Atmospheric particulate matter source analysis method

Abstract

The invention discloses an atmospheric particulate matter source analysis method, which comprises the following steps: collecting samples, constructing an emission source component spectrum, carrying out data statistics, establishing a metal element source analysis model, and carrying out source analysis on air particulate matters by adopting a PPR (pentatricopeptide repeats) model. Taking the mass fraction of metal in an emission source in the air particulate matter emission source component spectrum as a PPR model modeling factor, taking the mass fraction of metal elements in a particulate matter sample as a dependent variable, performing data generation processing, optimizing a parameter matrix, obtaining a weight contribution rate of a factor xj to a dependent variable y, and calculating the weight contribution rate of the factor xj to the dependent variable y; the contribution rate of each particulate matter pollution emission source to the mass fraction of the metal elements at the sampling point is obtained. According to the air suspended particulate matter metal source analytical model constructed based on PPR, a large amount of data does not need to be measured at sampling points, the source analytical model can be constructed by measuring the metal element content of one air suspended particulate matter sample, the response time from sample collection to model construction is shortened, and the method has the advantages of being convenient to operate, rapid in calculation and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of atmospheric pollutant source tracing and pollution prevention and control, and in particular to a method for analyzing the source of atmospheric particulate matter. Background Art

[0002] The current status of environmental pollution in my country has made environmental problems increasingly prominent. Some of the difficult-to-degrade metal pollution has become the focus of research due to its damage to the environment and harm to human body. As for metal pollution, due to the invisible and colorless nature of atmospheric pollutants, it is easier to be ignored than metals in water or soil environment. The metal elements in the atmosphere are mainly enriched in suspended particles in the air. These particles migrate with the flow of air. Some larger air-suspended particles settle in the migration path, and the rest continue to suspend in the air and move with the airflow. At present, there are many studies on the content of metal elements in atmospheric settling particles and their sources, but there are few studies on metals in suspended particles accompanying airflow. In order to analyze the source of metal elements in air-suspended particles accompanying atmospheric movement, based on the receptor model in particulate matter source analysis, by analyzing the metal element components of the particulate matter samples at the sampling point and the emission source, the contribution value of the pollution emission source to the metal elements in the air-suspended particles at the sampling point is finally determined.

[0003] There are many types of receptor models for source analysis, such as principal component analysis, multivariate linear regression, and positive definite matrix factorization. Since the content of some metal elements in the suspended particulate matter of the emission source composition spectrum is extremely low or below the detection limit, there are many zero values ​​in the receptor modeling matrix data, and the composition spectrum data structure is complex, which cannot reflect the nonlinear situation in actual problems. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide an atmospheric particulate matter source analysis method with strong model analysis capability, high accuracy and no restrictions on the structure and characteristics of data.

[0005] The technical solution adopted by the present invention to solve the technical problem is: a method for source analysis of atmospheric particulate matter, comprising the following steps:

[0006] S1. Sample collection: Collect air particulate matter samples from sampling points in the area where the metal elements in the polluted particulate matter need to be analyzed, and record the sample collection location and time; the atmospheric particulate matter includes PM2.5, PM0 and TSP and other polluted particulate matter;

[0007] S2. Construct emission source composition spectrum: Take the sampling point as the starting point of the backward trajectory, use the backward trajectory method to calculate the backward trajectory of the air mass located at the sampling point within n hours before the start of sampling, calculate a backward trajectory every hour, and take the maximum straight-line distance from the starting point of the trajectory to the sampling point in the n air mass backward trajectories as the effective distance D; in the actual area of ​​composition spectrum sampling, take the sampling point as the center, and take the shortest distance from the sampling point to the boundary of the composition spectrum sampling area as the maximum effective distance D max , D≤D max ; The composition spectrum collects the corresponding atmospheric particulate matter emission source information and determines the mass fraction of metal elements in the atmospheric particulate matter emitted by the emission source, thereby constructing the metal element composition spectrum emitted by potential emission sources within the effective distance of the sampling point;

[0008] Taking the sampling point as the starting point of the backward trajectory, the backward trajectory method is used to calculate the n-hour backward trajectory of the air mass located at the sampling point n hours before the start of sampling. A backward trajectory is calculated every 1 hour, and the maximum straight-line distance from the starting point of the trajectory to the sampling point in the n air mass backward trajectories is taken as the effective distance D; the actual area of ​​component spectrum sampling is determined, and within the actual area of ​​component spectrum sampling, the shortest distance from the sampling point to the boundary of the component spectrum sampling area is taken as the maximum effective distance D with the sampling point as the center. max , when D≤D max When the n-hour value is determined, the composition spectrum should collect the corresponding atmospheric particulate matter emission source information and determine the mass fraction of metal elements in the atmospheric particulate matter emitted by the emission source, so as to construct the metal element composition spectrum emitted by the potential emission source within the effective distance of the sampling point;

[0009] S3, data statistics: experimentally analyze the particulate matter samples collected in step S1 to obtain the mass fraction of metal elements in the particulate matter samples, and standardize the obtained mass fraction of metal elements. The standardization method adopts the maximum and minimum standardization method, and the processing formula is as follows:

[0010]

[0011] Where x is the original data of the mass fraction of the metal element before processing, and x' is the standardized data;

[0012] S4. Establishing a metal element source analysis model: using a set of ridge functions G m (Z m ) and to approximate the regression function f(x), and establish a source analysis model of metal elements in the air pollution process based on PPR. The formula of f(x) is shown in formula (1) and formula (2). Formula (2) is the matrix form of the PPR model.

[0013]

[0014] Where M is the number of ridge functions, βm is the mth ridge function G m Contribution weight to the regression function f(x);

[0015] Z m As the independent variable of the ridge function, it is a P-dimensional vector exist The projection in the direction and satisfy Where j is the dimension of the P-dimensional vector, and its formula is shown in formula (3):

[0016]

[0017] S5. Use the PPR model to analyze the source of air particles:

[0018] S51. The mass fraction of metals in emission source j in the air particulate matter emission source composition spectrum is regarded as the factor x for PPR modeling. j ;

[0019] S52, the mass fraction of the metal element in the particulate matter sample obtained by the analysis in step S3 is regarded as the dependent variable y, and the mass fraction of the metal element x in the emission source composition spectrum in step S2 is regarded as the dependent variable y. j One-to-one correspondence with y;

[0020] S53, perform data generation processing, optimize parameter matrix, and obtain factor x j The weighted contribution rate to the dependent variable y is the contribution rate of each particulate matter emission source to the mass fraction of metal elements at the sampling point.

[0021] Furthermore, according to the atmospheric particulate matter source analysis method of claim 1, it is characterized in that: in step S53, a simulated annealing algorithm is used to optimize the parameters in the PPR model, that is, the PPR adopts the least square method as the minimization criterion, by selecting β in formula (2) m , G m (Z m ) and the number of ridge functions M, and design the objective function L2 so that the PPR model satisfies the criterion of formula (4).

[0022] The simulated annealing algorithm is used to optimize the parameter matrix (β1, β2, ..., β M )and Get factor x j The weighted contribution rate of the dependent variable y is the contribution rate of each particulate matter emission source to the mass fraction of metal elements at the sampling point, where E is the sum, It is a prediction of the mass fraction y of metal elements in the particulate matter sample.

[0023] Furthermore, the simulated annealing algorithm is implemented by programming in Matlab language.

[0024] Furthermore, the parameters of the simulated annealing algorithm when optimizing the PPR model are set as follows:

[0025] Population size 200; annealing form: T(t+1)=γT(t), γ=0.9, t is the number of iterations; initial temperature T0=1×1010, termination temperature T f =0; Acceptance probability formula: exp(Δf / T)>rand, where rand is a random number between (0, 1);

[0026] The simulated annealing algorithm is used to perform iterative optimization repeatedly. When the annealing temperature T f = 0, the iteration ends and the optimized parameter matrix is ​​obtained.

[0027] (β1,β2,...,β M )and

[0028] The contribution rates of the air particulate matter emission sources (j=1, 2, …, P) to the mass fraction of metal elements in the samples at the sampling points can be obtained as follows:

[0029] Furthermore, in the step S1, the sample is collected using a polypropylene membrane, and a medium-flow atmospheric sampler is used to continuously sample at a rate of 100 liters / minute for 24 hours; at the same time, a blank sample is collected, and the blank sample is obtained when the sampler is loaded with a filter membrane and is not turned on for 24 hours.

[0030] Furthermore, the polypropylene film is placed in a constant temperature and humidity chamber (20±1°C) and (50±5)% for 24 hours before and after sampling, and then weighed using a 100,000th balance. The mass difference between the polypropylene film before and after sampling is used as the mass of the air particulate sample.

[0031] Furthermore, in addition to the constant temperature and humidity treatment before and after weighing, the samples were stored in a freezer until laboratory analysis, and the storage temperature was set at -4°C.

[0032] Furthermore, the S3 data statistics include statistics of water-soluble and water-insoluble metal elements in particulate matter. Water-soluble metal elements on polypropylene film are measured by ion chromatograph; and water-insoluble metal elements on polypropylene film are measured by X-ray fluorescence spectrometer. The metal elements in the particulate matter sample on polypropylene film are divided into soluble and insoluble in water, and different methods are used to measure them respectively. First, cut half of the polypropylene film, cut it into pieces and place it in a stoppered tube, add deionized water, and the volume of deionized water corresponds to 100ml of deionized water added to 1 polypropylene film; then ultrasonically extract and filter the solution to constant volume, and use ion chromatograph to measure water-soluble metal ions. In addition, for water-insoluble metal ions on polypropylene film, use X-ray fluorescence spectrometer to measure their mass. Use the remaining half of the polypropylene film to press the sample into a sheet, and use X-ray fluorescence spectrometer to measure the content of metal elements contained in the air particulate matter on the polypropylene film. The mass of the particles on the entire filter membrane is the weight difference obtained by weighing the filter membrane before and after collecting the particles. The mass content (%) of the metal element in the collected particle sample is the mass of the corresponding metal measured by experimental analysis divided by the mass of the corresponding particle sample on the filter membrane.

[0033] Furthermore, M is the number of ridge functions, and its value is 2-5.

[0034] The beneficial effects of the present invention are:

[0035] 1. The present invention detects the percentage of metal elements in the air particulate matter of the sampling point and the potential emission sources around it, and constructs a metal source analysis model in the air particulate matter based on PPR, which can avoid the contradiction that only using a simple linear regression method cannot reflect the nonlinear situation in the actual problem, and the PPR model does not limit the structure and characteristics of the modeling data. It can be based on a modeling matrix with many zero values ​​in the metal source component spectrum of the pollution source, and a set of ridge functions are used to approximate the regression function, that is, the metal content in the particulate matter at the sampling point. The optimal model parameters are obtained through the optimization algorithm to judge the contribution of potential pollution emission sources to the metal content in the sampling point sample;

[0036] 2. The metal source analysis model of air suspended particulate matter constructed based on PPR does not need to measure a large amount of data (i.e., multiple samples) at the sampling point. The source analysis model can be constructed by measuring the metal element content of an air suspended particulate matter sample, which shortens the response time from sample collection to model construction. It has the advantages of easy operation and fast calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is the effective distance of potential pollution emission sources around the sampling point in Example 1. DETAILED DESCRIPTION

[0038] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0039] Embodiment 1:

[0040] Taking a sampling point in Shuangliu District, Chengdu City, Sichuan Province on December 31, 2015 as an example, the specific steps of the source analysis method of metal elements in suspended air particles at this sampling point are as follows:

[0041] 1. Collect air suspended particulate matter samples at the sampling point on a certain day. The sampling time was from 8:00 on December 31, 2015 to 8:00 on January 1, 2016. The sampling point was set on the top floor of the teaching and science building of the Aviation Port Campus of Chengdu University of Information Technology. The specific geographical location was 30°35′2.62″N, 103°59′19.5″E, and the sampling height was 15 meters. 48 hours before the start of sampling, the polypropylene filter membrane was placed in a constant temperature and humidity chamber (20±1℃), (50±5)% balanced for 24 hours, and the weight was weighed with a 100,000th balance to be 0.58364g; at 8:00 on December 26, 2015, a medium-flow atmospheric sampler was used at 100L·min -1 The samples were continuously sampled at a rate of 24 hours; the polypropylene film with the suspended particulate matter sample collected was placed in a constant temperature and humidity chamber (20±1°C) and balanced at (50±5)% for 24 hours, and the weight was weighed on a 100,000th balance to find it was 0.60127 g; finally, it was stored in a freezer until laboratory analysis, and the storage temperature was set to -4°C.

[0042] 2. Construct the composition spectrum of potential emission sources within the effective distance of the sampling point. Taking the sampling point as the starting point, the shortest distance D from the sampling point to the boundary of the sampling area for the composition survey is max The value is 105 km. Based on HYSPLIT, the backward trajectory method is used to calculate the 24-hour backward trajectory of the air mass at the sampling point from the start of sampling to the end of sampling. One backward trajectory is calculated every hour, and the maximum straight-line distance from the starting point of the trajectory to the sampling point in the 24 air mass backward trajectories is taken as the effective distance D. Figure 1 As shown, D is 100 km, D ≤ D max , n can be selected as 24 hours; then, through the combination of component spectrum investigation and remote sensing images, the main sources of air suspended particulate matter emission in the component spectrum collection area are metallurgical dust sources, comprehensive dust sources, soil dust sources, construction dust sources, coal smoke dust sources and traffic dust sources. After previous field tests and literature research, the mass percentage (%) of metal elements in the air particulate matter emitted by these pollution sources is as shown in Table 1. The metal element component spectrum emitted by potential emission sources within the effective distance of the sampling point is constructed.

[0043] Table 1 Metal element composition of potential emission sources within the effective distance around the sampling point

[0044] element Metallurgical Dust x1 (%) Comprehensive dust x2 (%) Soil dust x3(%) Construction dust x4 (%) Soot x5 (%) Traffic dust x6(%) Pb 1.30 0.00 0.00 0.02 0.13 0.18 Cd 0.01 0.00 0.00 0.00 0.00 0.00 Se 0.00 0.00 0.00 0.00 0.00 0.00 As 0.02 0.00 0.00 0.00 0.08 0.00 Cu 0.10 0.02 0.00 0.00 0.02 0.05 Ni 0.01 0.00 0.00 0.00 0.01 0.01 Co 0.00 0.00 0.00 0.00 0.00 0.00 Fe 14.60 1.26 5.41 2.03 2.90 0.00 Mn 1.14 0.02 0.10 0.07 0.03 0.03 Cr 0.14 0.00 0.01 0.07 0.02 0.10 V 0.02 0.00 0.01 0.00 0.03 0.00 Ti 0.05 0.06 0.44 0.65 0.74 0.17 Ca 8.67 2.70 1.52 14.12 2.07 1.81 K 1.80 0.68 3.76 1.55 1.63 0.29 S 0.81 0.78 0.10 2.37 2.03 3.59 Si 4.56 4.76 27.22 14.30 16.10 5.81 Al 0.80 1.15 13.77 3.85 9.70 1.71 Mg 2.12 0.13 0.91 0.21 0.62 0.07 Zn 3.09 0.06 0.02 0.14 0.12 1.00

[0045] 3. Determine the mass fraction of metal elements in the particulate sample. First, determine the water-soluble metal elements in the particulate sample on the polypropylene film. Cut half of the polypropylene film, cut it into pieces and place it in a stoppered tube, and add 50 ml of deionized water; then perform ultrasonic extraction and filtration to make up the volume, and use an ion chromatograph to determine the water-soluble metal ions; secondly, determine the content of water-insoluble metal elements on the polypropylene film. Use the remaining half of the polypropylene film to press the sample into a sheet, and use an X-ray fluorescence spectrometer to determine the content of metal elements contained in the air particles on the polypropylene film; the mass of the particulate matter on the entire filter membrane is the weight difference obtained by weighing the filter membrane before and after collecting the particulate matter, which is 0.01763 g. The mass content (%) of the metal elements in the collected particulate sample is the mass of the corresponding metal measured by the experimental analysis divided by the mass of the particulate sample on the corresponding filter membrane, as shown in Table 2.

[0046] Table 2 Content of metal elements in suspended particulate matter samples

[0047] element y(％) element y(％) Pb 0.2705 V 0.0081 Cd 0.0135 Ti 0.03 Se 0.04 Ca 4.8049 As 0.2394 K 3.3472 Cu 0.0784 S 0.0037 Ni 0.0773 Si 9.0102 Co 0.0094 Al 1.221 Fe 1.2382 Mg 0.3996 Mn 0.0469 Zn 1.2328 Cr 0.038 - -

[0048] The metal mass content and emission source component spectrum in the samples measured experimentally were standardized using the maximum and minimum standardization method. The results are shown in Table 3.

[0049] Table 3 Data used for PPR modeling after standardization

[0050]

[0051]

[0052] 4. Establish a PPR-based metal element source analysis model in the air pollution process. Substitute the standardized component spectrum (x) and sample data (y) in Table 3 into formula (2), take the number of ridge functions M as 2, and establish a PPR air suspended pollution particulate matter metal element source analysis model for 6 emission sources and 19 metal elements, as shown below:

[0053]

[0054] The parameter α in the formula mj and β m (m=1,2;j=1,2,…,6) is determined by optimizing using simulated annealing algorithm.

[0055] 5. Based on the simulated annealing algorithm, the parameters in the PPR model are optimized to obtain the optimal combination of parameters, and the contribution rate of the particulate matter emission source to the metal elements at the sampling point is calculated. Under the condition of satisfying the optimization target criterion formula (4), the simulated annealing algorithm is used to repeatedly iteratively optimize the parameters α and β in formula (5). When the iterative optimization satisfies the minimum target value, the iteration stops and the optimized parameter matrix is ​​obtained. The analysis results show that the contribution of the six emission source categories to the metal elements in the air suspended particulate matter at the sampling points from 8:00 on December 31, 2015 to 8:00 on January 1, 2016 are metallurgical dust 18.22%, comprehensive dust 27.27%, soil dust 7.06%, construction dust 10.3%, coal smoke dust 18.43% and traffic dust 18.72%, among which the important The contributing dust sources are comprehensive dust, traffic dust and coal smoke dust, accounting for 64.42% of the total emission sources; among these six pollution emission sources, the one that contributes the most to the metal content in the air suspended particulate matter at the sampling point is urban comprehensive dust. The contribution of this emission source usually comes from the uncertain comprehensive emission of waste generated in people's daily lives, such as certain emissions: plant (straw) burning, urban garbage and urban dust, etc. The proportion of the first two emissions from these pollution sources is relatively high, which has the greatest impact on the content of metal elements in suspended particulate matter. The potential source contribution factor analysis method (PSCF) combined with remote sensing images is one of the methods to verify whether the interpretation of source analysis results is reasonable. The PSCF method identifies the location of possible pollution sources based on backward trajectories, and remote sensing images provide spatial distribution information of pollution sources. This method is used to verify that the source analysis results of this embodiment are reliable and accurate.

[0056] Comparative Example:

[0057] Taking a sampling point in Shuangliu District, Chengdu City, Sichuan Province on December 31, 2015 as an example, the specific steps of the source analysis method of metal elements in suspended air particles at this sampling point are as follows:

[0058] Steps 1-3 are the same as in the embodiment.

[0059] 4. Establish a metal element source analysis model in the air pollution process based on the least squares method. The least squares method is a technique for solving a system of linear equations by minimizing the sum of squares of errors. It can be used to analyze the composition of metal elements in the sample in this example. In the given linear equation system A·x=y, where A is the metal element matrix of the composition spectrum, y is the sample metal element composition vector, and x is the dust source composition that needs to be solved; the goal of the least squares method is to find a vector x such that ||Ax-y|| 2 minimize.

[0060] 5. Based on the least squares method, the source analysis model of metal elements in the air pollution process, the contributions of metallurgical dust, comprehensive dust, soil dust, construction dust, coal smoke and traffic dust in the sample particulate matter are 19.92%, -140.75%, 386.36%, 123.78%, -337.95% and 65.65% respectively; the reason for the negative scores is that the least squares method does not perform ridge regression regularization. Ridge regression can introduce constraints in the least squares method, increase the stability of the solution, and prevent overfitting (negative scores) in the regression analysis.

Claims

1. A method for source analysis of atmospheric particulate matter, characterized in that: The following steps are involved: S1. Sample collection: Collect air particulate matter samples from sampling points in areas where metal element sources in polluted particulate matter need to be analyzed, and record the location and time of sample collection; S2. Construct emission source composition spectrum: Take the sampling point as the starting point of the backward trajectory, use the backward trajectory method to calculate the backward trajectory of the air mass located at the sampling point within n hours before the start of sampling, calculate a backward trajectory every hour, and take the maximum straight-line distance from the starting point of the trajectory to the sampling point in the n air mass backward trajectories as the effective distance D; in the actual area of ​​composition spectrum sampling, take the sampling point as the center, and take the shortest distance from the sampling point to the boundary of the composition spectrum sampling area as the maximum effective distance D max , D≤D max ; The composition spectrum collects the corresponding atmospheric particulate matter emission source information and determines the mass fraction of metal elements in the atmospheric particulate matter emitted by the emission source, thereby constructing the metal element composition spectrum emitted by potential emission sources within the effective distance of the sampling point; S3, data statistics: experimentally analyze the particulate matter samples collected in step S1 to obtain the mass fraction of metal elements in the particulate matter samples, and standardize the obtained mass fraction of metal elements. The standardization method adopts the maximum and minimum standardization method, and the processing formula is as follows: Where x is the original data of the mass fraction of the metal element before processing, and x' is the standardized data; S4. Establishing a metal element source analysis model: using a set of ridge functions G m (Z m ) and to approximate the regression function f(x), and establish a source analysis model of metal elements in the air pollution process based on PPR. The formula of f(x) is shown in formula (1) and formula (2). Formula (2) is the matrix form of the PPR model. Where M is the number of ridge functions, β m is the mth ridge function G m Contribution weight to the regression function f(x); Z m As the independent variable of the ridge function, it is a P-dimensional vector exist The projection in the direction and satisfy Where j is the dimension of the P-dimensional vector, and its formula is shown in formula (3): S5. Use the PPR model to analyze the source of air particles: S51. The mass fraction of metals in emission source j in the air particulate matter emission source composition spectrum is regarded as the factor x for PPR modeling. j ; S52, the mass fraction of the metal element in the particulate matter sample obtained by the analysis in step S3 is regarded as the dependent variable y, and the mass fraction of the metal element x in the emission source composition spectrum in step S2 is regarded as the dependent variable y. j One-to-one correspondence with y; S53, perform data generation processing, optimize parameter matrix, and obtain factor x j The weighted contribution rate to the dependent variable y is the contribution rate of each particulate matter emission source to the mass fraction of metal elements at the sampling point.

2. The method for source analysis of atmospheric particulate matter according to claim 1, characterized in that: In step S53, the simulated annealing algorithm is used to optimize the parameters in the PPR model, that is, the PPR uses the least square method as the minimization criterion, by selecting β in formula (2) m , G m (Z m ) and the number of ridge functions M, and design the objective function L2 so that the PPR model satisfies the criterion of formula (4). The simulated annealing algorithm is used to optimize the parameter matrix (β1, β2, ..., β M )and Get factor x j The weighted contribution rate of the dependent variable y is the contribution rate of each particulate matter emission source to the mass fraction of metal elements at the sampling point, where E is the sum, It is a prediction of the mass fraction y of metal elements in the particulate matter sample.

3. A method for source analysis of atmospheric particulate matter according to claim 2, characterized in that: The simulated annealing algorithm is implemented by programming in Matlab language.

4. The method for source analysis of atmospheric particulate matter according to claim 2, characterized in that: The parameters of the simulated annealing algorithm when optimizing the PPR model are set as follows: Population size 200; annealing form: T(t+1)=γT(t), γ=0.9, t is the number of iterations; initial temperature T0=1×1010, termination temperature T f =0; Acceptance probability formula: exp(Δf / T)>rand, where rand is a random number between (0, 1); The simulated annealing algorithm is used to perform iterative optimization repeatedly. f = 0, the iteration ends and the optimized parameter matrix is ​​obtained. (β1, β2,..., β M ) and The contribution rates of the air particulate matter emission sources (j=1, 2, …, P) to the mass fraction of metal elements in the samples at the sampling points can be obtained as follows:

5. The method for source analysis of atmospheric particulate matter according to claim 1, characterized in that: In the step S1, a polypropylene membrane is used for sample collection, and a medium-flow atmospheric sampler is used to continuously sample at a rate of 100 liters / minute for 24 hours; at the same time, a blank sample is collected, and the blank sample is obtained when the sampler is loaded with a filter membrane and is not turned on for 24 hours.

6. A method for source analysis of atmospheric particulate matter according to claim 5, characterized in that: The polypropylene film was placed in a constant temperature and humidity chamber (20±1° C.) and (50±5)% for 24 hours before and after sampling, and then weighed with a 100,000th balance. The mass difference of the polypropylene film before and after sampling was taken as the mass of the air particle sample.

7. A method for source analysis of atmospheric particulate matter according to claim 6, characterized in that: In addition to the constant temperature and humidity treatment before and after weighing, the samples were stored in a freezer until laboratory analysis, and the storage temperature was set at -4°C.

8. The method for source analysis of atmospheric particulate matter according to claim 5, characterized in that: The S3 data statistics include statistics of water-soluble and water-insoluble metal elements in the particulate matter. The water-soluble metal elements on the polypropylene film are measured by ion chromatography; the water-insoluble metal elements on the polypropylene film are measured by X-ray fluorescence spectrometer.

9. The method for source analysis of atmospheric particulate matter according to claim 1, characterized in that: The M is the number of ridge functions, and its value ranges from 2 to 5.

Images