameter information of each pollution source; the receptor model is a kind of technology that determines the contribution rate of each pollution source through chemical and microscopic analysis of the receptor sample, and the ultimate goal is to identify the pollution sources that contribute to the receptor and quantitatively calculate the sharing rate of each pollution source. Compared with the predictive diffusion model, the receptor model is a diagnostic model, which explains the past rather than the future, and can be used to realize the quantitative analysis of pollution sources.

[0004] The chemical mass balance model is a relatively mature and accurate pollution source identification method, but its application conditions are relatively strict. In addition to knowing the pollution source fingerprint in advance, the phenomenon of "mixed sources" cannot appear between each pollution source, which is often not consistent with the actual application conditions. SUMMARY

[0005] The technical problem to be solved by the present application is that the single pollution source analysis model in the prior art is not suitable for the typical organic pollution of a complex petroleum contaminated site, including poor accuracy of source analysis results and poor stability of the number of pollution sources.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] A typical organic pollution source tracing analysis system for a petroleum contaminated site comprises an NMFCMB composite receptor system, wherein the composite receptor system comprises an NMF module, a coupling module and a CMB module.

[0008] The NMF module comprises an initial receptor unit, a data processing unit, a factor extraction unit and a factor rotation unit.

[0009] The coupling module comprises an organic pollution source tracing identification unit.

[0010] The CMB module comprises a pollution source contribution rate calculation unit.

[0011] Preferably, in the NMF module, the production data stream is processed by four units of initial receptor unit, data processing unit, extraction of significant factor unit, and non-negative constraint factor rotation unit, to obtain linearly independent pollution source fingerprint spectrum.

[0012] Preferably, the data processing unit in the NMF module comprises the following calculation formula:

[0013]

[0014] wherein m is the number of sampling points, n is the number of pollutants, S is a diagonal matrix, R is a diagonal element, S ik is the diagonal matrix of the i-th row and the k-th column, R kj is the diagonal element of the k-th row and the j-th column, and are the measured value and the predicted value of the pollutant, respectively; is the uncertainty of measurement; is the weighted residual error.

[0015] Preferably, the extraction of significant factor unit in the NMF module comprises the following calculation formula:

[0016]

[0017] wherein p is the number of significant factors, m is the number of pollutants, is the eigenvalue.

[0018] Preferably, the non-negative constraint factor rotation unit in the NMF module comprises the following calculation formula:

[0019]

[0020] wherein, C i is the concentration of variable i; is the multiple regression constant of the pollutant i ; is the scaling value of the rotation factor based on sample data; is the multiple regression coefficient of the i-th pollution source of the pollutant i ; q is the contribution rate of the i-th pollution source to the heavy metal ; q represents the contribution rate of the i-th pollution source to the heavy metal i .

[0021] Preferably, in the coupling module, the pollution source fingerprint obtained from the NMF module is used as the pollution receptor data of the next module, the CMB module, and the positive factor decomposition data obtained after the pollution receptor data is processed by the CMB module is coupled by the organic pollution source tracing identification unit to obtain accurate pollution source data.

[0022] Preferably, the organic pollution source tracing identification unit in the coupling module is used to put the eigenvalues and principal component factors generated by the extraction of significant factor unit and the non-negative constraint factor rotation unit in the NMF module into the CMB module, analyze the pollution source by using the calculation formula of the pollution source contribution rate calculation unit in the CMB module, and compare the generated results with the pollution source setting simulation data to identify the pollution source.

[0023] Preferably, in the CMB module, the pollution source fingerprint output by the NMF module and the pollution receptor data output by the coupling module are used as the calculation data of the CMB module to obtain the pollution source type and the pollution source contribution rate data.

[0024] Preferably, the pollution source contribution rate calculation unit in the CMB module comprises the following calculation formula:

[0025]

[0026] wherein, is the standard deviation calculated by the receptor environment measurement value, and the maximum likelihood solution is the measurement value of the pollutant in the i th sampling point, j is the predicted value generated under the given model complexity.

[0027] A typical organic pollution source tracing analysis method for a petroleum contaminated site, wherein the method uses the typical organic pollution source tracing analysis system for the petroleum contaminated site, and comprises the following steps:

[0028] S1, data preprocessing is performed by using the initial receptor unit and the data processing unit of the NMF module to obtain typical pollutant data.

[0029] Preferably, the data preprocessing unit comprises identification, judgment and processing of untested items, missing items and outliers of data, and identification of pollutant variables unsuitable for model input according to the preprocessed data, and finally, the preprocessed receptor data is standardized.

[0030] Preferably, the typical pollutant data is used to calculate the pollution load in step S3 for use in combination with the CMB model analysis in step S5.

[0031] ​S2, extracting principal component factors from the typical pollutant data and obtaining pollutant receptor data eigenvalues by using the extraction of saliency factor unit and the non-negative constraint factor rotation unit in the NMF module.

[0032] S3, performing non-negative constraint factor decomposition on the principal component factors by using the non-negative constraint factor rotation unit of the NMF module to obtain pollution source load and pollution source fingerprint.

[0033] Preferably, the principal component factors and the pollutant receptor data eigenvalues are used for substituting into step S4 to analyze the pollution source by combining the pollution source load with the pollution source fingerprint and for step S5 of the CMB model analysis.

[0034] S4, analyzing the main pollution source by combining the pollution source load with the pollution source fingerprint by using the organic pollution source tracing identification unit of the coupling module.

[0035] Preferably, the organic pollution source tracing identification unit contains a pollutant load analysis method framework, and the pollutant load analysis method framework is based on the spatiotemporal distribution evolution characteristics of the organic pollutants depicted by the GIS interpolation, and combines the quantitative index method to comprehensively represent the pollution degree.

[0036] S5, calculating the contribution rate of the main pollution source and performing uncertainty analysis by using the pollution source contribution rate calculation unit of the CMB module.

[0037] Preferably, the pollution source contribution rate calculation unit combines the pollution source load and the pollution source fingerprint results of steps S3 and S4, and uses the CMB module to perform comprehensive pollution source analysis.

[0038] S6, repeating steps S1 to S5 to perform iterative calculation until a convergence condition is met.

[0039] Preferably, the uncertainty analysis uses the following uncertainty calculation formula to calculate the uncertainty:

[0040]

[0041] wherein, DL j is the detection limit of the pollutant j , CV j is the coefficient of variation of the pollutant j.

[0042] Preferably, the convergence condition is that the difference between in two consecutive iterations is less than 10 -8 , that is, the iterative calculation stopping criterion is met.

[0043] Compared with the prior art, the technical scheme provided by the present application has at least the following beneficial effects:

[0044] The petroleum contaminated site typical organic pollution source tracing analysis system provided by the present application improves the accuracy of the pollution source analysis result of the petroleum contaminated site, and verifies the organic pollution source tracing analysis system by using artificial simulation data, so as to ensure the accuracy of the pollution cause analysis.

[0045] The petroleum contaminated site typical organic pollution source tracing analysis method provided by the present application obtains the pollution source emission component spectrum and the corresponding uncertainty value by using the non-negative constraint factor decomposition, and then calculates the corresponding pollution source contribution rate by using the composite receptor system, so as to ensure that the analyzed pollution source result meets the preset petroleum contaminated site pollution characteristics, that is, the petroleum contaminated site typical organic pollution source tracing analysis system provided by the present application has a significantly improved analysis rate accuracy, and the analysis result is closer to the preset pollution source. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0047] Figure 1 A schematic diagram of a petroleum contaminated site typical organic pollution source tracing analysis system of embodiment 1;

[0048] Figure 2 An algorithm logic diagram of a petroleum contaminated site typical organic pollution source tracing analysis method;

[0049] Figure 3 A pollutant load analysis method framework;

[0050] Figure 4 A fingerprint spectrum of embodiment 1, and the error line indicates a 10% relative error;

[0051] Figure 5 A factor load diagram of the NMFCMB composite receptor system of embodiment 1;

[0052] Figure 6 A comparison of the preset source contribution value of embodiment 1 and the calculation value of the single CMB model and the NMFCMB composite receptor system;

[0053] Figure 7 A fingerprint spectrum of embodiment 2, and the error line indicates a 10% relative error;

[0054] Figure 8 A factor load diagram of the NMFCMB composite receptor system of embodiment 2;

[0055] Figure 9 Comparison analysis of the preset source contribution value of Example 2 with the single CMB model and the NMFCMB composite receptor system calculation value;

[0056] Figure 10 CMB source analysis calculation process for Example 1. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application and the technical problems solved by the embodiments will be described below in conjunction with the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments.

[0058] A typical organic pollution source tracing analysis method for a petroleum contaminated site using a NMFCMB composite receptor system, comprising the following steps:

[0059] S1, using the initial receptor unit and data processing unit of the NMF module to perform data preprocessing to obtain typical pollutant data.

[0060] Further, the data processing unit in the NMF module comprises the following calculation formula:

[0061]

[0062] Wherein, m is the number of sampling points, n is the number of pollutants, S is a diagonal matrix, R is a diagonal element, S ik is the diagonal matrix of the ith row and the kth column, R kj is the diagonal element of the kth row and the jth column, and are the measured value and the predicted value of the pollutant, respectively; is the measurement uncertainty; is the weighted residual error.

[0063] Further, the data preprocessing unit includes identification, judgment and processing of untested items, missing items and outliers of data, and identifies pollutant variables that are not suitable for model input according to the preprocessed data, and finally standardizes the preprocessed receptor data.

[0064] Further, the typical pollutant data is used to calculate the pollution load in step S3 for use in combination with the CMB model analysis in step S5.

[0065] S2, using the extraction of significant factor unit and non-negative constraint factor rotation unit in the NMF module to extract principal component factors from the typical pollutant data and obtain the eigenvalues of the pollution receptor data.

[0066] Further, the extraction of significant factor unit in the NMF module comprises the following calculation formula:

[0067]

[0068] wherein p is the number of significant factors; m is the number of pollutants; is the eigenvalue.

[0069] Further, the non-negative constraint factor rotation unit in the NMF module comprises the following calculation formula:

[0070]

[0071] wherein, C i is the concentration of variable i; is the multiple regression constant of pollutant i ; is the scaling value of the rotation factor based on sample data; is the multiple regression coefficient of the first i pollution source of pollutant q ; represents the contribution rate of the first q pollution source to heavy metal i .

[0072] S3, performing non-negative constraint factor decomposition on the principal component factors by using the non-negative constraint factor rotation unit of the NMF module to obtain pollution source load and pollution source fingerprint spectrum.

[0073] Further, the principal component factors and pollution receptor data eigenvalues are used for substitution in step S4 to analyze pollution sources by combining pollution source load with pollution source fingerprint spectrum and for step S5 CMB model analysis.

[0074] S4, using the organic pollution source tracing identification unit of the coupling module to analyze the main pollution sources by combining pollution source load with pollution source fingerprint spectrum.

[0075] Further, the organic pollution source tracing identification unit comprises a pollutant load analysis method framework as shown in Figure 3 , which is used to comprehensively represent the pollution degree according to the temporal and spatial distribution evolution characteristics of the organic pollutants depicted by GIS interpolation.

[0076] S5, using the pollution source contribution rate calculation unit of the CMB module to calculate the contribution rate of the main pollution sources and perform uncertainty analysis.

[0077] Further, the pollution source contribution rate calculation unit combines the pollution source load and pollution source fingerprint spectrum results of steps S3 and S4 and uses the CMB module to carry out comprehensive pollution source analysis.

[0078] Further, the pollution source contribution rate calculation unit in the CMB module comprises the following calculation formula:

[0079]

[0080] wherein, is the standard deviation calculated from the receptor environment measurement value, the maximum likelihood solution is the measurement value of the pollutant in the i th sampling point, j is the predicted value generated under the given model complexity.

[0081] S6, repeat steps S1 to S5 for iterative calculation until the convergence condition is met.

[0082] Preferably, the uncertainty analysis calculates the uncertainty using the following uncertainty calculation formula:

[0083]

[0084] wherein, DL j is the detection limit of the pollutant j , CV j is the coefficient of variation of the pollutant j.

[0085] Preferably, the convergence condition is that the difference between in two consecutive iterations is less than 10 -8 , i.e. the iterative calculation stopping criterion is met.

[0086] Embodiment 1

[0087] The geographical conditions of the Northeast region, characterized by low temperature (minimum temperature in winter below -30 degrees) and complex stratum structure, result in a large difference between the pollution source analysis results in the Northeast cold region and the actual situation. A single pollution source analysis model cannot accurately analyze the enrichment state of typical pollutants in oil pollution sites, which easily leads to a series of management difficulties such as poor remediation effect and difficult reuse management of subsequent pollution sites.

[0088] Therefore, the present embodiment 1 aims to solve the problem of inaccurate and unclear tracing of the current organic pollution source tracing technology in the Northeast cold region oil pollution site. From the comprehensive perspective of risk source, migration path and pollution receptor, a set of oil pollution site typical organic pollution source tracing analysis system suitable for the Northeast cold region is preliminarily established, and the system flow chart is as shown in Figure 1 .

[0089] Table 1 is the pollution investigation data and data source characteristic index of a certain oil pollution site in the Northeast cold region.

[0090] ​The sampling points in Table 1 are selected from A1-A5, B1-B4, C1-C6 ten groups of data as shown in the following table.

[0091] Table 1: Pollution investigation data of a petroleum contaminated site in northeast cold region

[0092]

[0093] 0, data preprocessing and model establishment

[0094] (1) Pollution source spectrum data processing

[0095] According to the pollution investigation data of a petroleum contaminated site in northeast cold region in Table 1, 10 groups of pollution source spectrum data are randomly generated by computer program, and are assumed to be represented by SP1, SP2, … SPN.

[0096] The generated pollution source spectrum array is grouped, wherein each group of source spectrum data SP1-SP5 includes 4 types of pollution sources, each type of pollution source contains 12 types of pollutants, which are represented by P1-P12; each group of source spectrum data SP6-SP10 includes 6 types of pollution sources, each type of pollution source contains 18 types of pollutants, which are represented by P1-P18.

[0097] Considering the non-collinearity requirement of the receptor model for pollution source spectrum, each group of artificially simulated pollution source spectrum is quite different, and there is no collinearity feature. Thus, a total of 10 groups of data matrices are obtained, of which the first five groups are 12x4 data matrices, and the last five groups are 18x6 data matrices.

[0098] (2) Receptor data processing

[0099] Using the above-mentioned 10 groups of artificially simulated pollution source spectrum array, 10 groups of corresponding receptor data can be generated. Here, it is specified that each group of receptor sample data generates 50 sample data, and the first five groups of receptor data contain 12 types of pollutant concentration records, and each group of receptor sample is a 50x12 data matrix; the last five groups of receptors contain 18 types of pollutant concentration records, and each group of receptor data is a 50x18 data matrix.

[0100] 2, using the typical organic pollution source tracing analysis method of petroleum contaminated site for model analysis

[0101] Figure 2 The algorithm logic diagram of the typical organic pollution source tracing analysis method of petroleum contaminated site for this embodiment is shown in the following figure, and the calculation formulas used in steps S1 to S5 are completed by the algorithm program in Figure 2 .

[0102] S1, data in Table 1 is substituted into the formula Data preprocessing is performed to obtain typical pollutant data. This typical pollutant data is then substituted into step S3 to calculate the pollution load, which is then used for analysis in conjunction with the CMB model in step S5.

[0103] S2. Substitute the typical pollutant data obtained in step S1 into the formula. and formula In this process, the characteristic values ​​of the pollution receptor data are calculated and the principal component factors are determined for analysis in the next step S5 using the CMB model.

[0104] S3. The principal component factors are decomposed using the non-negative constraint factor rotation unit of the NMF module to obtain the pollution source load and pollution source fingerprint spectrum.

[0105] S4. Substitute the eigenvalues ​​and principal component factors generated in steps S2 and S3 into the formula. The pollution source was analyzed by combining pollution source load with pollution source fingerprinting, and the results are as follows: Figure 6 As shown.

[0106] S5, The generated results and Figure 5 The system identifies pollution sources by comparing them with simulated data set against the same pollution source. The comparison results are then used to... Figure 6 The data is displayed together. Combining the load data generated in S2 and the pollution source data generated in S4 further improves the accuracy of the contribution rate.

[0107] S6. Repeat steps S1 to S5, using the formula. Perform iterative calculations until the convergence condition is met;

[0108] 3. Monte Carlo Simulated Receptor Sample Data Processing

[0109] The sample data is processed using the formulas in steps S3-S5. Considering the inherent uncertainties in both pollutant source spectrum data and receptor data in real-world environments, and the variations in the pollutant migration process from the emission source to the receptor, a Monte Carlo simulation method is used to generate receptor data for a more realistic simulation of practical applications.

[0110]

[0111] In the formula D ij That is, the concentration of the i-th type of pollutant in the j-th sample, taking into account receptor sample error; A ij The initial receptor sample concentration matrix is ​​obtained from the contaminant source spectrum and the defined source contribution matrix according to the formula; C ij erf is the coefficient of variation of the concentration of pollutant of type i in sample j; -1 It is the inverse Gaussian error function; R ij It is a random number between 0 and 1.

[0112] 4. Simulation verification results

[0113] In this embodiment, the Monte Carlo method is used to construct artificial simulation data, and the NMFCMB composite receptor system is verified by combining the pollution source load index.

[0114] The receptor data generated by the four pollution sources are used to verify the model. From the above generated first five groups of receptor data, a group is randomly selected. It is assumed that the second group is selected, the significant factor identification matrix is calculated, and the calculation results are shown in Table 1. The pollution source spectrum and the real contribution rate are analyzed, and the results are shown in Table 2. Figure 4

[0115] Figure 4 The real pollution source spectrum generated by artificial simulation and the corresponding pollution source contribution value are shown in the figure. The pollution source contribution value shows that there are four pollution sources.

[0116] Figure 5 The pollution source component spectrum obtained based on the NMFCMB composite receptor system shows that there are four pollution sources, which represent industrial pollution sources, agricultural pollution sources, natural emission sources and other mixed pollution sources.

[0117] Comparison Figure 4 and Figure 5 It can be seen that the principal component factor 1 represents the first pollution source in the pollution source spectrum, the principal component factor 2 represents the third pollution source in the pollution source, and the principal component factors 3 and 4 represent the fourth and second pollution sources, respectively.

[0118] After extracting the pollution source fingerprint spectrum, the pollution source contribution rate of the sample sequence is calculated by using the CMB model, Figure 6 The comparison of the four pollution source emission contributions of the artificial simulation preset and the model calculation is shown.

[0119] Table 2 is the significant factor identification matrix, including eigenvalue, determination coefficient, cumulative variance rate and Exner function, which shows that the pollution source analysis calculation result is consistent with the pollution background investigation result, indicating the accuracy of the pollution source analysis result.

[0120] Table 2

[0121]

[0122] Table 3 is the source analysis result of the artificial simulation receptor data NMFCMB composite receptor system, which shows the comparison of the source contribution rate preset value and the different model calculation values.

[0123] Table 3

[0124]

[0125] ​From the above calculation results, it can be seen that the pollution source fingerprint spectrum obtained by the NMFCMB composite receptor system can be identified according to the system in a relatively clear manner although there are some errors compared with the pre-set pollution source component spectrum. In the source contribution calculation process, the R variance is 0.88, the P / M value reaches 92%, and the residual value is 0.17. The model result is relatively reliable. The source contribution value calculated according to the pollution source fingerprint spectrum reversely obtained by the model is consistent with the pre-set source contribution value. Although there is a certain difference between the total source contribution rate and the pre-set source contribution rate, the relative error is within 15%. The specific calculation process is shown in Figure 10 .

[0126] Example 2

[0127] A set of 6-pollution-source receptor samples is randomly selected from the last five groups of receptor data generated above. In this embodiment, the eighth group is assumed to be selected. The real pollution source spectrum artificially generated and the corresponding pollution source contribution value are shown in Figure 8 .

[0128] Figure 7 The artificially generated real pollution source spectrum and the corresponding pollution source contribution value are shown in FIG. 8. As shown in the figure, the pollution source contribution value shows that there are 6 pollution sources.

[0129] Figure 8 The pollution source component spectrum obtained based on the NMFCMB composite receptor system is shown in FIG. 9. As shown in the figure, the pollution source contribution value shows that there are 6 pollution sources, which are a refinery, an oil production plant, a calcium carbide plant 1, a calcium carbide plant 2, a chemical fertilizer plant and natural emission.

[0130] Figure 9 The comparison diagram of the pollution source contribution value calculated by the NMFCMB composite receptor system and the pre-set source contribution value is shown in FIG. 10. As shown in the figure, the pollution source contribution value shows that there are 6 pollution sources.

[0131] The pollution source component spectrum obtained based on the NMFCMB composite receptor system and the sample sequence pollution source contribution value calculated are shown in Figure 8 , 9 As can be seen from the comparison of FIGS. 11 and 12, Figure 7 and Figure 8 , the principal component factors 1-6 correspond to the pre-set 6th, 2nd, 4th, 5th and 1st pollution sources respectively.

[0132] Table 4 shows the comparison of the pre-set source contribution rate and the model calculation value.

[0133] Table 4

[0134]

[0135] Similar to the simulation data of 4 pollution sources, the receptor data generated by 6 artificial pollution sources can be analyzed based on the NMFCMB composite receptor system to obtain a pollution source fingerprint spectrum with obvious characteristics, and the source contribution calculation value is consistent with the pre-set source contribution value.

[0136] The simulation verification results of the above two cases can show that the factor decomposition of the receptor data by using the NMFCMB composite receptor system can reduce the data dimension while maintaining the original data information, so as to obtain the component spectrum representing the pollution emission characteristics, and the CMB model based on the effective variance square method can calculate the relatively accurate source contribution rate. Therefore, under the condition of lacking source spectrum, the NMFCMB composite receptor system can improve the pollution source analysis accuracy.

[0137] The above is the preferred embodiment of the present application, it should be noted that for those skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can also be made, which should be considered as the protection scope of the present application.

Claims

1. A system for tracing and analyzing typical organic pollution of a petroleum contaminated site, comprising a NMFCMB composite receptor system, characterized in that, The composite receptor system comprises an NMF module, a coupling module and a CMB module; The NMF module comprises an initial receptor unit, a data processing unit, an extraction of significant factor unit and a non-negative constraint factor rotation unit, in the NMF module, the production data flow is processed by the initial receptor unit, the data processing unit, the extraction of significant factor unit and the non-negative constraint factor rotation unit to obtain linearly independent pollution source fingerprint spectrum; The data processing unit in the NMF module comprises the following calculation formula: ； where m is the number of sampling points, n is the number of pollutants, S is a diagonal matrix, R is a diagonal element, S ik is the diagonal matrix of the ith row and the kth column, R kj is the diagonal element of the kth row and the jth column, and are the measured value and the predicted value of the pollutant, respectively; is the uncertainty of the measurement; is the weighted residual error; The extraction of significant factor unit in the NMF module comprises the following calculation formula: ； where p is the number of significance factors; m is the number of pollutants; is the eigenvalue; The non-negative constraint factor rotation unit in the NMF module comprises the following calculation formula: ； wherein, C i is the concentration of the variable i; is the multiple regression constant of the pollutant i ; is the scaled value of the rotation factor based on the sample data; is the multiple regression coefficient of the pollutant i ; q is the multiple regression coefficient of the i-th pollution source of the pollutant ; q represents the contribution rate of the i-th pollution source to the heavy metal i ; The coupling module comprises an organic pollution source tracing recognition unit, in the coupling module, the pollution source fingerprint spectrum obtained from the NMF module is taken as the pollution receptor data of the next module CMB module, the pollution receptor data is processed by the CMB module to obtain positive factor decomposition data, the organic pollution source tracing recognition unit is coupled to the positive factor decomposition data to obtain accurate pollution source data; The organic pollution source tracing recognition unit in the coupling module is to substitute the eigenvalues and principal component factors generated by the extraction of significant factor unit and the non-negative constraint factor rotation unit in the NMF module into the CMB module, analyze the pollution source by using the pollution source contribution rate calculation unit in the CMB module, compare the generated result with the pollution source setting simulation data, and identify the pollution source; The CMB module comprises a pollution source contribution rate calculation unit, in the CMB module, the pollution source fingerprint spectrum output by the NMF module and the pollution receptor data output by the coupling module are taken as the calculation data of the CMB module to obtain pollution source types and pollution source contribution rate data; The pollution source contribution rate calculation unit in the CMB module comprises the following calculation formula: ； in, It is the standard deviation calculated from the receptor environmental measurements, the maximum likelihood solution. It is the first i Among the sampling points j The measured values ​​of pollutants, It is the predicted value generated given the model complexity.

2. A method for tracing and analyzing typical organic pollution of a petroleum contaminated site, characterized by, The method uses the petroleum contaminated site typical organic pollution source tracing analysis system of claim 1, and comprises the following steps: S1, data preprocessing is performed by using the initial receptor unit and the data processing unit of the NMF module to obtain typical pollutant data; S2, principal component factors are extracted from the typical pollutant data by using the extraction of significant factor unit and the non-negative constraint factor rotation unit in the NMF module, and eigenvalues of pollution receptor data are obtained; S3, the principal component factors are subjected to non-negative constraint factor decomposition by using the non-negative constraint factor rotation unit of the NMF module to obtain pollution source load and pollution source fingerprint spectrum; S4, the main pollution source is analyzed by using the organic pollution source tracing recognition unit of the coupling module to combine the pollution source load and the pollution source fingerprint spectrum; S5, the contribution rate of the main pollution source is calculated by using the pollution source contribution rate calculation unit of the CMB module, and uncertainty analysis is performed; S6, steps S1 to S5 are repeated for iterative calculation until the convergence condition is met.

3. The method according to claim 2, wherein the petroleum contaminated site is a typical organic pollution site. In step S1, the data preprocessing unit comprises identification, judgment and processing of untested items, missing items and outliers of data, and pollution variable unsuitable for model input is identified according to the preprocessed data, and finally the preprocessed receptor data is subjected to standardization processing; The typical pollutant data is used to substitute into step S3 to calculate the pollution source load for use in combination with the CMB model in step S5.

4. The method according to claim 2, wherein the petroleum contaminated site is a typical organic pollution site. In steps S2 and S3, the principal component factor and pollution receptor data eigenvalue are used to substitute into step S4 to analyze the pollution source load, combine the pollution source fingerprint, and analyze the pollution source, and are used in step S5 of the CMB model analysis.

5. The method according to claim 2, wherein the method is a method for tracing and analyzing typical organic pollutants in a petroleum contaminated site, characterized by, In step S4, the organic pollution source tracing recognition unit contains a pollution load analysis method framework, which is used to combine the quantitative index method to comprehensively represent the pollution degree according to the GIS interpolation depicted temporal and spatial distribution evolution characteristics of the organic pollutants.

6. The method according to claim 2, wherein the method is a method for tracing and analyzing typical organic pollutants in a petroleum contaminated site. In step S5, the pollution source contribution rate calculation unit combines the pollution source load and the pollution source fingerprint of steps S3 and S4, and uses the CMB module to carry out comprehensive pollution source analysis and analysis; The uncertainty analysis uses the following uncertainty calculation formula to calculate the uncertainty: ； wherein DL j is the limit of detection of the pollutant j , CV j is the coefficient of variation of the pollutant j.

7. The method according to claim 2, wherein the method is a method for tracing and analyzing typical organic pollutants in a petroleum contaminated site. In step S6, the convergence condition is that the difference between two successive iterations is less than 10 -8 i.e. the iteration calculation stopping criterion is met.​

Images