Method and apparatus for fine particulate pollution source apportionment based on mass concentration across multiple particle size ranges

Abstract

This invention discloses a method and apparatus for apportioning fine particulate pollution sources based on multi-size range mass concentrations, relating to the field of particulate matter analysis technology and applicable to environmental monitoring. Its main purpose is to address the problem of low efficiency in existing fine particulate pollution source apportionment methods. The method primarily includes: constructing corresponding target multiple linear regression models based on historical concentration contribution values ​​and historical multi-size range mass concentrations for various pollution sources within the same historical period; obtaining real-time multi-size range mass concentrations for each type of pollution source in response to a fine particulate pollution source apportionment command; predicting the real-time multi-size range mass concentrations for each type of pollution source using the target multiple linear regression models to obtain predicted concentration contribution values ​​for each type of pollution source; and calculating the fine particulate pollution source apportionment results based on the predicted concentration contribution values ​​for each type of pollution source. This method is mainly used for apportioning fine particulate pollution sources.

Description

Technical Field

[0001] This invention relates to the field of particulate matter analysis technology, and in particular to a method and apparatus for source apportionment of fine particulate pollution based on mass concentration of multiple particle size ranges. Background Technology

[0002] With rapid industrialization and urbanization, fine particulate matter (PM2.5) pollution has harmed air quality, human health, and the ecological environment. Accurately identifying the sources of fine particulate matter pollution and clarifying the contribution of various pollution sources to the concentration of fine particulate matter is a key prerequisite for formulating effective pollution control strategies and improving air quality.

[0003] The most common method for apportioning fine particulate pollution sources is the source modeling approach. This method establishes complex physicochemical models to simulate the emission, diffusion, and transformation of pollutants, thereby determining the contribution of different pollution sources. However, this method requires a large amount of basic data, including detailed emission inventories, meteorological data, and topographic data. In practice, obtaining this accurate and comprehensive data often faces numerous difficulties, and data uncertainty can significantly affect the simulation accuracy of the model. Moreover, the source modeling method is extremely complex, involving a large amount of numerical calculation and iterative computation, which places high demands on computer hardware performance and is computationally time-consuming, resulting in low apportionment efficiency. Summary of the Invention

[0004] In view of this, the present invention provides a method and apparatus for apportionment of fine particulate pollution sources based on mass concentration of multiple particle size ranges, the main purpose of which is to solve the problem of low apportionment efficiency of existing fine particulate pollution sources.

[0005] According to one aspect of the present invention, a method for apportioning fine particulate pollution sources based on mass concentration across multiple particle size ranges is provided, comprising:

[0006] Based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, respective target multiple linear regression models are constructed.

[0007] In response to fine particulate pollution source apportionment commands, real-time multi-particle size range mass concentrations corresponding to various pollution sources are obtained;

[0008] For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-particle size range mass concentration to obtain the concentration contribution prediction value of each type of pollution source.

[0009] Based on the predicted concentration contribution values ​​of the various pollution sources, the analysis results of fine particulate pollution sources are calculated.

[0010] Further, the historical concentration contribution value includes the first historical concentration contribution value within the first historical period, and the historical multi-particle size range mass concentration includes the first historical multi-particle size range mass concentration within the first historical period. For any type of pollution source, a target multiple linear regression model is constructed based on the historical concentration contribution value and historical multi-particle size range mass concentration of the pollution source within the same historical period, including:

[0011] The first historical concentration contribution value is used as the target output, and the first historical multi-particle size segment mass concentration is used as the independent variable to construct an initial multiple linear regression model, wherein the first historical multi-particle size segment mass concentration includes the mass concentration distribution of the pollution source in a preset number of particle size channels.

[0012] The mass concentration distribution weights of the pollution source in the particle size channel are obtained by solving the initial multiple linear regression model.

[0013] The weight distribution of the pollution source in each particle size channel is iteratively optimized based on a preset error threshold, and a target multiple linear regression model is constructed based on the optimized weight distribution.

[0014] Further, the step of obtaining the mass concentration distribution weights of the pollution source in the particle size channel based on the initial multiple linear regression model includes:

[0015] The multiple linear regression equation of the initial multiple linear regression model is solved using the least squares method to obtain the weight distribution of the pollution source in each particle size channel. The multiple linear regression equation is expressed as follows:

[0016] ;

[0017] in, This represents the concentration contribution value of the kth type of pollution source; the preset number of particle size channels is 11. - These represent the mass concentrations of the 1st to 11th particle size channels, respectively. This indicates the weight of particles whose size does not fall within the range of particle size channels 1 to 11. - This represents the mass concentration distribution weight corresponding to each of the 1st to 11th particle size channels.

[0018] Further, the historical concentration contribution value includes the second historical concentration contribution value within the second historical period, and the historical multi-particle size segment mass concentration includes the second historical multi-particle size segment mass concentration within the second historical period. The iterative optimization of the weight distribution of the pollution source in each particle size channel based on a preset error threshold, and the construction of a target multiple linear regression model based on the optimized weight distribution, includes:

[0019] Obtain the mass concentration of the second historical multi-particle size segment within the second historical period, substitute the mass concentration of the second historical multi-particle size segment into the initial multiple linear regression equation, and calculate the predicted concentration contribution values ​​of various pollution sources in the second historical period.

[0020] The model error value is obtained by calculating the error between the predicted concentration contribution values ​​of various pollution sources and the second historical concentration contribution values.

[0021] If the model error value is less than or equal to a preset error threshold, the initial multiple linear regression model is used as the target multiple linear regression model.

[0022] If the model error value is greater than the preset error threshold, the initial multiple linear regression model is iteratively optimized until the model error value of the iteratively optimized multiple linear regression model is less than or equal to the preset error threshold, and the iteratively optimized multiple linear regression model is taken as the target multiple linear regression model.

[0023] Furthermore, the historical concentration contribution value is obtained based on the analysis of historical pollution source chemical composition data. Before constructing the corresponding target multiple linear regression model based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, the method further includes:

[0024] Acquire historical pollution source chemical composition data within a historical period, wherein the historical pollution source chemical composition data is collected by super monitoring stations according to a preset period;

[0025] By using a positive matrix factorization model to analyze the chemical composition data of historical pollution sources, the historical concentration contribution values ​​of various pollution sources are obtained.

[0026] Furthermore, after predicting the real-time multi-particle size range mass concentration using the target multiple linear regression model for each type of pollution source to obtain the predicted concentration contribution values ​​of each type of pollution source, the method further includes:

[0027] In response to the triggering of a preset time interval, acquire a set of historical pollution source chemical component data with the shortest time interval between the acquisition time point and the real-time multi-particle size mass concentration;

[0028] The historical pollution source chemical composition data were analyzed to obtain the reference concentration contribution values ​​of various pollution sources;

[0029] For each type of pollution source, the root mean square error between the reference concentration contribution value and the predicted concentration contribution value is calculated to obtain the model error value of the target multiple linear regression model corresponding to each type of pollution source.

[0030] If the model error value is greater than the preset error threshold, the model parameters of the target multiple linear regression model are updated according to the reference concentration contribution value, so as to perform subsequent pollution source real-time concentration contribution value prediction operation based on the updated target multiple linear regression model.

[0031] Furthermore, the calculation of the fine particulate pollution source apportionment results based on the concentration contribution prediction values ​​of the various pollution sources includes:

[0032] For each type of pollution source, the ratio of the predicted concentration contribution of that pollution source to the sum of the predicted concentration contributions of all pollution sources is calculated to obtain the contribution ratio of each type of pollution source.

[0033] Based on the aforementioned contribution ratios and the predicted concentration contribution values ​​of various pollution sources, the analysis results of fine particulate pollution sources are generated.

[0034] According to another aspect of the present invention, a fine particulate pollution source apportionment device based on mass concentration of multiple particle size ranges is provided, comprising:

[0035] The module is used to construct the corresponding target multiple linear regression model based on the historical concentration contribution value and historical multi-particle size mass concentration of various pollution sources within the same historical period.

[0036] The acquisition module is used to obtain the real-time multi-particle size mass concentration corresponding to various pollution sources in response to fine particulate pollution source analysis instructions;

[0037] The prediction module is used to predict the real-time multi-particle size mass concentration for each type of pollution source using the target multiple linear regression model, so as to obtain the concentration contribution prediction value of each type of pollution source.

[0038] The calculation module is used to calculate the fine particulate pollution source analysis results based on the concentration contribution prediction values ​​of the various pollution sources.

[0039] According to another aspect of the present invention, a storage medium is provided, wherein at least one executable instruction is stored therein, the executable instruction causing a processor to perform operations corresponding to the above-described method for analyzing fine particulate pollution sources based on mass concentration of multiple particle sizes.

[0040] According to another aspect of the present invention, a terminal is provided, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0041] The memory is used to store at least one executable instruction, which causes the processor to perform the operation corresponding to the above-described fine particulate pollution source analysis method based on the mass concentration of multiple particle sizes.

[0042] By employing the above-described technical solutions, the technical solutions provided by the embodiments of the present invention have at least the following advantages:

[0043] This invention provides a method and apparatus for apportioning fine particulate pollution sources based on multi-size segment mass concentration. In this embodiment, a target multiple linear regression model is constructed based on the historical concentration contribution values ​​and historical multi-size segment mass concentrations of various pollution sources within the same historical period. Responding to a fine particulate pollution source apportionment command, the real-time multi-size segment mass concentrations corresponding to various pollution sources are obtained. For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-size segment mass concentration, obtaining the predicted concentration contribution value of each pollution source. Based on the predicted concentration contribution values ​​of each pollution source, the fine particulate pollution source apportionment result is calculated. By predicting the real-time multi-size segment mass concentration using the target multiple linear regression model, the predicted concentration contribution values ​​of various pollution sources can be obtained quickly, greatly reducing the time for data collection and analysis. Simultaneously, the target multiple linear regression model can accurately capture the intrinsic correlation between each pollution source and the multi-size segment mass concentration, ensuring the accuracy of the prediction results, thereby significantly improving the apportionment efficiency of fine particulate pollution sources.

[0044] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the contents of the specification. Furthermore, in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0045] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0046] Figure 1 This invention provides a flowchart of a fine particulate pollution source apportionment method based on mass concentration across multiple particle size ranges, according to an embodiment of the present invention.

[0047] Figure 2 This invention provides a flowchart illustrating the construction of a target multiple linear regression model for historical concentration contribution values ​​and historical multi-particle size range mass concentrations of pollution sources, as provided in an embodiment of the invention.

[0048] Figure 3 This diagram illustrates a block diagram of a fine particulate pollution source apportionment device based on mass concentration across multiple particle size ranges, as provided in an embodiment of the present invention.

[0049] Figure 4A schematic diagram of the structure of a terminal provided in an embodiment of the present invention is shown. Detailed Implementation

[0050] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0051] To address the problem of low efficiency in existing fine particulate pollution source apportionment methods, this invention provides a fine particulate pollution source apportionment method based on mass concentration across multiple particle size ranges, such as... Figure 1 As shown, the method includes:

[0052] 101. Based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, construct their respective target multiple linear regression models.

[0053] Traditional PM2.5 pollution source apportionment mainly relies on high-precision chemical composition data obtained from super monitoring stations (also known as superstations) and uses positive matrix factorization (PMF) for analysis. However, the construction and operation and maintenance costs of superstations are high, making it difficult to achieve large-scale deployment, which limits the coverage and timeliness of pollution source apportionment.

[0054] In this embodiment of the invention, chemical component data (used to calculate concentration contribution values) collected from super-monitoring stations and multi-particle size range mass concentration data collected from particle size monitoring equipment are collected within the coverage area of ​​the super-monitoring station. Specifically, this includes historical concentration contribution values ​​and historical multi-particle size range mass concentration data for various pollution sources within the same historical period. Since the historical concentration contribution values ​​obtained from chemical component data analysis are close to the actual values ​​and have high accuracy, while the multi-particle size range mass concentration data comes from existing particulate matter size monitoring equipment and is typically used for rough analysis of particle size distribution, it has good real-time performance. Therefore, a multiple linear regression analysis method is used to deeply explore the intrinsic relationship between historical concentration contribution values ​​and historical multi-particle size range mass concentration data, constructing a dedicated target multiple linear regression model for each type of pollution source. This allows pollution source analysis to break free from excessive reliance on the high cost and long-term data from super-monitoring stations. Based on existing particle size monitoring equipment resources and real-time collected multi-particle size range mass concentration data, the analysis of pollution source concentration contribution values ​​is achieved, providing a reliable analytical tool for subsequent accurate and efficient pollution source analysis.

[0055] 102. In response to the fine particulate pollution source analysis command, obtain the real-time multi-particle size mass concentration corresponding to various pollution sources.

[0056] Upon receiving the instruction, this invention responds rapidly, utilizing existing particulate matter size monitoring equipment to acquire real-time multi-size particle concentrations corresponding to various pollution sources. These monitoring devices can accurately measure the mass concentration of fine particulate matter in different size ranges in real time and transmit the data promptly to the data processing system. Compared to relying on superstations for data acquisition, this process is more efficient and convenient. Its technical advantage lies in its ability to promptly capture dynamic changes in pollution sources in the environment, obtaining the latest data reflecting the current environmental situation. This provides timely and accurate information support for subsequent pollution source analysis, making the analysis results closer to the actual situation and facilitating the timely implementation of effective pollution control measures to protect environmental quality and public health.

[0057] 103. For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-particle size range mass concentration to obtain the concentration contribution prediction value of each type of pollution source.

[0058] For each type of pollution source, this invention utilizes the target multiple linear regression model constructed in step 101 to predict the real-time multi-particle size range mass concentration obtained in step 102. During the prediction process, the real-time multi-particle size range mass concentration is substituted as an input variable into the target multiple linear regression model. Based on a pre-established functional relationship between the pollution source contribution rate and particle size distribution, the model quickly calculates the predicted concentration contribution values ​​of various pollution sources. This technique is highly effective, avoiding the complex on-site monitoring and experimental analysis processes of traditional methods, eliminating the need to wait for data from super-stations, and significantly improving the efficiency of pollution source analysis. Through model prediction, the predicted contribution values ​​of various pollution sources to PM2.5 concentration at the current moment can be obtained in a short time, providing strong support for real-time monitoring of pollution source dynamics and enabling environmental management departments to make timely decisions and take corresponding control measures.

[0059] 104. Based on the predicted concentration contribution values ​​of the various pollution sources, the analysis results of fine particulate pollution sources are calculated.

[0060] After obtaining the predicted concentration contributions of various pollution sources, this invention performs comprehensive calculations based on these predictions to obtain the source apportionment results for fine particulate pollution. By summarizing, analyzing, and comparing the predicted concentration contributions of various pollution sources, the relative contributions of different pollution sources to PM2.5 pollution can be clearly determined, thereby identifying the main pollution sources.

[0061] In one embodiment of the present invention, for further illustration and limitation, such as Figure 2 As shown, for any type of pollution source, a target multiple linear regression model is constructed based on the historical concentration contribution value and historical multi-particle size range mass concentration of the pollution source within the same historical period, including:

[0062] 201. Using the contribution value of the first historical concentration as the target output, and using the mass concentration of the first historical multi-particle size segment as the independent variable, construct an initial multiple linear regression model.

[0063] 202. The mass concentration distribution weights of the pollution source in the particle size channel are obtained by solving the initial multiple linear regression model.

[0064] 203. Based on a preset error threshold, the weight distribution of the pollution source in each particle size channel is iteratively optimized, and a target multiple linear regression model is constructed based on the optimized weight distribution.

[0065] In this embodiment of the invention, concentration analysis and prediction for any type of pollution source are based on historical data-driven multiple linear regression modeling. First, key data within the first historical period are focused on, using the historical concentration contribution value of this pollution source during this period as the target output variable for model fitting. Simultaneously, the mass concentration distribution of the pollution source in a preset number of particle size channels within the same period is selected as the independent variable, constructing an initial multiple linear regression framework. The model is solved using mathematical tools such as the least squares method to initially quantify the weight coefficients of each particle size channel's contribution to the total concentration, revealing the original correlation between different particle size segments and pollution concentration. Then, a preset error threshold is introduced as the optimization objective, and algorithms such as gradient descent or iterative reweighting are used to finely adjust the initial weight distribution. During training, the model complexity and prediction accuracy are dynamically balanced until the model's error on the validation set stably converges to within the threshold range; the final target multiple linear regression model is formed. This model can not only accurately characterize the changes in pollution source concentration with particle size distribution but also identify key particle size control intervals through weight coefficient sorting. The historical concentration contribution value includes the first historical concentration contribution value within the first historical period. The historical multi-size particle size range mass concentration includes the first historical multi-size particle size range mass concentration within the first historical period. The first historical multi-size particle size range mass concentration includes the mass concentration distribution of the pollution source across a preset number of particle size channels. The preset number of particle size channels can be 11, for example, channel 1: 0.3-0.5 μm, channel 2: 0.5-1 μm, channel 11: 5-10 μm.

[0066] Through a collaborative optimization mechanism of data models, the spatiotemporal adaptability and interpretability of pollution source apportionment are significantly improved, providing a quantitative decision-making tool for scenarios such as precise source tracing of air pollution and graded control of particulate matter. The first historical period can be any historical time period, and the length of the time period can be a single super-station data collection cycle, or it can be customized; this embodiment of the invention does not impose specific limitations.

[0067] In one embodiment of the present invention, for further explanation and limitation, the step of obtaining the mass concentration distribution weights of the pollution source in the particle size channel based on the initial multiple linear regression model includes:

[0068] The weight distribution of the pollution source in each particle size channel is obtained by solving the multiple linear regression equation of the initial multiple linear regression model using the least squares method.

[0069] In this embodiment of the invention, the weights of the mass concentration distribution of the pollution source in the particle size channel are determined for the initial multiple linear regression model, specifically using the least squares method for parameter estimation. The multiple linear regression equation is expressed as:

[0070] ;

[0071] in, This represents the concentration contribution value of the kth type of pollution source; the preset number of particle size channels is 11. - These represent the mass concentrations of the 1st to 11th particle size channels, respectively. This indicates the weight of particles whose size does not fall within the range of particle size channels 1 to 11. - This represents the mass concentration distribution weights corresponding to each of the 1st to 11th particle size channels. Based on the principle of least squares, the weight coefficients are solved by minimizing the sum of squared residuals, resulting in a set of normal equations concerning the weight coefficients. Then, matrix operations or numerical calculation methods are used to solve this set of normal equations, thus obtaining the regression coefficient matrix, which represents the weight distribution of the pollution source in each particle size channel. - These weights reflect the relative influence of the mass concentration of different particle size channels on the historical concentration of pollution sources. By solving for the weights using the least squares method, the contribution weight of each particle size channel can be objectively and accurately determined based on historical data, providing a reliable parameter basis for subsequent model optimization and pollution source analysis.

[0072] In one embodiment of the present invention, for further explanation and limitation, the step of iteratively optimizing the weight distribution of the pollution source in each particle size channel according to a preset error threshold, and constructing a target multiple linear regression model based on the optimized weight distribution, includes:

[0073] Obtain the mass concentration of the second historical multi-particle size segment within the second historical period, substitute the mass concentration of the second historical multi-particle size segment into the initial multiple linear regression equation, and calculate the predicted concentration contribution values ​​of various pollution sources in the second historical period.

[0074] The model error value is obtained by calculating the error between the predicted concentration contribution values ​​of various pollution sources and the second historical concentration contribution values.

[0075] If the model error value is less than or equal to a preset error threshold, the initial multiple linear regression model is used as the target multiple linear regression model.

[0076] If the model error value is greater than the preset error threshold, the initial multiple linear regression model is iteratively optimized until the model error value of the iteratively optimized multiple linear regression model is less than or equal to the preset error threshold, and the iteratively optimized multiple linear regression model is taken as the target multiple linear regression model.

[0077] In this embodiment of the invention, the historical concentration contribution value includes the second historical concentration contribution value within the second historical period. The historical multi-particle size segment mass concentration includes the second historical multi-particle size segment mass concentration within the second historical period. That is, the target multiple linear regression model is trained based on multiple sets of historical concentration contribution values ​​and historical multi-particle size segment mass concentrations from different historical periods. After constructing the initial multiple linear regression model based on the first historical concentration contribution value within the first historical period, monitoring data within a new time period is acquired, i.e., mass concentration monitoring data of different particle size channels within the second historical period. These actual monitoring data are substituted into the initial multiple linear regression equation, and the concentration contribution prediction values ​​of various pollution sources in this historical period are obtained through equation calculation. Then, the error is calculated based on these prediction values ​​and the actual hourly concentration contribution values ​​of various pollution sources in the second historical period to obtain the model error value. If the model error value is less than or equal to the preset error threshold, it indicates that the accuracy of the current initial multiple linear regression model meets the standard and can be used as the target multiple linear regression model. If the model error value is greater than the preset error threshold, it indicates that the model accuracy is insufficient, and the initial multiple linear regression model needs to be iteratively optimized. This process is repeated until the model error value after iterative optimization is less than or equal to the preset error threshold. At this point, the iteratively optimized model is used as the target multiple linear regression model. This iterative optimization method can improve the accuracy of the model's prediction of the contribution of pollution source concentrations. The model error value can be the root mean square error between the predicted contribution values ​​of various pollution source concentrations and the second historical concentration contribution values. The preset error threshold can be 15%, or it can be customized according to actual needs; this embodiment of the invention does not impose specific limitations.

[0078] In one embodiment of the present invention, for further explanation and limitation, before constructing the corresponding target multiple linear regression model based on the historical concentration contribution values ​​and historical multi-particle size range mass concentrations of various pollution sources within the same historical period, the method further includes:

[0079] Obtain historical pollution source chemical composition data within a historical period;

[0080] By using a positive matrix factorization model to analyze the chemical composition data of historical pollution sources, the historical concentration contribution values ​​of various pollution sources are obtained.

[0081] In this embodiment of the invention, historical pollution source chemical composition data are collected from super monitoring stations according to a preset cycle. Historical concentration contribution values ​​are obtained by analyzing the historical pollution source chemical composition data. Before constructing the target multiple linear regression model, historical pollution source chemical composition data within the historical cycle are collected from the super monitoring stations according to a preset cycle (e.g., daily, weekly). This data covers various chemical components emitted by different pollution sources. Then, the collected historical pollution source chemical composition data is analyzed using positive matrix factorization, a factor-based environmental data analysis technique mainly used for source apportionment of complex pollution sources such as atmospheric particulate matter (e.g., PM2.5, PM10). Its core principle is to decompose a non-negative concentration matrix into two non-negative matrices: a source composition matrix and a source contribution matrix. By minimizing the residual of the objective function, the historical concentration contribution values ​​of various pollution sources are obtained. The data collected by the super monitoring station is comprehensive and accurate, providing a reliable foundation for subsequent analysis. The positive matrix factorization model can accurately extract the concentration contribution information of various pollution sources from complex chemical composition data, making the obtained concentration contribution values ​​more accurately reflect the actual situation. This provides high-quality input data for constructing the target multiple linear regression model, which helps to improve the prediction accuracy and reliability of the final model.

[0082] In one application example, a regression model was trained using super-station component data and particle size monitoring data for a certain month in 2024. A regression equation was established between particle size and component data. The contribution percentages of various sources calculated using the regression coefficients deviated from those obtained by the PMF method by only 0.1% to 4.1%, showing a high degree of consistency. Using the regression equation to extrapolate the contribution percentages of the five pollution sources at the same station for the following month, the deviation from the PMF method was found to be between 0.1% and 5.6%, with overall results close and an error of <6%, indicating that the particle size source apportionment results extrapolated by multiple linear regression are reliable.

[0083] In one embodiment of the present invention, for further explanation and limitation, after predicting the real-time multi-particle size range mass concentration using the target multiple linear regression model for each type of pollution source to obtain the predicted concentration contribution values ​​of each type of pollution source, the method further includes:

[0084] In response to the triggering of a preset time interval, acquire a set of historical pollution source chemical component data with the shortest time interval between the acquisition time point and the real-time multi-particle size mass concentration;

[0085] The historical pollution source chemical composition data were analyzed to obtain the reference concentration contribution values ​​of various pollution sources;

[0086] For each type of pollution source, the root mean square error between the reference concentration contribution value and the predicted concentration contribution value is calculated to obtain the model error value of the target multiple linear regression model corresponding to each type of pollution source.

[0087] If the model error value is greater than the preset error threshold, the model parameters of the target multiple linear regression model are updated according to the reference concentration contribution value, so as to perform subsequent pollution source real-time concentration contribution value prediction operation based on the updated target multiple linear regression model.

[0088] In this embodiment of the invention, after obtaining the target multiple linear regression model and using it to predict the real-time multi-particle size range mass concentration, to ensure the long-term stability and accuracy of the model, it is necessary to periodically (at a preset time interval) verify the accuracy of the model using the concentration contribution values ​​of various pollution sources obtained from positive matrix decomposition as reference standard values, so as to correct the model parameters in a timely manner and maintain the continuous stability of the model. The preset time interval can be customized according to actual needs. Since the collection of pollution source chemical component data has a certain interval period, when selecting the historical pollution source chemical component data corresponding to the reference concentration contribution value, the data obtained from the collection process closest to the real-time multi-particle size range mass concentration collection time corresponding to the concentration contribution prediction value should be taken. That is, obtain a set of historical pollution source chemical component data with the shortest time interval between the collection time point and the real-time multi-particle size range mass concentration collection. These data are closest to the real-time multi-particle size range mass concentration collection in time and space, and can reflect the true situation of the pollution source at that time. The model verification process is consistent with the parameter optimization process from the initial multiple linear regression model to the target multiple linear regression model, and will not be described again here.

[0089] By periodically acquiring historical data close to the real-time data as a reference, changes in pollution sources can be reflected in a timely manner. The root mean square error calculation can accurately quantify the model prediction deviation. When the error exceeds the limit, the model parameters are updated in a timely manner, enabling the model to dynamically adapt to changes in pollution sources. This improves the accuracy of subsequent predictions and ensures that the prediction of the real-time concentration contribution value of pollution sources is more in line with reality, providing a more reliable basis for environmental management and pollution prevention and control.

[0090] In one embodiment of the present invention, for further explanation and limitation, the fine particulate pollution source apportionment results are calculated based on the concentration contribution prediction values ​​of the various pollution sources, including:

[0091] For each type of pollution source, the ratio of the predicted concentration contribution of that pollution source to the sum of the predicted concentration contributions of all pollution sources is calculated to obtain the contribution ratio of each type of pollution source.

[0092] Based on the aforementioned contribution ratios and the predicted concentration contribution values ​​of various pollution sources, the analysis results of fine particulate pollution sources are generated.

[0093] In this embodiment of the invention, after obtaining the predicted concentration contribution values ​​of various pollution sources, for each type of pollution source, the ratio of the predicted concentration contribution value of that pollution source to the sum of the predicted concentration contribution values ​​of all pollution sources is calculated. Specifically, the predicted concentration contribution values ​​of each pollution source type are summed to obtain the sum of the predicted concentration contribution values ​​of all pollution sources. Then, the predicted concentration contribution value of each pollution source type is divided by this sum, and the result is the contribution percentage of each type of pollution source. The formula for calculating the contribution percentage of any type of pollution source is as follows:

[0094] (k = 1, 2, ..., 5);

[0095] The concentration contribution value of any type of pollution source is expressed as: The contribution percentage is expressed as The contribution percentage visually represents the share of each pollution source in the overall pollution, while the concentration contribution prediction reflects the specific concentration contribution of each source. Calculating the contribution percentage allows for a quick and intuitive understanding of the relative importance of different pollution sources in fine particulate matter (PM2.5) pollution, helping to identify key areas for pollution control and prioritize measures against sources with large contribution percentages. Furthermore, combining the concentration contribution prediction with the calculation provides a more precise quantification of the specific impact of each pollution source on PM2.5 pollution, offering detailed and accurate data support for developing targeted pollution prevention and control strategies.

[0096] This invention provides a method for apportioning fine particulate pollution sources based on multi-size segment mass concentration. In this embodiment, a target multiple linear regression model is constructed based on the historical concentration contribution values ​​and historical multi-size segment mass concentrations of various pollution sources within the same historical period. Responding to a fine particulate pollution source apportionment command, the real-time multi-size segment mass concentrations corresponding to various pollution sources are obtained. For each type of pollution source, the real-time multi-size segment mass concentration is predicted using the target multiple linear regression model to obtain the predicted concentration contribution value of each pollution source. Based on the predicted concentration contribution values ​​of each pollution source, the fine particulate pollution source apportionment result is calculated. By predicting the real-time multi-size segment mass concentration using the target multiple linear regression model, the predicted concentration contribution values ​​of various pollution sources can be obtained quickly, significantly reducing the time for data collection and analysis. Simultaneously, the target multiple linear regression model can accurately capture the intrinsic correlation between each pollution source and the multi-size segment mass concentration, ensuring the accuracy of the prediction results and thus greatly improving the apportionment efficiency of fine particulate pollution sources.

[0097] Furthermore, as a response to the above Figure 1 The implementation of the method shown in this invention provides a fine particulate pollution source apportionment device based on the mass concentration of multiple particle size ranges, such as... Figure 3As shown, the device includes:

[0098] Module 31 is used to construct the corresponding target multiple linear regression model based on the historical concentration contribution value and historical multi-particle size mass concentration of various pollution sources within the same historical period.

[0099] The acquisition module 32 is used to obtain the real-time multi-particle size mass concentration corresponding to various pollution sources in response to the fine particulate pollution source analysis command;

[0100] The prediction module 33 is used to predict the real-time multi-particle size mass concentration for each type of pollution source using the target multiple linear regression model, so as to obtain the concentration contribution prediction value of each type of pollution source.

[0101] The calculation module 34 is used to calculate the fine particulate pollution source analysis results based on the concentration contribution prediction values ​​of the various pollution sources.

[0102] Furthermore, the construction module 31 includes:

[0103] The first construction unit is used to construct an initial multiple linear regression model with the first historical concentration contribution value as the target output and the first historical multi-particle size segment mass concentration as the independent variable, wherein the first historical multi-particle size segment mass concentration includes the mass concentration distribution of the pollution source in a preset number of particle size channels.

[0104] The solution unit is used to solve for the mass concentration distribution weights of the pollution source in the particle size channel based on the initial multiple linear regression model.

[0105] The optimization unit is used to iteratively optimize the weight distribution of the pollution source in each particle size channel according to a preset error threshold, and to construct a target multiple linear regression model based on the optimized weight distribution.

[0106] Furthermore, in specific application scenarios, the solving unit is specifically used to solve the multiple linear regression equation of the initial multiple linear regression model using the least squares method, to obtain the weight distribution of the pollution source in each particle size channel, wherein the multiple linear regression equation is expressed as:

[0107] ;

[0108] in, This represents the concentration contribution value of the kth type of pollution source; the preset number of particle size channels is 11. - These represent the mass concentrations of the 1st to 11th particle size channels, respectively. This indicates the weight of particles whose size does not fall within the range of particle size channels 1 to 11. - This represents the mass concentration distribution weight corresponding to each of the 1st to 11th particle size channels.

[0109] Furthermore, in a specific application scenario, the optimization unit is specifically used to obtain the mass concentration of the second historical multi-particle size segment within the second historical period, substitute the mass concentration of the second historical multi-particle size segment into the initial multiple linear regression equation, and calculate the predicted concentration contribution values ​​of various pollution sources in the second historical period.

[0110] The model error value is obtained by calculating the error between the predicted concentration contribution values ​​of various pollution sources and the second historical concentration contribution values.

[0111] If the model error value is less than or equal to a preset error threshold, the initial multiple linear regression model is used as the target multiple linear regression model.

[0112] If the model error value is greater than the preset error threshold, the initial multiple linear regression model is iteratively optimized until the model error value of the iteratively optimized multiple linear regression model is less than or equal to the preset error threshold, and the iteratively optimized multiple linear regression model is taken as the target multiple linear regression model.

[0113] Furthermore, the device also includes:

[0114] The acquisition module 32 is also used to acquire historical pollution source chemical composition data within a historical period, wherein the historical pollution source chemical composition data is collected by the super monitoring station according to a preset period.

[0115] The first analysis module is used to analyze the historical chemical composition data of pollution sources using a positive matrix factorization model to obtain the historical concentration contribution values ​​of various pollution sources.

[0116] Furthermore, the device also includes:

[0117] The acquisition module 32 is also used to acquire a set of historical pollution source chemical component data with the shortest acquisition time interval between the acquisition time point and the real-time multi-particle size mass concentration in response to the triggering of the preset time interval.

[0118] The second analysis module is used to analyze the historical pollution source chemical composition data to obtain the reference concentration contribution value of various pollution sources;

[0119] The calculation module 34 is also used to calculate the root mean square error between the reference concentration contribution value and the concentration contribution prediction value for each type of pollution source, so as to obtain the model error value of the target multiple linear regression model corresponding to each type of pollution source.

[0120] The update module is used to update the model parameters of the target multiple linear regression model based on the reference concentration contribution value if the model error value is greater than a preset error threshold, so as to perform subsequent pollution source real-time concentration contribution value prediction operations based on the updated target multiple linear regression model.

[0121] Furthermore, the computing module 34 includes:

[0122] The calculation unit is used to calculate the ratio of the predicted concentration contribution of each pollution source to the sum of the predicted concentration contribution of all pollution sources for each type of pollution source, so as to obtain the contribution ratio of each type of pollution source.

[0123] The generation unit is used to generate fine particulate pollution source analysis results based on the contribution ratio and the concentration contribution prediction values ​​of various pollution sources.

[0124] This invention provides a fine particulate pollution source apportionment device based on multi-size segment mass concentration. In this embodiment, a target multiple linear regression model is constructed based on the historical concentration contribution values ​​and historical multi-size segment mass concentrations of various pollution sources within the same historical period. Responding to a fine particulate pollution source apportionment command, the real-time multi-size segment mass concentrations corresponding to various pollution sources are obtained. For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-size segment mass concentration, obtaining the predicted concentration contribution value of each pollution source. Based on the predicted concentration contribution values ​​of each pollution source, the fine particulate pollution source apportionment result is calculated. By predicting the real-time multi-size segment mass concentration using the target multiple linear regression model, the predicted concentration contribution values ​​of various pollution sources can be obtained quickly, greatly reducing the time for data collection and analysis. Simultaneously, the target multiple linear regression model can accurately capture the intrinsic correlation between each pollution source and the multi-size segment mass concentration, ensuring the accuracy of the prediction results, thereby significantly improving the apportionment efficiency of fine particulate pollution sources.

[0125] According to one embodiment of the present invention, a storage medium is provided, the storage medium storing at least one executable instruction, the computer-executable instruction being able to execute the fine particulate pollution source apportionment method based on multi-size segment mass concentration in any of the above method embodiments.

[0126] Figure 4 The diagram shows a structural schematic of a terminal according to an embodiment of the present invention. The specific implementation of the present invention is not limited to the specific implementation of the terminal.

[0127] like Figure 4 As shown, the terminal may include: a processor 402, a communication interface 404, a memory 406, and a communication bus 408.

[0128] The processor 402, communication interface 404, and memory 406 communicate with each other via communication bus 408.

[0129] Communication interface 404 is used for network communication with other devices such as clients or other servers.

[0130] The processor 402 is used to execute program 410, which can specifically execute the relevant steps in the above embodiments of the fine particulate pollution source apportionment method based on the mass concentration of multiple particle size segments.

[0131] Specifically, program 410 may include program code that includes computer operation instructions.

[0132] Processor 402 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The terminal may include one or more processors of the same type, such as one or more CPUs; or it may include processors of different types, such as one or more CPUs and one or more ASICs.

[0133] Memory 406 is used to store program 410. Memory 406 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0134] Specifically, program 410 can be used to cause processor 402 to perform the following operations:

[0135] Based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, respective target multiple linear regression models are constructed.

[0136] In response to fine particulate pollution source apportionment commands, real-time multi-particle size range mass concentrations corresponding to various pollution sources are obtained;

[0137] For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-particle size range mass concentration to obtain the concentration contribution prediction value of each type of pollution source.

[0138] Based on the predicted concentration contribution values ​​of the various pollution sources, the analysis results of fine particulate pollution sources are calculated.

[0139] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those presented herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0140] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for apportioning fine particulate pollution sources based on mass concentration across multiple particle size ranges, characterized in that, include: Based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, respective target multiple linear regression models are constructed. In response to fine particulate pollution source apportionment commands, real-time multi-particle size range mass concentrations corresponding to various pollution sources are obtained; For each type of pollution source, the target multiple linear regression model is used to predict the real-time multi-particle size range mass concentration to obtain the concentration contribution prediction value of each type of pollution source. Based on the predicted concentration contributions of the various pollution sources, the apportionment results for fine particulate pollution sources are calculated, specifically including: For each type of pollution source, the ratio of the predicted concentration contribution of that pollution source to the sum of the predicted concentration contributions of all pollution sources is calculated to obtain the contribution ratio of each type of pollution source. Based on the aforementioned contribution ratios and the predicted concentration contribution values ​​of various pollution sources, the analysis results of fine particulate pollution sources are generated; Wherein, the historical concentration contribution value includes the first historical concentration contribution value within the first historical period, and the historical multi-particle size range mass concentration includes the first historical multi-particle size range mass concentration within the first historical period. For any type of pollution source, a target multiple linear regression model is constructed based on the historical concentration contribution value and historical multi-particle size range mass concentration of the pollution source within the same historical period, including: The first historical concentration contribution value is used as the target output, and the first historical multi-particle size segment mass concentration is used as the independent variable to construct an initial multiple linear regression model, wherein the first historical multi-particle size segment mass concentration includes the mass concentration distribution of the pollution source in a preset number of particle size channels. The mass concentration distribution weights of the pollution source in the particle size channel are obtained by solving the initial multiple linear regression model. The weight distribution of the pollution source in each particle size channel is iteratively optimized based on a preset error threshold, and a target multiple linear regression model is constructed based on the optimized weight distribution.

2. The method according to claim 1, characterized in that, The step of obtaining the mass concentration distribution weights of the pollution source in the particle size channel based on the initial multiple linear regression model includes: The multiple linear regression equation of the initial multiple linear regression model is solved using the least squares method to obtain the weight distribution of the pollution source in each particle size channel. The multiple linear regression equation is expressed as follows: ； in, This represents the concentration contribution value of the kth type of pollution source; the preset number of particle size channels is 11. - These represent the mass concentrations of the 1st to 11th particle size channels, respectively. This indicates the weight of particles whose size does not fall within the range of particle size channels 1 to 11. - This represents the mass concentration distribution weight corresponding to each of the 1st to 11th particle size channels.

3. The method according to claim 1, characterized in that, The historical concentration contribution value includes the second historical concentration contribution value within the second historical period, and the historical multi-particle size segment mass concentration includes the second historical multi-particle size segment mass concentration within the second historical period. The iterative optimization of the weight distribution of the pollution source in each particle size channel based on a preset error threshold, and the construction of a target multiple linear regression model based on the optimized weight distribution, includes: Obtain the mass concentration of the second historical multi-particle size segment within the second historical period, substitute the mass concentration of the second historical multi-particle size segment into the initial multiple linear regression equation, and calculate the predicted concentration contribution value of various pollution sources within the second historical period. The model error value is obtained by calculating the error between the predicted concentration contribution values ​​of various pollution sources and the second historical concentration contribution values. If the model error value is less than or equal to a preset error threshold, the initial multiple linear regression model is used as the target multiple linear regression model. If the model error value is greater than the preset error threshold, the initial multiple linear regression model is iteratively optimized until the model error value of the iteratively optimized multiple linear regression model is less than or equal to the preset error threshold, and the iteratively optimized multiple linear regression model is taken as the target multiple linear regression model.

4. The method according to claim 1, characterized in that, The historical concentration contribution value is obtained based on the analysis of historical pollution source chemical composition data. Before constructing the corresponding target multiple linear regression model based on the historical concentration contribution values ​​and historical multi-particle size mass concentrations of various pollution sources within the same historical period, the method further includes: Acquire historical pollution source chemical composition data within a historical period, wherein the historical pollution source chemical composition data is collected by super monitoring stations according to a preset period; By using a positive matrix factorization model to analyze the chemical composition data of historical pollution sources, the historical concentration contribution values ​​of various pollution sources are obtained.

5. The method according to claim 1, characterized in that, After predicting the real-time multi-particle-size mass concentration for each type of pollution source using the target multiple linear regression model to obtain the predicted concentration contribution values ​​for each type of pollution source, the method further includes: In response to the triggering of a preset time interval, acquire a set of historical pollution source chemical component data with the shortest time interval between the acquisition time point and the real-time multi-particle size mass concentration; The historical pollution source chemical composition data were analyzed to obtain the reference concentration contribution values ​​of various pollution sources; For each type of pollution source, the root mean square error between the reference concentration contribution value and the predicted concentration contribution value is calculated to obtain the model error value of the target multiple linear regression model corresponding to each type of pollution source. If the model error value is greater than the preset error threshold, the model parameters of the target multiple linear regression model are updated according to the reference concentration contribution value, so as to perform subsequent pollution source real-time concentration contribution value prediction operation based on the updated target multiple linear regression model.

6. A fine particulate pollution source apportionment device based on mass concentration across multiple particle size ranges, characterized in that, include: The construction module is used to construct corresponding target multiple linear regression models for each type of pollution source based on the historical concentration contribution values ​​and historical multi-particle size range mass concentrations within the same historical period. The historical concentration contribution values ​​include the first historical concentration contribution values ​​within the first historical period, and the historical multi-particle size range mass concentrations include the first historical multi-particle size range mass concentrations within the first historical period. For any type of pollution source, the construction of the target multiple linear regression model based on the historical concentration contribution values ​​and historical multi-particle size range mass concentrations within the same historical period includes: The first historical concentration contribution value is used as the target output, and the first historical multi-particle size segment mass concentration is used as the independent variable to construct an initial multiple linear regression model, wherein the first historical multi-particle size segment mass concentration includes the mass concentration distribution of the pollution source in a preset number of particle size channels. The mass concentration distribution weights of the pollution source in the particle size channel are obtained by solving the initial multiple linear regression model. The weight distribution of the pollution source in each particle size channel is iteratively optimized based on a preset error threshold, and a target multiple linear regression model is constructed based on the optimized weight distribution. The acquisition module is used to obtain the real-time multi-particle size mass concentration corresponding to various pollution sources in response to fine particulate pollution source analysis instructions; The prediction module is used to predict the real-time multi-particle size mass concentration for each type of pollution source using the target multiple linear regression model, so as to obtain the concentration contribution prediction value of each type of pollution source. The calculation module is used to calculate the fine particulate pollution source apportionment results based on the concentration contribution prediction values ​​of the various pollution sources. The calculation module includes: The calculation unit is used to calculate the ratio of the predicted concentration contribution of each pollution source to the sum of the predicted concentration contribution of all pollution sources for each type of pollution source, so as to obtain the contribution ratio of each type of pollution source. The generation unit is used to generate fine particulate pollution source analysis results based on the contribution ratio and the concentration contribution prediction values ​​of various pollution sources.

7. A storage medium, characterized in that, The storage medium stores at least one executable instruction that causes the processor to perform the operation corresponding to the fine particulate pollution source apportionment method based on the mass concentration of multiple particle size segments as described in any one of claims 1-5.

8. A terminal, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation corresponding to the fine particulate pollution source apportionment method based on the mass concentration of multiple particle size segments as described in any one of claims 1-5.

Images