A PM2.5 and ozone collaborative source tracing analysis method and system based on multi-source data coupling

Abstract

This application discloses a method and system for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling. The method includes the following steps: establishing a multi-dimensional dataset; inputting the multi-dimensional dataset into a multi-phase co-existing pollution source tracing model to obtain preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point; correcting the preliminary estimates based on a dual-pollutant cross-sensitivity model to obtain corrected pollution source contribution values; determining the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring point based on the corrected pollution source contribution values; constructing a synergistic source tracing analysis matrix of PM2.5 and ozone based on the optimal transmission paths; and generating synergistic source tracing results of PM2.5 and ozone based on the synergistic source tracing analysis matrix. This application, through multi-model concatenation to correct biases and combined with dynamic path search, can accurately identify key pollution sources that significantly contribute to the monitoring points and the dominant transmission paths of pollutants, achieving high source tracing accuracy.

Description

Technical Field

[0001] This application relates to the field of data processing, and more specifically, to a method and system for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling. Background Technology

[0002] As air pollution control becomes more refined, PM2.5 and ozone, as key pollutants affecting air quality, exhibit complex coupling relationships in atmospheric transport, diffusion, and chemical reactions. Accurately identifying their pollution sources and transport pathways has become a core requirement for improving regional air quality. Currently, atmospheric environmental monitoring has established a multi-source data acquisition system encompassing ground stations, satellite remote sensing, and mobile devices, accumulating massive amounts of pollution source emissions, meteorological fields, and geospatial data. However, effectively coupling multi-source data and combining it with pollutant synergistic mechanisms to achieve source tracing analysis still faces technical bottlenecks. Traditional source tracing methods often target single pollutants, making them unsuitable for the coordinated management of PM2.5 and ozone. Existing technologies for synergistic source tracing of PM2.5 and ozone have two significant drawbacks: First, existing source tracing models often operate independently, lacking deep coupling between multiple models. For example, some methods only use a single pollution source tracing model to estimate the contribution of pollution sources, without combining a dual-pollutant cross-sensitivity model to correct estimation biases caused by the chemical reactions between the two pollutants. This results in the source tracing results failing to accurately reflect the actual pollution sources under the synergistic effects of pollutants. Second, existing technologies have limitations in utilizing multi-source data. They have failed to construct a unified multi-source coupled pollution source tracing platform. Multi-source data is often stored in a scattered manner and accessed individually, making it impossible to achieve real-time data transmission, efficient management, and dynamic interaction with model algorithms. This makes it difficult to fully realize the value of the data. Furthermore, during the pollution path search process, the search strategy is not dynamically adjusted in conjunction with multi-source data, affecting the accuracy and timeliness of transmission path analysis.

[0003] Therefore, how to provide a collaborative source tracing technology system that can integrate multi-source data and fuse multi-model algorithms to improve the accuracy of pollution source identification and the efficiency of transmission path analysis for PM2.5 and ozone collaborative source tracing has become an urgent problem to be solved in this field. Summary of the Invention

[0004] To address the aforementioned issues, this application proposes a method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling, comprising the following steps: establishing a multi-dimensional dataset; inputting the multi-dimensional dataset into a multi-phase co-existing pollution source tracing model to obtain preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point; correcting the preliminary estimates based on a dual-pollutant cross-sensitivity model to obtain corrected pollution source contribution values; determining the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring point based on the corrected pollution source contribution values, as well as the pollutant transmission intensity and duration corresponding to each optimal transmission path; constructing a PM2.5 and ozone synergistic source tracing analysis matrix based on the optimal transmission paths and their corresponding transmission intensity and duration, and determining the comprehensive contribution coefficient; identifying, based on the synergistic source tracing analysis matrix, calibrated pollution sources whose contribution to PM2.5 and ozone concentrations at each monitoring point exceeds a preset threshold, and simultaneously determining the dominant transmission paths of pollutants from the calibrated pollution sources to the monitoring points, thus forming the PM2.5 and ozone synergistic source tracing results.

[0005] As mentioned above, establishing a multi-dimensional dataset includes collecting PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to establish a multi-dimensional dataset.

[0006] As mentioned above, inputting the multi-dimensional dataset into the multi-phase pollution source tracing model to obtain preliminary estimates of the contribution of different pollution sources to PM2.5 and ozone concentrations at each monitoring point includes: converting the multi-dimensional dataset according to the input format of the multi-phase pollution source tracing model; determining the parameters required for the multi-phase pollution source tracing model; and inputting the multi-dimensional dataset into the multi-phase pollution source tracing model after completing the input format conversion and determining the required parameters.

[0007] As shown above, the multiphase co-occurring pollution source tracing model is represented as follows: ; in Let t be the simulated concentration of pollutant of type k at the j-th altitude layer of the i-th monitoring point. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.5. The time represents ozone.

[0008] As described above, the process of correcting the preliminary estimates based on the dual-pollutant cross-sensitivity model to obtain the corrected pollution source contribution values ​​includes the following sub-steps: extracting the preliminary estimates of PM2.5 and ozone concentrations at each monitoring point from the multiphase co-existing pollution source tracing model, and simultaneously extracting the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points in the same time series, establishing a comparison dataset of the preliminary estimates and actual monitored concentrations; calculating the Pearson correlation coefficient and partial correlation coefficient of PM2.5 and ozone concentrations based on the comparison dataset, determining the degree of linear correlation between the changes in the concentrations of the two pollutants, and analyzing the synergistic change patterns of the concentrations of the two pollutants under different temperature and humidity conditions, identifying the calibration meteorological threshold for the chemical reaction coupling effect; inputting the degree of linear correlation, synergistic change patterns, and calibration meteorological thresholds into the dual-pollutant cross-sensitivity model, adjusting the cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model, and outputting the correction amounts for PM2.5 and ozone concentrations.

[0009] A PM2.5 and ozone co-source tracing analysis system based on multi-source data coupling includes: a multi-dimensional data set establishment unit, a preliminary estimation value determination unit, a pollution source contribution value determination unit, an optimal transmission path determination unit, a comprehensive contribution coefficient determination unit, and a co-source tracing result formation unit. The multi-dimensional data set establishment unit is used to establish a multi-dimensional data set. The preliminary estimation value determination unit is used to input the multi-dimensional data set into a multi-phase co-existing pollution tracing model to obtain preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point. The pollution source contribution value determination unit is used to correct the preliminary estimates based on a dual-pollutant cross-sensitivity model to obtain the corrected pollution source contribution values. The optimal transmission path determination unit... The system comprises three units: a **determining unit**, which determines the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring points based on the corrected pollution source contribution values, as well as the pollutant transmission intensity and duration corresponding to each optimal transmission path; a **comprehensive contribution coefficient determination unit**, which constructs a PM2.5 and ozone co-source tracing analysis matrix based on the optimal transmission paths and their corresponding transmission intensity and duration, and determines the comprehensive contribution coefficient; and a **co-source tracing result formation unit**, which identifies, based on the co-source tracing analysis matrix, the calibrated pollution sources whose contribution to PM2.5 and ozone concentrations at each monitoring point exceeds a preset threshold, and simultaneously determines the dominant transmission paths of pollutants from the calibrated pollution sources to the monitoring points, thus forming the PM2.5 and ozone co-source tracing results.

[0010] As mentioned above, the multi-dimensional data set establishment unit establishes a multi-dimensional data set by collecting PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to establish a multi-dimensional data set.

[0011] As described above, the preliminary estimation value determination unit inputs the multi-dimensional data set into the multi-phase co-existing pollution source tracing model to obtain preliminary estimates of the contribution of different pollution sources to the PM2.5 and ozone concentrations at each monitoring point. This includes: converting the multi-dimensional data set according to the input format of the multi-phase co-existing pollution source tracing model; determining the parameters required by the multi-phase co-existing pollution source tracing model; and inputting the multi-dimensional data set into the multi-phase co-existing pollution source tracing model after completing the input format conversion and determining the required parameters.

[0012] As shown above, in the preliminary estimate determination unit, the multiphase co-existing pollution source tracing model is represented as follows: ; in Let t be the simulated concentration of pollutant of type k at the j-th altitude layer of the i-th monitoring point. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.5. The time represents ozone.

[0013] As described above, the pollution source contribution determination unit corrects the preliminary estimated values ​​based on the dual-pollutant cross-sensitivity model. Obtaining the corrected pollution source contribution values ​​involves the following sub-steps: extracting the preliminary estimated values ​​of PM2.5 and ozone concentrations at each monitoring point from the multi-phase co-existing pollution source tracing model; simultaneously extracting the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points under the same time series, establishing a comparison dataset of the preliminary estimated values ​​and actual monitored concentrations; calculating the Pearson correlation coefficient and partial correlation coefficient of PM2.5 and ozone concentrations based on the comparison dataset, determining the degree of linear correlation between the two pollutant concentration changes, and analyzing the synergistic change patterns of the two pollutant concentrations under different temperature and humidity conditions, identifying the calibration meteorological threshold for chemical reaction coupling effects; inputting the degree of linear correlation, synergistic change patterns, and calibration meteorological thresholds into the dual-pollutant cross-sensitivity model, adjusting the cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model, and outputting the correction amounts for PM2.5 and ozone concentrations.

[0014] This application has the following beneficial effects: (1) This application can integrate multi-source data such as ground monitoring, satellite remote sensing, and mobile monitoring to realize real-time data transmission, classified storage, rapid query and dynamic interaction, change the state of scattered storage and separate call of multi-source data in the prior art, fully explore the value of data and solve the problem of insufficient data utilization.

[0015] (2) This application connects a multi-phase co-occurring pollution source tracing model, a dual-pollutant cross-sensitivity model and a pollution path dynamic search algorithm. First, the multi-phase co-occurring model is used to preliminarily estimate the contribution of pollution sources. Then, the dual-pollutant cross-sensitivity model is used to correct the estimation deviation caused by PM2.5 and ozone chemical reaction. Finally, the path search strategy is dynamically adjusted by combining multi-source data, which breaks the limitation of independent operation of multiple models in the prior art and solves the problem of inaccurate source tracing results.

[0016] (3) This application uses multi-model concatenation to correct biases and combines dynamic path search to accurately identify key pollution sources and dominant pollutant transport paths that contribute significantly to monitoring sites, resulting in high source tracing accuracy. Simultaneously, it leverages a multi-source coupling platform to achieve deep collaboration between multi-source data and multiple models, avoiding the inefficiency of scattered data access and improving the overall efficiency of source tracing analysis, leading to more efficient data utilization. Through a complete technical system from data acquisition and model calculation to result output, it provides more targeted scientific support for regional joint prevention and control of air pollution. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart illustrating the PM2.5 and ozone co-source analysis method based on multi-source data coupling provided in the embodiments of this application. Figure 2 This is a schematic diagram of the internal structure of a PM2.5 and ozone co-source analysis system based on multi-source data coupling, provided according to an embodiment of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0020] Example 1 like Figure 1 As shown, this embodiment provides a method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling, specifically including the following steps: Step S1: Establish a multi-dimensional dataset.

[0021] The multi-source coupled pollution tracing platform collects PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to establish a multi-dimensional data set.

[0022] The concentration data includes PM2.5 mass concentration and ozone volume fraction at different monitoring points over a continuous time series; the meteorological field data includes wind speed, wind direction, temperature, humidity, and air pressure parameters; the pollution source emission data includes the emission intensity and emission time information of pollutants from stationary and mobile sources; and the geospatial data includes the coordinates of monitoring points and the geographical coordinates of pollution sources.

[0023] As a specific embodiment of the present invention, a multi-source coupled pollution tracing platform is used to complete multi-dimensional data collection and aggregation. This platform is a core platform that integrates multi-source data collection, storage, management, and model interaction, supporting the entire collaborative tracing process. It can provide basic data support for subsequent tracing analysis.

[0024] The following are the specific implementation methods for data collection on the platform:

[0025] The platform first collects data through ground-based monitoring terminals distributed throughout the monitoring area. The terminal deployment density is set at one terminal per 50 square kilometers. It continuously collects PM2.5 mass concentration and ozone volume fraction data at a frequency of once per hour to ensure the acquisition of time-series data for more than 30 consecutive days. The PM2.5 concentration monitoring range covers 0 to 1000, and the ozone volume fraction monitoring range covers 0 to 500.

[0026] Simultaneously, regional-scale atmospheric data transmitted back by polar-orbiting satellites are received via a satellite remote sensing module. The spatial resolution of the remote sensing data is set to 1 kilometer, and the temporal resolution is twice a day, supplementing the coverage gaps of ground monitoring over large areas.

[0027] Then, mobile monitoring devices are used to collect data in key areas such as main traffic routes and industrial parks. The sampling interval of the mobile devices is set to once every 5 minutes, and the collection time is no less than 2 hours each time.

[0028] Meteorological field data were obtained from regional meteorological stations, including wind speed (monitoring range 0 to 30), wind direction (0 to 360 degrees), temperature (-30 to 50), humidity (0 to 100%), and air pressure (800 to 1100), with a collection frequency of once per hour.

[0029] Pollution source emission data are obtained through enterprise declarations and environmental protection department verification. For stationary sources, the pollutant emission intensity (range 0 to 1000) and daily emission time period (accurate to the hour) of each emission outlet are recorded. For mobile sources, the average emission intensity (range 0 to 10) and average daily mileage are calculated according to vehicle type.

[0030] Geospatial data is acquired using a GIS system, accurately recording the latitude and longitude coordinates (accurate to 6 decimal places) and altitude (range -50 to 3000) of monitoring points and pollution sources.

[0031] All the collected data were categorized and organized according to the format of "data type-collection time-collection location". Data with a missing rate of more than 5% was removed, and finally a multi-dimensional data set was established, including 6 types of sub-data, a time span of more than 30 days, and a coverage area of ​​no less than 1,000 square kilometers.

[0032] Because the data sources of multi-dimensional datasets are heterogeneous, diverse, and complementary, using them directly without fusion would lead to data inconsistencies and accumulated errors. Therefore, this embodiment also requires data fusion of the collected data. The multi-source coupled pollution tracing platform adopts a data weight allocation model during the data fusion process, expressed as: ; in Let be the fusion weight of the data from the s-th data source at time t for the data from the ith monitoring point. Total number of data source types Let be the standard deviation of the concentration data of the s-th data source at the ith monitoring point at time t. Let be the data integrity coefficient of the s-th data source at the ith monitoring point at time t. The data source types include ground monitoring station data, satellite remote sensing data, and mobile monitoring vehicle data.

[0033] Specifically, the data fusion process of the multi-source coupled pollution tracing platform determines the fusion weight of each data source by quantifying the standard deviation and integrity of the data sources, thereby achieving efficient integration of multi-source data and improving data reliability. During implementation, the data source types are first clearly defined, including ground monitoring stations, satellite remote sensing, and mobile monitoring vehicles, with a total of no fewer than three types. The number of each type of data source is set according to monitoring needs: at least one ground monitoring station per 50 square kilometers, satellite remote sensing covering the entire monitoring area, and at least five mobile monitoring vehicles. The data standard deviation is calculated statistically from historical data. Ground monitoring station data has a relatively low standard deviation, ranging from 5 to 15; satellite remote sensing data has a medium standard deviation, ranging from 10 to 25; and mobile monitoring vehicle data has a relatively high standard deviation, ranging from 12 to 30. The data integrity coefficient is calculated as the proportion of valid data entries to the total number of collected data entries. The coefficient is 1.0 when the integrity reaches 95% or higher, decreasing by 0.1 for every 5% decrease, down to a minimum of 0.5. When calculating the fusion weight, the reciprocal of the standard deviation of each data source is first taken to eliminate the influence of dimensions. Then, the sum of the reciprocals of all data sources is calculated. The ratio of the reciprocal of a single data source to the sum is used as the basic weight. Finally, this ratio is multiplied by the data integrity coefficient to obtain the final fusion weight. For example, if the standard deviation of ground monitoring station data is 10, its reciprocal is 0.1. If the reciprocals of satellite remote sensing and mobile monitoring vehicle data are 0.05 and 0.04 respectively, the sum is 0.19, and the basic weight is 0.1 / 0.19 ≈ 0.526. If the integrity coefficient is 1.0, the final weight is 0.526. This model reduces the overall error after data fusion by 20% to 30%, while prioritizing the retention of data contributions with high reliability and good integrity, avoiding data failure caused by the failure of a single data source. This provides high-quality integrated data for subsequent model calculations, ensuring the accuracy and reliability of the basic data for source tracing analysis.

[0034] Step S2: Input the multi-dimensional dataset into the multi-phase pollution source tracing model to obtain preliminary estimates of the contribution of different pollution sources to the PM2.5 and ozone concentrations at each monitoring point.

[0035] Before inputting the multi-dimensional dataset into the multi-phase pollution source tracing model, the process also includes converting the multi-dimensional dataset according to the model input format.

[0036] The concentration data and meteorological field data are aligned in time series, with a time step of 1 hour. Pollution source emission data are linked with geospatial data to clarify the correspondence between the geographical location and emission parameters of each pollution source.

[0037] The transport and diffusion processes of PM2.5 and ozone in the atmosphere were simulated using a multiphase co-polluting source tracing model to obtain preliminary estimates of the contributions of different pollution sources to the PM2.5 and ozone concentrations at various monitoring sites. The multiphase co-polluting source tracing model is expressed as follows: ; in This represents the preliminary estimate of the k-th type of pollutant at the j-th altitude layer of the i-th monitoring point at time t. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.5. The time represents ozone.

[0038] When the multiphase pollution source tracing model is running, the simulation area is set to be consistent with the data coverage area in step S1, the spatial grid resolution is set to 1 km × 1 km, and the vertical direction is divided into 5 height layers (0-50 m, 50-100 m, 100-200 m, 200-500 m, 500-1000 m), and the simulation duration is the same as the data acquisition duration (more than 30 days). The model simulates the transport and diffusion process of PM2.5 and ozone in the atmosphere through numerical calculations, considering the influence of wind speed and direction on the horizontal transport of pollutants, the influence of temperature and humidity on the diffusion rate of pollutants, and the interference of geographical altitude differences on airflow. At the same time, it combines the emission intensity of pollution sources and time period to calculate the contribution of each pollution source to the pollutant concentration at different height layers of each monitoring point in each time step. During the simulation, intermediate results are verified every 12 hours to ensure that the deviation between the simulation results and the actual monitoring data is initially controlled within 30%. After the simulation is completed, the preliminary estimated value of the contribution of each pollution source to the PM2.5 and ozone concentrations at each monitoring point is output. The estimated value is stored in the format of "pollution source number-monitoring point number-time step-pollutant type", forming a preliminary estimation dataset with at least 1,000 records, which provides data input for subsequent correction work.

[0039] Furthermore, when the multiphase pollution source tracing model is running, the basic parameters required for model calculation are first determined, and the total number of sources is fixed. The number of mobile sources is set based on the actual number of cases investigated within the monitoring area, typically no less than 50. The statistics are categorized by vehicle type, including major types such as passenger and freight transport, with no fewer than 20 types. (Pollutant transport delay time) Based on the straight-line distance between the pollution source and the monitoring point and the average wind speed, the delay time increases by 0.5 to 2 hours for every 10 kilometers increase in distance; however, this needs to be dynamically adjusted based on real-time wind direction. Emission diffusion coefficient. Based on the height of the pollution source and the surrounding topography, the coefficient range for plains areas is set at 0.6 to 0.9, and for mountainous areas at 0.2 to 0.5; geographical attenuation factor. Based on the distance attenuation law, the factor is set at 1.0 within 1 kilometer of the pollution source, decreasing by 0.1 for every additional 5 kilometers, down to a minimum of 0.3. Atmospheric chemical conversion coefficient. Separate parameters were set for PM2.5 and ozone. The PM2.5 conversion coefficient is affected by humidity, increasing by 0.05 to 0.1 for every 10% increase in humidity. The ozone conversion coefficient is affected by temperature, increasing by 0.08 to 0.12 for every 5% increase in temperature. Calculations were performed hourly, first obtaining the emission intensity of pollution sources for the corresponding time period, then matching historical emission data with transmission delay time, and finally multiplying by the diffusion coefficient, geographical attenuation factor, and chemical conversion coefficient to obtain simulated pollutant concentration values ​​at different altitudes for each monitoring point. This model, through multi-parameter collaborative calculation, initially controlled the deviation between the simulated concentration results and actual monitoring data within 30%, laying a data foundation for subsequent correction work. It also achieved preliminary quantification of the contribution of pollution sources, avoiding excessive deviations caused by single-parameter estimation.

[0040] Step S3: Correct the preliminary estimate based on the dual-pollutant cross-sensitivity model to obtain the corrected pollution source contribution value.

[0041] The method involves using a dual-pollutant cross-sensitivity model to correct the initial estimates, eliminating biases caused by the interaction between PM2.5 and ozone, and improving the accuracy of pollution source contribution estimation.

[0042] Step S3 includes the following sub-steps: Step S31: Extract the preliminary estimated values ​​of PM2.5 and ozone concentrations at each monitoring point from the multiphase pollution source tracing model output, and simultaneously extract the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points in the same time series to establish a comparison dataset of preliminary estimated values ​​and actual monitored concentrations.

[0043] Specifically, preliminary estimates of PM2.5 and ozone concentrations at each time step are extracted from the preliminary estimate dataset output in step S2. At the same time, the actual monitored concentrations of PM2.5 and ozone at the same time step at the corresponding monitoring points are extracted from the multi-dimensional data set in step S1. A comparison dataset of "preliminary estimate - actual monitored value - time - location" is established to ensure that the number of comparison data is no less than 800 and the time span covers the entire collection period.

[0044] Step S32: Based on the comparison dataset, calculate the Pearson correlation coefficient and partial correlation coefficient between PM2.5 and ozone concentrations to determine the degree of linear correlation between the concentration changes of the two pollutants. At the same time, analyze the synergistic change pattern of the concentrations of the two pollutants under different temperature and humidity conditions, and identify the calibration meteorological threshold of chemical reaction coupling effect.

[0045] The Pearson correlation coefficient is calculated by quantifying the linear correlation between PM2.5 and ozone concentration, while the partial correlation coefficient is calculated after controlling for confounding variables such as temperature and humidity. Specific algorithms can be found in existing technologies and will not be elaborated upon here.

[0046] The interaction characteristics of the two pollutants were analyzed, and the correlation coefficients between PM2.5 and ozone concentrations were calculated in different temperature ranges (each range is 5) and humidity ranges (each range is 10%). The meteorological conditions with strong correlation were determined (usually the correlation is highest when the temperature is 15-35 and the humidity is 40%-80%). The interaction within these ranges has the greatest impact on the estimation bias and needs to be corrected.

[0047] Step S33: Input the degree of linear correlation, the law of coordinated change and the calibrated meteorological threshold into the dual-pollutant cross-sensitivity model, adjust the cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model, and output the correction amount of PM2.5 and ozone concentration.

[0048] The comparative dataset and correlation analysis results are input into the dual-pollutant cross-sensitivity model. The model first identifies the distribution pattern of deviations between the preliminary estimates and actual monitoring values. Then, based on the chemical reaction coupling relationship between PM2.5 and ozone (such as the inhibitory effect of PM2.5 on ozone formation and the oxidative effect of ozone on PM2.5 components), it determines the key parameters for deviation correction and identifies the key parameters affecting the correction amount of PM2.5 and ozone concentrations in the dual-pollutant cross-sensitivity model. These parameters are directly related to the interaction between the two pollutants and the influence of environmental conditions on the deviation, such as the cross-sensitivity coefficient, the total number of concentration coupling terms, the total number of chemical reaction types, and the influence weight of the p-th type chemical reaction on the correction of PM2.5 and ozone concentrations. The above parameters are dynamically optimized at a frequency of once per hour to ensure that the deviation between the corrected data and the actual monitoring data is reduced to within 15%. The adjusted parameters are substituted into the dual-pollutant cross-sensitivity model, and the preliminary estimates of PM2.5 and ozone concentrations at all monitoring points at each time point are corrected one by one according to the correction amount. The corrected contribution value of each pollution source corresponding to each monitoring point is generated and stored in the database of the multi-source coupled pollution tracing platform.

[0049] The contribution value is recorded in the format of "pollution source number - monitoring point number - pollutant type - correction deviation rate", and the correction deviation rate should indicate the correction range for each data point.

[0050] The dual-pollutant cross-sensitivity model is expressed as follows: ; in This is the correction amount for PM2.5 concentration. This is a correction factor for ozone concentration. The cross sensitivity coefficient, This represents the total number of chemical reaction types. The weights of the effects of the p-th type of chemical reaction on the correction of PM2.5 and ozone concentrations are respectively. Let be the reaction rate of the p-th type of chemical reaction at time t. This represents the total number of orders of the concentration coupling terms. These are the coefficients for the q-th order concentration coupling term, used to correct for PM2.5 and ozone concentrations, respectively. To correct the previous PM2.5 concentration value, The ozone concentration value before correction.

[0051] The dual-pollutant cross-sensitivity model quantifies the interaction and chemical reaction effects between PM2.5 and ozone, correcting initial estimation biases and improving the accuracy of pollution source contribution estimation. During implementation, the key coefficients and parameters in the model are first determined. The cross-sensitivity coefficients are calibrated based on historical monitoring data. The α value range for PM2.5 concentration correction is set to 0.1 to 0.3, and the β value range is set to 0.05 to 0.2. The γ value range for ozone concentration correction is set to 0.15 to 0.35, and the δ value range is set to 0.08 to 0.25. Chemical reaction types include photochemical reactions, redox reactions, etc., totaling no fewer than 8 categories. The influence weight of each type of reaction is determined based on the reaction rate; the faster the reaction rate, the higher the weight, with a range of 0.05 to 0.2. The reaction rate is affected by temperature and light intensity. For every 5°C increase in temperature, the photochemical reaction rate increases by 10% to 20%, and for every 10,000 lux increase in light intensity, the rate increases by 8% to 15%. The order of the concentration coupling term is set according to the actual correlation analysis results, typically 3 to 5 orders. The coefficient of each order coupling term is adjusted based on the concentration correlation coefficient; the higher the correlation, the larger the coefficient, with a range of 0.03 to 0.15. During calculation, the uncorrected concentration values ​​of each monitoring point are first extracted. Combined with the chemical reaction rate and concentration coupling term for the corresponding time period, the concentration correction amounts for PM2.5 and ozone are calculated separately: the PM2.5 correction amount needs to be superimposed with the influence of ozone concentration and its own concentration coupling effect, and the ozone correction amount needs to be superimposed with the influence of PM2.5 concentration and its corresponding coupling effect. After correction, the deviation between the estimated PM2.5 and ozone concentrations was reduced from less than 30% initially to less than 15%. At the same time, the estimation deviation caused by the mutual inhibition or promotion of the two pollutants was eliminated, ensuring that the corrected pollution source contribution value can truly reflect the actual pollution source, providing accurate data for subsequent path search, and avoiding the distortion of source tracing results caused by the lack of quantification of pollutant interactions.

[0052] Step S4: Based on the corrected pollution source contribution values, determine the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring points, as well as the pollutant transmission intensity and transmission duration corresponding to each optimal transmission path.

[0053] In this embodiment, the main transmission path of pollutants is determined based on the corrected contribution value and meteorological data, and the key parameters in the transmission process are identified, providing a path basis for the construction of the source tracing matrix.

[0054] Based on the corrected pollution source contribution value, the algorithm uses a dynamic pollution path search algorithm, combined with the spatiotemporal variation characteristics of wind speed and direction in meteorological field data, to search for the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring point, and determines the pollutant transmission intensity and transmission duration corresponding to each transmission path.

[0055] Step S4 includes the following sub-steps: Step S41: Extract the corrected pollution source contribution value, and at the same time retrieve the meteorological field data of the corresponding time period, filter out the effective time series of wind speed and wind direction data, remove the time periods with missing or abnormal data, and form the algorithm input dataset.

[0056] First, obtain the corrected pollution source contribution value from step S3, and select pollution sources whose contribution value accounts for more than 5% of the total contribution as key analysis objects to ensure that the number of key pollution sources is not less than 30% of the total number of pollution sources. At the same time, extract the wind speed and wind direction data of the area between these key pollution sources and the corresponding monitoring points from the meteorological field data in step S1, and adjust the time resolution to once every 30 minutes to capture the dynamic changes in meteorological conditions.

[0057] Understandably, the corrected pollution source contribution values ​​can be retrieved from the database of the multi-source coupled pollution tracing platform. The corrected pollution source contribution values ​​output in step S3 are retrieved from the multi-source coupled pollution tracing platform database. Simultaneously, meteorological field data for the corresponding time period (consistent with the contribution value time series, not less than 30 days) are retrieved. Abnormal data is eliminated using data validity checks (e.g., wind speed data exceeding the 0-30 degree range or wind direction data exceeding the 0-360 degree range are considered abnormal), ensuring that the proportion of valid data is not less than 90%, thus forming the algorithm input dataset.

[0058] Step S42: Group the algorithm input dataset according to the correspondence between monitoring points and pollution sources, and construct a spatial grid for transmission path search for each group of data.

[0059] Each set of data includes the corrected contribution values ​​of a single monitoring point and all pollution sources, as well as wind speed and direction data for the corresponding time period.

[0060] The spacing between grid nodes can be adaptively adjusted according to the distance between the pollution source and the monitoring point.

[0061] The input dataset is grouped according to the correspondence between "monitoring point - pollution source". Each group includes the corrected contribution values ​​of one monitoring point and all pollution sources, as well as the wind speed and direction data for the corresponding time period. A spatial grid is constructed for each group of data. The spacing between grid nodes is adaptively adjusted according to the distance between the pollution source and the monitoring point (0.3 km for distances less than 10 km, 0.5 km for distances between 10-50 km, and 1 km for distances greater than 50 km) to ensure that the grid can completely cover the area between the two.

[0062] Step S43: Start the pollution path dynamic search algorithm, take the grid nodes as path nodes, determine the initial direction of path search based on wind speed direction, use the corrected contribution value as the path weight, and calculate the cumulative value of pollutant transmission intensity on different paths.

[0063] During algorithm execution, the area between key pollution sources and monitoring points is divided into a fine grid of 0.5 km × 0.5 km. Grid nodes serve as potential nodes for path search. The algorithm starts at the grid node containing the pollution source and ends at the grid node containing the monitoring point. It determines the initial search direction based on real-time wind direction and calculates the pollutant transmission rate between grids by combining wind speed. The transmission rate = wind speed × pollutant diffusion coefficient, which is dynamically adjusted according to temperature and humidity, ranging from 0.1 to 1.0. During the search process, wind direction and wind speed data are updated hourly to dynamically adjust the search path and avoid path deviations caused by changes in meteorological conditions. At the same time, the pollutant transmission intensity of each potential path is calculated.

[0064] The transmission intensity is calculated as: Corrected contribution value × Transmission rate × Transmission time. The transmission time is calculated based on the estimated time it takes for the pollutant to travel from the origin to the destination (range 0.5 to 24 hours).

[0065] This also incorporates real-time adjustments to the search direction based on wind direction changes, preventing path search deviations caused by sudden changes in wind direction.

[0066] The dynamic search algorithm for pollution paths is expressed as follows: ; in Let be the optimal transmission path from the m-th pollution source to the i-th monitoring point at time t. For the set of all possible transmission paths, This is the starting time of the path search. for The average wind speed along path I at time point, for The cosine of the angle between the wind direction and the path direction on path I at time moment. for The pollutant concentration decay rate along path I from the m-th pollution source to the i-th monitoring point at any given time. is the geographical connectivity coefficient of path I, reflecting the degree to which the terrain on path I hinders the transport of pollutants.

[0067] The pollution path dynamic search algorithm accurately identifies the main transmission paths of pollutants from pollution sources to monitoring points by quantifying factors such as wind speed, wind direction, concentration decay, and geographical obstacles, thus clarifying key parameters in the transmission process. During implementation, the basic parameters for algorithm operation are first determined. The path search start time coincides with the data acquisition start time. All possible transmission paths are generated by combining grid nodes, with a grid node spacing of 0.5 kilometers to ensure coverage of the entire area between the pollution source and the monitoring point. Average wind speed... Take the average value of the corresponding path segment over a time period, updated every 30 minutes, and the cosine of the angle between the wind direction and the path direction. Calculated by angle, the cosine value is 1.0 when the included angle is 0 degrees, 0 when the included angle is 90 degrees, and -1.0 when the included angle is 180 degrees. Pollutant concentration decay rate. Based on distance, the transmission rate decreases by 5% to 10% per kilometer. Furthermore, it is affected by terrain, with the attenuation rate in mountainous areas being 3% to 8% higher than in plains. The geographic connectivity coefficient is set according to terrain type: 0.9 to 1.0 for plains, 0.6 to 0.8 for hilly areas, and 0.3 to 0.5 for mountainous areas. During the search, the initial search direction is determined by the wind direction, starting from the grid node where the pollution source is located. The cumulative value of pollutant transmission intensity along the path is calculated node by node: first, the product of wind speed and the cosine of the angle is calculated for each time period; then, combined with the concentration attenuation rate and the geographic connectivity coefficient, the path contribution value for that time period is obtained; finally, the cumulative value of pollutant transmission intensity is obtained by accumulating the values ​​over time.

[0068] When the cumulative value of a certain path exceeds 20% of the total cumulative value of all paths, it is determined to be the optimal transmission path. Simultaneously, the coordinates of the path's start and end points are recorded, and the path length (cumulative straight-line distances between nodes, ranging from 1 to 100 kilometers), transmission intensity (ranging from 5 to 100), and transmission duration (statistically calculated based on cumulative transmission time, ranging from 0.5 to 24 hours) are calculated. The information on the main transmission paths is then associated with and stored with the corrected contribution value. This algorithm, through dynamic adjustment of the search direction and multi-parameter quantification calculations, ensures that the consistency between the identified main transmission paths and the actual pollutant migration direction reaches over 70%, providing accurate path parameters for subsequent source tracing matrix construction and avoiding misjudgments of the transmission process due to static path assumptions.

[0069] In another embodiment of the invention, the transmission intensity is further dynamically adjusted. The transmission intensity is determined by considering minor changes in terrain and wind speed fluctuations. While the transmission intensity can be determined using the above method when only wind speed is considered, dynamic adjustment of the transmission intensity can be performed when considering both minor changes in terrain and wind speed fluctuations.

[0070] The dynamically adjusted transmission strength is represented as follows: ; in This represents the calibrated pollutant transmission intensity at time t, where the m-th pollution source reaches the i-th monitoring point via the l-th transmission path. Let represent the initial transmission intensity at time t, from the m-th pollution source to the i-th monitoring point via the l-th transmission path. This represents the transmission intensity calibration coefficient (value range: 0.1~0.3, calibrated according to the regional pollution type; 0.25~0.3 for industrial areas, and 0.1~0.15 for suburban areas). The topographic complexity coefficient of the l-th transmission path is represented by a value ranging from 1.0 to 2.5, where 1.0 to 1.2 is used for plain paths, 1.3 to 1.8 for hilly paths, and 1.9 to 2.5 for mountain / valley paths. It is calculated from the elevation undulation in the geospatial data. This represents the real-time average wind speed along the l-th transmission path at time t (which can be obtained from meteorological field data fused by the multi-source coupled pollution tracing platform). This represents the historical average wind speed along the l-th transmission path at time t (obtained from 30 days of meteorological data). Let represent the meteorological compatibility coefficient between the m-th pollution source and the i-th monitoring point at time t.

[0071] Calculated It can replace the original transmission strength and perform cumulative value sorting and main transmission path filtering.

[0072] By quantifying the hindering / promoting effects of terrain on transmission intensity, such as the faster attenuation of transmission intensity on mountainous paths, the calibrated values ​​are more in line with reality. At the same time, by correcting the misjudgment of transmission intensity caused by abnormal fluctuations in wind speed, the path selection intensity is improved.

[0073] Step S5: Based on the optimal transmission path and the corresponding transmission intensity and duration, construct a PM2.5 and ozone co-source analysis matrix and determine the comprehensive contribution coefficient.

[0074] A collaborative source tracing analysis matrix is ​​constructed based on the main transmission paths and related parameters. This matrix integrates the pollution source contribution with the transmission path information to form a structured source tracing data carrier, providing a quantitative basis for the identification of key pollution sources.

[0075] Step S5 includes the following sub-steps: Step S51: Determine the dimension parameters of the collaborative source tracing analysis matrix.

[0076] The number of rows in the matrix is ​​set to the total number of pollution sources, with each row vector corresponding to one pollution source. The number of columns is set to the total number of monitoring points, with each column vector corresponding to one monitoring point. The matrix elements are defined as the comprehensive contribution coefficient.

[0077] Preferably, the number of row dimensions is set to the total number of pollution sources (consistent with the number of pollution sources investigated in step S1, not less than 50), with each row vector corresponding to one pollution source and labeled with the pollution source number; the number of column dimensions is set to the total number of monitoring points (consistent with the number of monitoring points deployed in step S1, not less than one per 50 square kilometers), with each column vector corresponding to one monitoring point and labeled with the point number; the matrix elements are defined as comprehensive contribution coefficients, and the coefficient values ​​need to quantify the degree of pollution contribution of pollution sources to monitoring points.

[0078] Step S52: Determine the preliminary comprehensive contribution coefficient based on transmission strength and transmission duration.

[0079] Extract the transmission intensity and transmission duration of each major transmission path in step S4, and calculate the path impact factor, where the path impact factor = transmission intensity × transmission duration / total transmission duration, and the total transmission duration is the sum of the transmission durations from the pollution source to all monitoring points. Simultaneously, extract the emission intensity level of key pollution sources from the pollution source emission data in step S1 (divided into 5 levels based on emission intensity, with level coefficients ranging from 1 to 5), and multiply the emission intensity level coefficient by the path impact factor to obtain the preliminary comprehensive contribution coefficient.

[0080] Step S53: Determine the corrected comprehensive contribution coefficient based on the preliminary comprehensive contribution coefficient.

[0081] Then, the interaction correction coefficient (determined based on pollutant correlation, ranging from 0.8 to 1.2) output from the dual-pollutant cross-sensitivity model in step S3 is introduced. The preliminary comprehensive contribution coefficient is multiplied by the correction coefficient to obtain the final comprehensive contribution coefficient, ensuring that the coefficient reflects the impact of pollutant interactions on the contribution value. Finally, according to the correspondence between "key pollution source number - monitoring point number", the final comprehensive contribution coefficient is filled into the corresponding position in the matrix. After completing the matrix construction, the matrix is ​​row-normalized so that the sum of the elements in each row equals the total contribution percentage of the pollution source to all monitoring points (the sum of the total percentages is 100%). The matrix data precision is retained to 4 decimal places, forming a complete PM2.5 and ozone co-source analysis matrix including rows, columns, and element parameters. The matrix data volume is no less than the number of key pollution sources × the total number of monitoring points.

[0082] The contribution coefficient calibration model was used in the construction of the PM2.5 and ozone co-source analysis matrix, and its expression is: ; in Let be the calibrated comprehensive contribution coefficient of the m-th pollution source to the i-th monitoring point. This is the initial comprehensive contribution coefficient. The length of the time series. The time weight at time t, Let be the actual pollutant concentration at the i-th monitoring point at time t. Let t be the simulated concentration at the ith monitoring point considering only the m-th pollution source. Let be the meteorological adaptation coefficient between the m-th pollution source and the i-th monitoring point at time t.

[0083] Step S54: Determine the final PM2.5 and ozone co-source analysis matrix based on the comprehensive contribution coefficient.

[0084] According to the correspondence between pollution source number and monitoring point number, the corrected comprehensive contribution coefficient is filled into the corresponding position of the collaborative source tracing analysis matrix. After the matrix is ​​constructed, the matrix is ​​row-standardized so that the sum of the elements in each row is equal to the proportion of the total contribution of the pollution source to all monitoring points, thus forming the final PM2.5 and ozone collaborative source tracing analysis matrix.

[0085] Step S6: Based on the collaborative source tracing analysis matrix, identify the calibrated pollution sources whose contribution to PM2.5 and ozone concentrations at each monitoring point exceeds the preset threshold, and determine the dominant pathways for the transmission of pollutants from the calibrated pollution sources to the monitoring points, thus forming the collaborative source tracing results for PM2.5 and ozone.

[0086] Based on the collaborative source tracing analysis matrix, key pollution sources and dominant transmission paths are identified, resulting in the final source tracing results. This provides precise targeting for pollution control and completes the entire collaborative source tracing analysis process.

[0087] First, a threshold for identifying key pollution sources is set. The threshold is determined according to the regional pollution control targets. Generally, pollution sources whose comprehensive contribution coefficient accounts for more than 10% of the total comprehensive contribution coefficient of the monitoring point are identified as key pollution sources. If the sum of the top three pollution sources in the total comprehensive contribution coefficient of a monitoring point exceeds 50%, then these three pollution sources are included in the scope of key pollution sources. This ensures that the number of key pollution sources corresponding to each monitoring point is controlled at 2-5, so as to avoid too many key pollution sources leading to a dispersion of control focus.

[0088] Next, for each critical pollution source, the path with the highest transmission intensity from the pollution source to the corresponding monitoring point is selected from the list of main transmission paths in step S4 and determined as the dominant transmission path. The dominant transmission path must meet the requirement that the transmission intensity accounts for more than 40% of the total transmission intensity from the pollution source to the monitoring point. If there are multiple paths with similar transmission intensities (difference less than 5%), they are all determined as dominant transmission paths. The number of dominant transmission paths corresponding to each critical pollution source is controlled to be between 1 and 3.

[0089] Then, the information on key pollution sources and dominant transmission paths is integrated, and a source tracing result list is formed in the format of "monitoring point number - key pollution source number - dominant transmission path parameter - comprehensive contribution coefficient percentage". The list should indicate the emission type (stationary source / mobile source) of each key pollution source, the main areas through which the dominant transmission path passes (accurate to the township level administrative region) and the comprehensive contribution coefficient percentage (retaining two decimal places).

[0090] Finally, the source tracing results are verified for reasonableness. The verification method is to compare the overlap rate of key pollution sources at different monitoring points (overlap rate = number of common key pollution sources / total number of key pollution sources, reasonable range 30%-70%) and the consistency between the dominant transmission path and the meteorological dominant direction (consistency must reach more than 70%). If the verification fails, return to step S3 to revise the contribution value until the verification passes, and finally form a complete and accurate source tracing result for PM2.5 and ozone, providing specific basis for pollution source control and transmission path optimization for regional joint prevention and control of air pollution.

[0091] Example 2 like Figure 2 As shown in the figure, a PM2.5 and ozone co-source analysis system based on multi-source data coupling is provided in an embodiment of this application. Specifically, it includes: a multi-dimensional data set establishment unit 210, a preliminary estimate determination unit 220, a pollution source contribution value determination unit 230, an optimal transmission path determination unit 240, a comprehensive contribution coefficient determination unit 250, and a co-source analysis result formation unit 260.

[0092] Step S1: Establish a multi-dimensional dataset.

[0093] The multi-source coupled pollution tracing platform collects PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to establish a multi-dimensional data set.

[0094] The concentration data includes PM2.5 mass concentration and ozone volume fraction at different monitoring points over a continuous time series; the meteorological field data includes wind speed, wind direction, temperature, humidity, and air pressure parameters; the pollution source emission data includes the emission intensity and emission time information of pollutants from stationary and mobile sources; and the geospatial data includes the coordinates of monitoring points and the geographical coordinates of pollution sources.

[0095] As a specific embodiment of the present invention, a multi-source coupled pollution tracing platform is used to complete multi-dimensional data collection and aggregation. This platform is a core platform that integrates multi-source data collection, storage, management, and model interaction, supporting the entire collaborative tracing process. It can provide basic data support for subsequent tracing analysis.

[0096] The following are the specific implementation methods for data collection on the platform:

[0097] The platform first collects data through ground-based monitoring terminals distributed throughout the monitoring area. The terminal deployment density is set at one terminal per 50 square kilometers. It continuously collects PM2.5 mass concentration and ozone volume fraction data at a frequency of once per hour to ensure the acquisition of time-series data for more than 30 consecutive days. The PM2.5 concentration monitoring range covers 0 to 1000, and the ozone volume fraction monitoring range covers 0 to 500.

[0098] Simultaneously, regional-scale atmospheric data transmitted back by polar-orbiting satellites are received via a satellite remote sensing module. The spatial resolution of the remote sensing data is set to 1 kilometer, and the temporal resolution is twice a day, supplementing the coverage gaps of ground monitoring over large areas.

[0099] Then, mobile monitoring devices are used to collect data in key areas such as main traffic routes and industrial parks. The sampling interval of the mobile devices is set to once every 5 minutes, and the collection time is no less than 2 hours each time.

[0100] Meteorological field data were obtained from regional meteorological stations, including wind speed (monitoring range 0 to 30), wind direction (0 to 360 degrees), temperature (-30 to 50), humidity (0 to 100%), and air pressure (800 to 1100), with a collection frequency of once per hour.

[0101] Pollution source emission data are obtained through enterprise declarations and environmental protection department verification. For stationary sources, the pollutant emission intensity (range 0 to 1000) and daily emission time period (accurate to the hour) of each emission outlet are recorded. For mobile sources, the average emission intensity (range 0 to 10) and average daily mileage are calculated according to vehicle type.

[0102] Geospatial data is acquired using a GIS system, accurately recording the latitude and longitude coordinates (accurate to 6 decimal places) and altitude (range -50 to 3000) of monitoring points and pollution sources.

[0103] All the collected data were categorized and organized according to the format of "data type-collection time-collection location". Data with a missing rate of more than 5% was removed, and finally a multi-dimensional data set was established, including 6 types of sub-data, a time span of more than 30 days, and a coverage area of ​​no less than 1,000 square kilometers.

[0104] Because the data sources of multi-dimensional datasets are heterogeneous, diverse, and complementary, using them directly without fusion would lead to data inconsistencies and accumulated errors. Therefore, this embodiment also requires data fusion of the collected data. The multi-source coupled pollution tracing platform adopts a data weight allocation model during the data fusion process, expressed as: ; in Let be the fusion weight of the data from the s-th data source at time t for the data from the ith monitoring point. Total number of data source types Let be the standard deviation of the concentration data of the s-th data source at the ith monitoring point at time t. Let be the data integrity coefficient of the s-th data source at the ith monitoring point at time t. The data source types include ground monitoring station data, satellite remote sensing data, and mobile monitoring vehicle data.

[0105] Specifically, the data fusion process of the multi-source coupled pollution tracing platform determines the fusion weight of each data source by quantifying the standard deviation and integrity of the data sources, thereby achieving efficient integration of multi-source data and improving data reliability. During implementation, the data source types are first clearly defined, including ground monitoring stations, satellite remote sensing, and mobile monitoring vehicles, with a total of no fewer than three types. The number of each type of data source is set according to monitoring needs: at least one ground monitoring station per 50 square kilometers, satellite remote sensing covering the entire monitoring area, and at least five mobile monitoring vehicles. The data standard deviation is calculated statistically from historical data. Ground monitoring station data has a relatively low standard deviation, ranging from 5 to 15; satellite remote sensing data has a medium standard deviation, ranging from 10 to 25; and mobile monitoring vehicle data has a relatively high standard deviation, ranging from 12 to 30. The data integrity coefficient is calculated as the proportion of valid data entries to the total number of collected data entries. The coefficient is 1.0 when the integrity reaches 95% or higher, decreasing by 0.1 for every 5% decrease, down to a minimum of 0.5. When calculating the fusion weight, the reciprocal of the standard deviation of each data source is first taken to eliminate the influence of dimensions. Then, the sum of the reciprocals of all data sources is calculated. The ratio of the reciprocal of a single data source to the sum is used as the basic weight. Finally, this ratio is multiplied by the data integrity coefficient to obtain the final fusion weight. For example, if the standard deviation of ground monitoring station data is 10, its reciprocal is 0.1. If the reciprocals of satellite remote sensing and mobile monitoring vehicle data are 0.05 and 0.04 respectively, the sum is 0.19, and the basic weight is 0.1 / 0.19 ≈ 0.526. If the integrity coefficient is 1.0, the final weight is 0.526. This model reduces the overall error after data fusion by 20% to 30%, while prioritizing the retention of data contributions with high reliability and good integrity, avoiding data failure caused by the failure of a single data source. This provides high-quality integrated data for subsequent model calculations, ensuring the accuracy and reliability of the basic data for source tracing analysis.

[0106] The preliminary estimation value determination unit 220 is used to input the multi-dimensional data set into the multi-phase co-existing pollution source tracing model to obtain preliminary estimates of the contribution of different pollution sources to the PM2.5 and ozone concentrations at each monitoring point.

[0107] Before inputting the multi-dimensional dataset into the multi-phase pollution source tracing model, the process also includes converting the multi-dimensional dataset according to the model input format.

[0108] The concentration data and meteorological field data are aligned in time series, with a time step of 1 hour. Pollution source emission data are linked with geospatial data to clarify the correspondence between the geographical location and emission parameters of each pollution source.

[0109] The transport and diffusion processes of PM2.5 and ozone in the atmosphere were simulated using a multiphase co-polluting source tracing model to obtain preliminary estimates of the contributions of different pollution sources to the PM2.5 and ozone concentrations at various monitoring sites. The multiphase co-polluting source tracing model is expressed as follows: ; in This represents the preliminary estimate of the k-th type of pollutant at the j-th altitude layer of the i-th monitoring point at time t. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.5. The time represents ozone.

[0110] When the multiphase pollution source tracing model is running, the simulation area is set to be consistent with the data coverage area in step S1, the spatial grid resolution is set to 1 km × 1 km, and the vertical direction is divided into 5 height layers (0-50 m, 50-100 m, 100-200 m, 200-500 m, 500-1000 m), and the simulation duration is the same as the data acquisition duration (more than 30 days). The model simulates the transport and diffusion process of PM2.5 and ozone in the atmosphere through numerical calculations, considering the influence of wind speed and direction on the horizontal transport of pollutants, the influence of temperature and humidity on the diffusion rate of pollutants, and the interference of geographical altitude differences on airflow. At the same time, it combines the emission intensity of pollution sources and time period to calculate the contribution of each pollution source to the pollutant concentration at different height layers of each monitoring point in each time step. During the simulation, intermediate results are verified every 12 hours to ensure that the deviation between the simulation results and the actual monitoring data is initially controlled within 30%. After the simulation is completed, the preliminary estimated value of the contribution of each pollution source to the PM2.5 and ozone concentrations at each monitoring point is output. The estimated value is stored in the format of "pollution source number-monitoring point number-time step-pollutant type", forming a preliminary estimation dataset with at least 1,000 records, which provides data input for subsequent correction work.

[0111] Furthermore, when the multiphase pollution source tracing model is running, the basic parameters required for model calculation are first determined, and the total number of sources is fixed. The number of mobile sources is set based on the actual number of cases investigated within the monitoring area, typically no less than 50. The statistics are categorized by vehicle type, including major types such as passenger and freight transport, with no fewer than 20 types. (Pollutant transport delay time) Based on the straight-line distance between the pollution source and the monitoring point and the average wind speed, the delay time increases by 0.5 to 2 hours for every 10 kilometers increase in distance; however, this needs to be dynamically adjusted based on real-time wind direction. Emission diffusion coefficient. Based on the height of the pollution source and the surrounding topography, the coefficient range for plains areas is set at 0.6 to 0.9, and for mountainous areas at 0.2 to 0.5; geographical attenuation factor. Based on the distance attenuation law, the factor is set at 1.0 within 1 kilometer of the pollution source, decreasing by 0.1 for every additional 5 kilometers, down to a minimum of 0.3. Atmospheric chemical conversion coefficient. Separate parameters were set for PM2.5 and ozone. The PM2.5 conversion coefficient is affected by humidity, increasing by 0.05 to 0.1 for every 10% increase in humidity. The ozone conversion coefficient is affected by temperature, increasing by 0.08 to 0.12 for every 5% increase in temperature. Calculations were performed hourly, first obtaining the emission intensity of pollution sources for the corresponding time period, then matching historical emission data with transmission delay time, and finally multiplying by the diffusion coefficient, geographical attenuation factor, and chemical conversion coefficient to obtain simulated pollutant concentration values ​​at different altitudes for each monitoring point. This model, through multi-parameter collaborative calculation, initially controlled the deviation between the simulated concentration results and actual monitoring data within 30%, laying a data foundation for subsequent correction work. It also achieved preliminary quantification of the contribution of pollution sources, avoiding excessive deviations caused by single-parameter estimation.

[0112] The pollution source contribution value determination unit 230 is used to correct the preliminary estimated value based on the dual pollutant cross-sensitivity model and obtain the corrected pollution source contribution value.

[0113] The method involves using a dual-pollutant cross-sensitivity model to correct the initial estimates, eliminating biases caused by the interaction between PM2.5 and ozone, and improving the accuracy of pollution source contribution estimation.

[0114] The pollution source contribution value determination unit 230 performs the following sub-steps: Step V1: Extract the preliminary estimated values ​​of PM2.5 and ozone concentrations at each monitoring point from the multiphase pollution source tracing model output, and simultaneously extract the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points in the same time series to establish a comparison dataset of preliminary estimated values ​​and actual monitored concentrations.

[0115] Specifically, preliminary estimates of PM2.5 and ozone concentrations at each time step are extracted from the preliminary estimate dataset output in step S2. At the same time, the actual monitored concentrations of PM2.5 and ozone at the same time step at the corresponding monitoring points are extracted from the multi-dimensional data set in step S1. A comparison dataset of "preliminary estimate - actual monitored value - time - location" is established to ensure that the number of comparison data is no less than 800 and the time span covers the entire collection period.

[0116] Step V2: Based on the comparison dataset, calculate the Pearson correlation coefficient and partial correlation coefficient between PM2.5 and ozone concentrations to determine the degree of linear correlation between the concentration changes of the two pollutants. At the same time, analyze the synergistic change pattern of the concentrations of the two pollutants under different temperature and humidity conditions, and identify the calibration meteorological threshold of chemical reaction coupling effect.

[0117] The Pearson correlation coefficient is calculated by quantifying the linear correlation between PM2.5 and ozone concentration, while the partial correlation coefficient is calculated after controlling for confounding variables such as temperature and humidity. Specific algorithms can be found in existing technologies and will not be elaborated upon here.

[0118] The interaction characteristics of the two pollutants were analyzed, and the correlation coefficients between PM2.5 and ozone concentrations were calculated in different temperature ranges (each range is 5) and humidity ranges (each range is 10%). The meteorological conditions with strong correlation were determined (usually the correlation is highest when the temperature is 15-35 and the humidity is 40%-80%). The interaction within these ranges has the greatest impact on the estimation bias and needs to be corrected.

[0119] Step V3: Input the degree of linear correlation, the law of coordinated change and the calibrated meteorological threshold into the dual-pollutant cross-sensitivity model, adjust the cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model, and output the correction amount of PM2.5 and ozone concentration.

[0120] The comparative dataset and correlation analysis results are input into the dual-pollutant cross-sensitivity model. The model first identifies the distribution pattern of deviations between the preliminary estimates and actual monitoring values. Then, based on the chemical reaction coupling relationship between PM2.5 and ozone (such as the inhibitory effect of PM2.5 on ozone formation and the oxidative effect of ozone on PM2.5 components), it determines the key parameters for deviation correction and identifies the key parameters affecting the correction amount of PM2.5 and ozone concentrations in the dual-pollutant cross-sensitivity model. These parameters are directly related to the interaction between the two pollutants and the influence of environmental conditions on the deviation, such as the cross-sensitivity coefficient, the total number of concentration coupling terms, the total number of chemical reaction types, and the influence weight of the p-th type chemical reaction on the correction of PM2.5 and ozone concentrations. The above parameters are dynamically optimized at a frequency of once per hour to ensure that the deviation between the corrected data and the actual monitoring data is reduced to within 15%. The adjusted parameters are substituted into the dual-pollutant cross-sensitivity model, and the preliminary estimates of PM2.5 and ozone concentrations at all monitoring points at each time point are corrected one by one according to the correction amount. The corrected contribution value of each pollution source corresponding to each monitoring point is generated and stored in the database of the multi-source coupled pollution tracing platform.

[0121] The contribution value is recorded in the format of "pollution source number - monitoring point number - pollutant type - correction deviation rate", and the correction deviation rate should indicate the correction range for each data point.

[0122] The dual-pollutant cross-sensitivity model is expressed as follows: ; in This is the correction amount for PM2.5 concentration. This is a correction factor for ozone concentration. The cross sensitivity coefficient, This represents the total number of chemical reaction types. The weights of the effects of the p-th type of chemical reaction on the correction of PM2.5 and ozone concentrations are respectively. Let be the reaction rate of the p-th type of chemical reaction at time t. This represents the total number of orders of the concentration coupling terms. These are the coefficients for the q-th order concentration coupling term, used to correct for PM2.5 and ozone concentrations, respectively. To correct the previous PM2.5 concentration value, The ozone concentration value before correction.

[0123] The dual-pollutant cross-sensitivity model quantifies the interaction and chemical reaction effects between PM2.5 and ozone, correcting initial estimation biases and improving the accuracy of pollution source contribution estimation. During implementation, the key coefficients and parameters in the model are first determined. The cross-sensitivity coefficients are calibrated based on historical monitoring data. The α value range for PM2.5 concentration correction is set to 0.1 to 0.3, and the β value range is set to 0.05 to 0.2. The γ value range for ozone concentration correction is set to 0.15 to 0.35, and the δ value range is set to 0.08 to 0.25. Chemical reaction types include photochemical reactions, redox reactions, etc., totaling no fewer than 8 categories. The influence weight of each type of reaction is determined based on the reaction rate; the faster the reaction rate, the higher the weight, with a range of 0.05 to 0.2. The reaction rate is affected by temperature and light intensity. For every 5°C increase in temperature, the photochemical reaction rate increases by 10% to 20%, and for every 10,000 lux increase in light intensity, the rate increases by 8% to 15%. The order of the concentration coupling term is set according to the actual correlation analysis results, typically 3 to 5 orders. The coefficient of each order coupling term is adjusted based on the concentration correlation coefficient; the higher the correlation, the larger the coefficient, with a range of 0.03 to 0.15. During calculation, the uncorrected concentration values ​​of each monitoring point are first extracted. Combined with the chemical reaction rate and concentration coupling term for the corresponding time period, the concentration correction amounts for PM2.5 and ozone are calculated separately: the PM2.5 correction amount needs to be superimposed with the influence of ozone concentration and its own concentration coupling effect, and the ozone correction amount needs to be superimposed with the influence of PM2.5 concentration and its corresponding coupling effect. After correction, the deviation between the estimated PM2.5 and ozone concentrations was reduced from less than 30% initially to less than 15%. At the same time, the estimation deviation caused by the mutual inhibition or promotion of the two pollutants was eliminated, ensuring that the corrected pollution source contribution value can truly reflect the actual pollution source, providing accurate data for subsequent path search, and avoiding the distortion of source tracing results caused by the lack of quantification of pollutant interactions.

[0124] The optimal transmission path determination unit 240 is used to determine the optimal transmission path of PM2.5 and ozone from each pollution source to the monitoring point based on the corrected pollution source contribution value, as well as the pollutant transmission intensity and transmission duration corresponding to each optimal transmission path.

[0125] In this embodiment, the main transmission path of pollutants is determined based on the corrected contribution value and meteorological data, and the key parameters in the transmission process are identified, providing a path basis for the construction of the source tracing matrix.

[0126] Based on the corrected pollution source contribution value, the algorithm uses a dynamic pollution path search algorithm, combined with the spatiotemporal variation characteristics of wind speed and direction in meteorological field data, to search for the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring point, and determines the pollutant transmission intensity and transmission duration corresponding to each transmission path.

[0127] The optimal transmission path determination unit 240 performs the following sub-steps: Step W1: Extract the corrected pollution source contribution value, and at the same time retrieve the meteorological field data for the corresponding time period, filter out the effective time series of wind speed and wind direction data, remove time periods with missing or abnormal data, and form the algorithm input dataset.

[0128] First, obtain the corrected pollution source contribution value from step S3, and select pollution sources whose contribution value accounts for more than 5% of the total contribution as key analysis objects to ensure that the number of key pollution sources is not less than 30% of the total number of pollution sources. At the same time, extract the wind speed and wind direction data of the area between these key pollution sources and the corresponding monitoring points from the meteorological field data in step S1, and adjust the time resolution to once every 30 minutes to capture the dynamic changes in meteorological conditions.

[0129] Understandably, the corrected pollution source contribution values ​​can be retrieved from the database of the multi-source coupled pollution tracing platform. The corrected pollution source contribution values ​​output in step S3 are retrieved from the multi-source coupled pollution tracing platform database. Simultaneously, meteorological field data for the corresponding time period (consistent with the contribution value time series, not less than 30 days) are retrieved. Abnormal data is eliminated using data validity checks (e.g., wind speed data exceeding the 0-30 degree range or wind direction data exceeding the 0-360 degree range are considered abnormal), ensuring that the proportion of valid data is not less than 90%, thus forming the algorithm input dataset.

[0130] Step W2: Group the algorithm input dataset according to the correspondence between monitoring points and pollution sources, and construct a spatial grid for transmission path search for each group of data.

[0131] Each set of data includes the corrected contribution values ​​of a single monitoring point and all pollution sources, as well as wind speed and direction data for the corresponding time period.

[0132] The spacing between grid nodes can be adaptively adjusted according to the distance between the pollution source and the monitoring point.

[0133] The input dataset is grouped according to the correspondence between "monitoring point - pollution source". Each group includes the corrected contribution values ​​of one monitoring point and all pollution sources, as well as the wind speed and direction data for the corresponding time period. A spatial grid is constructed for each group of data. The spacing between grid nodes is adaptively adjusted according to the distance between the pollution source and the monitoring point (0.3 km for distances less than 10 km, 0.5 km for distances between 10-50 km, and 1 km for distances greater than 50 km) to ensure that the grid can completely cover the area between the two.

[0134] Step W3: Start the dynamic search algorithm for pollution paths. Use grid nodes as path nodes, determine the initial direction of path search based on wind speed and direction, use the corrected contribution value as the path weight, and calculate the cumulative value of pollutant transmission intensity on different paths.

[0135] During algorithm execution, the area between key pollution sources and monitoring points is divided into a fine grid of 0.5 km × 0.5 km. Grid nodes serve as potential nodes for path search. The algorithm starts at the grid node containing the pollution source and ends at the grid node containing the monitoring point. It determines the initial search direction based on real-time wind direction and calculates the pollutant transmission rate between grids by combining wind speed. The transmission rate = wind speed × pollutant diffusion coefficient, which is dynamically adjusted according to temperature and humidity, ranging from 0.1 to 1.0. During the search process, wind direction and wind speed data are updated hourly to dynamically adjust the search path and avoid path deviations caused by changes in meteorological conditions. At the same time, the pollutant transmission intensity of each potential path is calculated.

[0136] The transmission intensity is calculated as: Corrected contribution value × Transmission rate × Transmission time. The transmission time is calculated based on the estimated time it takes for the pollutant to travel from the origin to the destination (range 0.5 to 24 hours).

[0137] This also incorporates real-time adjustments to the search direction based on wind direction changes, preventing path search deviations caused by sudden changes in wind direction.

[0138] The dynamic search algorithm for pollution paths is expressed as follows: ; in Let be the optimal transmission path from the m-th pollution source to the i-th monitoring point at time t. For the set of all possible transmission paths, This is the starting time of the path search. for The average wind speed along path I at time point, for The cosine of the angle between the wind direction and the path direction on path I at time moment. for The pollutant concentration decay rate along path I from the m-th pollution source to the i-th monitoring point at any given time. is the geographical connectivity coefficient of path I, reflecting the degree to which the terrain on path I hinders the transport of pollutants.

[0139] The pollution path dynamic search algorithm accurately identifies the main transmission paths of pollutants from pollution sources to monitoring points by quantifying factors such as wind speed, wind direction, concentration decay, and geographical obstacles, thus clarifying key parameters in the transmission process. During implementation, the basic parameters for algorithm operation are first determined. The path search start time coincides with the data acquisition start time. All possible transmission paths are generated by combining grid nodes, with a grid node spacing of 0.5 kilometers to ensure coverage of the entire area between the pollution source and the monitoring point. Average wind speed... Take the average value of the corresponding path segment over a time period, updated every 30 minutes, and the cosine of the angle between the wind direction and the path direction. Calculated by angle, the cosine value is 1.0 when the included angle is 0 degrees, 0 when the included angle is 90 degrees, and -1.0 when the included angle is 180 degrees. Pollutant concentration decay rate. Based on distance, the transmission rate decreases by 5% to 10% per kilometer. Furthermore, it is affected by terrain, with the attenuation rate in mountainous areas being 3% to 8% higher than in plains. The geographic connectivity coefficient is set according to terrain type: 0.9 to 1.0 for plains, 0.6 to 0.8 for hilly areas, and 0.3 to 0.5 for mountainous areas. During the search, the initial search direction is determined by the wind direction, starting from the grid node where the pollution source is located. The cumulative value of pollutant transmission intensity along the path is calculated node by node: first, the product of wind speed and the cosine of the angle is calculated for each time period; then, combined with the concentration attenuation rate and the geographic connectivity coefficient, the path contribution value for that time period is obtained; finally, the cumulative value of pollutant transmission intensity is obtained by accumulating the values ​​over time.

[0140] When the cumulative value of a certain path exceeds 20% of the total cumulative value of all paths, it is determined to be the optimal transmission path. Simultaneously, the coordinates of the path's start and end points are recorded, and the path length (cumulative straight-line distances between nodes, ranging from 1 to 100 kilometers), transmission intensity (ranging from 5 to 100), and transmission duration (statistically calculated based on cumulative transmission time, ranging from 0.5 to 24 hours) are calculated. The information on the main transmission paths is then associated with and stored with the corrected contribution value. This algorithm, through dynamic adjustment of the search direction and multi-parameter quantification calculations, ensures that the consistency between the identified main transmission paths and the actual pollutant migration direction reaches over 70%, providing accurate path parameters for subsequent source tracing matrix construction and avoiding misjudgments of the transmission process due to static path assumptions.

[0141] In another embodiment of the invention, the transmission intensity is further dynamically adjusted. The transmission intensity is determined by considering minor changes in terrain and wind speed fluctuations. While the transmission intensity can be determined using the above method when only wind speed is considered, dynamic adjustment of the transmission intensity can be performed when considering both minor changes in terrain and wind speed fluctuations.

[0142] The dynamically adjusted transmission strength is represented as follows: ; in This represents the calibrated pollutant transmission intensity at time t, where the m-th pollution source reaches the i-th monitoring point via the l-th transmission path. Let represent the initial transmission intensity at time t, from the m-th pollution source to the i-th monitoring point via the l-th transmission path. This represents the transmission intensity calibration coefficient (value range: 0.1~0.3, calibrated according to the regional pollution type; 0.25~0.3 for industrial areas, and 0.1~0.15 for suburban areas). The topographic complexity coefficient of the l-th transmission path is represented by a value ranging from 1.0 to 2.5, where 1.0 to 1.2 is used for plain paths, 1.3 to 1.8 for hilly paths, and 1.9 to 2.5 for mountain / valley paths. It is calculated from the elevation undulation in the geospatial data. This represents the real-time average wind speed along the l-th transmission path at time t (which can be obtained from meteorological field data fused by the multi-source coupled pollution tracing platform). This represents the historical average wind speed along the l-th transmission path at time t (obtained from 30 days of meteorological data). Let represent the meteorological compatibility coefficient between the m-th pollution source and the i-th monitoring point at time t.

[0143] Calculated It can replace the original transmission strength and perform cumulative value sorting and main transmission path filtering.

[0144] By quantifying the hindering / promoting effects of terrain on transmission intensity, such as the faster attenuation of transmission intensity on mountainous paths, the calibrated values ​​are more in line with reality. At the same time, by correcting the misjudgment of transmission intensity caused by abnormal fluctuations in wind speed, the path selection intensity is improved.

[0145] The comprehensive contribution coefficient determination unit 250 is used to construct a PM2.5 and ozone co-source analysis matrix based on the optimal transmission path and the corresponding transmission intensity and duration, and to determine the comprehensive contribution coefficient.

[0146] A collaborative source tracing analysis matrix is ​​constructed based on the main transmission paths and related parameters. This matrix integrates the pollution source contribution with the transmission path information to form a structured source tracing data carrier, providing a quantitative basis for the identification of key pollution sources.

[0147] The comprehensive contribution coefficient determination unit 250 performs the following sub-steps: Step Y1: Determine the dimensional parameters of the collaborative source tracing analysis matrix.

[0148] The number of rows in the matrix is ​​set to the total number of pollution sources, with each row vector corresponding to one pollution source. The number of columns is set to the total number of monitoring points, with each column vector corresponding to one monitoring point. The matrix elements are defined as the comprehensive contribution coefficient.

[0149] Preferably, the number of row dimensions is set to the total number of pollution sources (consistent with the number of pollution sources investigated in step S1, not less than 50), with each row vector corresponding to one pollution source and labeled with the pollution source number; the number of column dimensions is set to the total number of monitoring points (consistent with the number of monitoring points deployed in step S1, not less than one per 50 square kilometers), with each column vector corresponding to one monitoring point and labeled with the point number; the matrix elements are defined as comprehensive contribution coefficients, and the coefficient values ​​need to quantify the degree of pollution contribution of pollution sources to monitoring points.

[0150] Step Y2: Determine the preliminary comprehensive contribution coefficient based on transmission strength and transmission duration.

[0151] Extract the transmission intensity and transmission duration of each major transmission path in step S4, and calculate the path impact factor, where the path impact factor = transmission intensity × transmission duration / total transmission duration, and the total transmission duration is the sum of the transmission durations from the pollution source to all monitoring points. Simultaneously, extract the emission intensity level of key pollution sources from the pollution source emission data in step S1 (divided into 5 levels based on emission intensity, with level coefficients ranging from 1 to 5), and multiply the emission intensity level coefficient by the path impact factor to obtain the preliminary comprehensive contribution coefficient.

[0152] Step Y3: Determine the corrected comprehensive contribution coefficient based on the preliminary comprehensive contribution coefficient.

[0153] Then, the interaction correction coefficient (determined based on pollutant correlation, ranging from 0.8 to 1.2) output from the dual-pollutant cross-sensitivity model in step S3 is introduced. The preliminary comprehensive contribution coefficient is multiplied by the correction coefficient to obtain the final comprehensive contribution coefficient, ensuring that the coefficient reflects the impact of pollutant interactions on the contribution value. Finally, according to the correspondence between "key pollution source number - monitoring point number", the final comprehensive contribution coefficient is filled into the corresponding position in the matrix. After completing the matrix construction, the matrix is ​​row-normalized so that the sum of the elements in each row equals the total contribution percentage of the pollution source to all monitoring points (the sum of the total percentages is 100%). The matrix data precision is retained to 4 decimal places, forming a complete PM2.5 and ozone co-source analysis matrix including rows, columns, and element parameters. The matrix data volume is no less than the number of key pollution sources × the total number of monitoring points.

[0154] The contribution coefficient calibration model was used in the construction of the PM2.5 and ozone co-source analysis matrix, and its expression is: ; in Let be the calibrated comprehensive contribution coefficient of the m-th pollution source to the i-th monitoring point. This is the initial comprehensive contribution coefficient. The length of the time series. The time weight at time t, Let be the actual pollutant concentration at the i-th monitoring point at time t. Let t be the simulated concentration at the ith monitoring point considering only the m-th pollution source. Let be the meteorological adaptation coefficient between the m-th pollution source and the i-th monitoring point at time t.

[0155] Step Y4: Determine the final PM2.5 and ozone co-source analysis matrix based on the comprehensive contribution coefficient.

[0156] According to the correspondence between pollution source number and monitoring point number, the corrected comprehensive contribution coefficient is filled into the corresponding position of the collaborative source tracing analysis matrix. After the matrix is ​​constructed, the matrix is ​​row-standardized so that the sum of the elements in each row is equal to the proportion of the total contribution of the pollution source to all monitoring points, thus forming the final PM2.5 and ozone collaborative source tracing analysis matrix.

[0157] The collaborative source tracing result generation unit 260 is used to identify, based on the collaborative source tracing analysis matrix, the calibrated pollution sources whose contribution to PM2.5 and ozone concentrations at each monitoring point exceeds a preset threshold, and at the same time determine the dominant pathway for the transmission of pollutants from the calibrated pollution sources to the monitoring points, thereby generating collaborative source tracing results for PM2.5 and ozone.

[0158] Based on the collaborative source tracing analysis matrix, key pollution sources and dominant transmission paths are identified, resulting in the final source tracing results. This provides precise targeting for pollution control and completes the entire collaborative source tracing analysis process.

[0159] First, a threshold for identifying key pollution sources is set. The threshold is determined according to the regional pollution control targets. Generally, pollution sources whose comprehensive contribution coefficient accounts for more than 10% of the total comprehensive contribution coefficient of the monitoring point are identified as key pollution sources. If the sum of the top three pollution sources in the total comprehensive contribution coefficient of a monitoring point exceeds 50%, then these three pollution sources are included in the scope of key pollution sources. This ensures that the number of key pollution sources corresponding to each monitoring point is controlled at 2-5, so as to avoid too many key pollution sources leading to a dispersion of control focus.

[0160] Next, for each critical pollution source, the path with the highest transmission intensity from the pollution source to the corresponding monitoring point is selected from the list of main transmission paths in step S4 and determined as the dominant transmission path. The dominant transmission path must meet the requirement that the transmission intensity accounts for more than 40% of the total transmission intensity from the pollution source to the monitoring point. If there are multiple paths with similar transmission intensities (difference less than 5%), they are all determined as dominant transmission paths. The number of dominant transmission paths corresponding to each critical pollution source is controlled to be between 1 and 3.

[0161] Then, the information on key pollution sources and dominant transmission paths is integrated, and a source tracing result list is formed in the format of "monitoring point number - key pollution source number - dominant transmission path parameter - comprehensive contribution coefficient percentage". The list should indicate the emission type (stationary source / mobile source) of each key pollution source, the main areas through which the dominant transmission path passes (accurate to the township level administrative region) and the comprehensive contribution coefficient percentage (retaining two decimal places).

[0162] Finally, the source tracing results are verified for reasonableness. The verification method is to compare the overlap rate of key pollution sources at different monitoring points (overlap rate = number of common key pollution sources / total number of key pollution sources, reasonable range 30%-70%) and the consistency between the dominant transmission path and the meteorological dominant direction (consistency must reach more than 70%). If the verification fails, return to step S3 to revise the contribution value until the verification passes, and finally form a complete and accurate source tracing result for PM2.5 and ozone, providing specific basis for pollution source control and transmission path optimization for regional joint prevention and control of air pollution.

[0163] This application also provides a computer storage medium storing computer instructions, which, when invoked, are used to execute the PM2.5 and ozone co-source analysis method based on multi-source data coupling.

[0164] The embodiments disclosed in this invention provide a computer-readable storage medium storing computer program instructions. When the computer program instructions are executed on a computer, the computer executes the aforementioned method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling.

[0165] This invention provides a processor for processing the above-described method for synergistic source analysis of PM2.5 and ozone based on multi-source data coupling.

[0166] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0167] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0168] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0169] Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).

[0170] This application has the following beneficial effects: (1) This application can integrate multi-source data such as ground monitoring, satellite remote sensing, and mobile monitoring to realize real-time data transmission, classified storage, rapid query and dynamic interaction, change the state of scattered storage and separate call of multi-source data in the prior art, fully explore the value of data and solve the problem of insufficient data utilization.

[0171] (2) This application connects a multi-phase co-occurring pollution source tracing model, a dual-pollutant cross-sensitivity model and a pollution path dynamic search algorithm. First, the multi-phase co-occurring model is used to preliminarily estimate the contribution of pollution sources. Then, the dual-pollutant cross-sensitivity model is used to correct the estimation deviation caused by PM2.5 and ozone chemical reaction. Finally, the path search strategy is dynamically adjusted by combining multi-source data, which breaks the limitation of independent operation of multiple models in the prior art and solves the problem of inaccurate source tracing results.

[0172] (3) This application uses multi-model concatenation to correct biases and combines dynamic path search to accurately identify key pollution sources and dominant pollutant transport paths that contribute significantly to monitoring sites, resulting in high source tracing accuracy. Simultaneously, it leverages a multi-source coupling platform to achieve deep collaboration between multi-source data and multiple models, avoiding the inefficiency of scattered data access and improving the overall efficiency of source tracing analysis, leading to more efficient data utilization. Through a complete technical system from data acquisition and model calculation to result output, it provides more targeted scientific support for regional joint prevention and control of air pollution.

[0173] Although the examples referenced in this application are described for illustrative purposes only and not for limiting the scope of this application, changes, additions and / or deletions to the implementation may be made without departing from the scope of this application.

[0174] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling, characterized in that, Includes the following steps: Establish a multi-dimensional dataset; By inputting multi-dimensional datasets into the multi-phase pollution source tracing model, preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point are obtained. The preliminary estimates were corrected based on the dual-pollutant cross-sensitivity model to obtain the corrected pollution source contribution values. Based on the corrected pollution source contribution values, the optimal transmission paths of PM2.5 and ozone from each pollution source to the monitoring points are determined, as well as the pollutant transmission intensity and transmission duration corresponding to each optimal transmission path. Based on the optimal transmission path and the corresponding transmission intensity and duration, a synergistic source analysis matrix for PM2.5 and ozone is constructed to determine the comprehensive contribution coefficient. Based on the collaborative source tracing analysis matrix, the calibrated pollution sources that contribute more than a preset threshold to the PM2.5 and ozone concentrations at each monitoring point are identified. At the same time, the dominant pathways for the transmission of pollutants from the calibrated pollution sources to the monitoring points are determined, thus forming the collaborative source tracing results for PM2.5 and ozone.

2. The method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling as described in claim 1, characterized in that, Establishing a multi-dimensional dataset involves collecting PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to create a multi-dimensional dataset.

3. The method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling as described in claim 1, characterized in that, By inputting multi-dimensional datasets into a multi-phase pollution source tracing model, preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point are obtained, including: The multi-dimensional dataset is converted to the input format of the multi-phase co-occurring pollution source tracing model; Determine the parameters required for the multiphase co-existing pollution source tracing model; After completing the input format conversion and determining the required parameters, the multi-dimensional data set is input into the multi-phase co-existing pollution source tracing model.

4. The method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling as described in claim 3, characterized in that, The multiphase co-existing pollution source tracing model is expressed as follows: ； in Let t be the simulated concentration of pollutant of type k at the j-th altitude layer of the i-th monitoring point. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.

5. The time represents ozone.

5. The method for synergistic source tracing analysis of PM2.5 and ozone based on multi-source data coupling as described in claim 4, characterized in that, The preliminary estimates are corrected based on the dual-pollutant cross-sensitivity model. The corrected pollution source contribution values ​​are obtained through the following sub-steps: Preliminary estimates of PM2.5 and ozone concentrations at each monitoring point were extracted from the multiphase pollution source tracing model. At the same time, the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points in the same time series were extracted to establish a comparison dataset of preliminary estimates and actual monitored concentrations. Based on the comparative dataset, the Pearson correlation coefficient and partial correlation coefficient between PM2.5 and ozone concentrations were calculated to determine the degree of linear correlation between the concentration changes of the two pollutants. At the same time, the synergistic change patterns of the concentrations of the two pollutants under different temperature and humidity conditions were analyzed to identify the calibration meteorological threshold of chemical reaction coupling effect. The degree of linear correlation, the law of coordinated change, and the calibrated meteorological threshold are input into the dual-pollutant cross-sensitivity model. The cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model is adjusted, and the correction amount of PM2.5 and ozone concentration is output.

6. A PM2.5 and ozone co-source analysis system based on multi-source data coupling, characterized in that, include: The system includes a multi-dimensional data set establishment unit, a preliminary estimate determination unit, a pollution source contribution value determination unit, an optimal transmission path determination unit, a comprehensive contribution coefficient determination unit, and a collaborative source tracing result formation unit. Multi-dimensional dataset creation unit, used to create multi-dimensional datasets; The preliminary estimation value determination unit is used to input multi-dimensional data sets into the multi-phase co-existing pollution source tracing model to obtain preliminary estimates of the contribution of different pollution sources to the PM2.5 and ozone concentrations at each monitoring point; The pollution source contribution value determination unit is used to correct the preliminary estimated value based on the dual pollutant cross-sensitivity model and obtain the corrected pollution source contribution value. The optimal transmission path determination unit is used to determine the optimal transmission path of PM2.5 and ozone from each pollution source to the monitoring point based on the corrected pollution source contribution value, as well as the pollutant transmission intensity and transmission duration corresponding to each optimal transmission path. The comprehensive contribution coefficient determination unit is used to construct a PM2.5 and ozone co-source analysis matrix based on the optimal transmission path and the corresponding transmission intensity and duration, and to determine the comprehensive contribution coefficient. The collaborative source tracing result generation unit is used to identify, based on the collaborative source tracing analysis matrix, the calibrated pollution sources whose contribution to PM2.5 and ozone concentrations at each monitoring point exceeds a preset threshold, and at the same time determine the dominant pathways for the transmission of pollutants from the calibrated pollution sources to the monitoring points, thereby generating collaborative source tracing results for PM2.5 and ozone.

7. The PM2.5 and ozone co-source analysis system based on multi-source data coupling as described in claim 6, characterized in that, The multi-dimensional dataset establishment unit involves collecting PM2.5 and ozone concentration data, meteorological field data, pollution source emission data, and geospatial data in the atmospheric environment to establish a multi-dimensional dataset.

8. The PM2.5 and ozone co-source analysis system based on multi-source data coupling as described in claim 6, characterized in that, The preliminary estimation unit inputs the multi-dimensional data set into the multi-phase pollution source tracing model to obtain preliminary estimates of the contributions of different pollution sources to PM2.5 and ozone concentrations at each monitoring point, including: The multi-dimensional dataset is converted to the input format of the multi-phase co-occurring pollution source tracing model; Determine the parameters required for the multiphase co-existing pollution source tracing model; After completing the input format conversion and determining the required parameters, the multi-dimensional data set is input into the multi-phase co-existing pollution source tracing model.

9. The PM2.5 and ozone co-source analysis system based on multi-source data coupling as described in claim 8, characterized in that, In the preliminary estimated value determination unit, the multiphase co-occurring pollution source tracing model is expressed as follows: ； in Let t be the simulated concentration of pollutant of type k at the j-th altitude layer of the i-th monitoring point. For a fixed total number of sources, For the total number of mobile sources, for The combined emission intensity of the m-th stationary source and the n-th mobile source at time m. Let be the transport delay time of pollutants from the combination of the m-th stationary source and the n-th mobile source to the j-th altitude layer of the i-th monitoring point. Let be the emission diffusion coefficient of pollutant type k at the j-th altitude layer at the i-th monitoring point, for the combination of the m-th stationary source and the n-th mobile source. Let be the geographic attenuation factor at time t for the combination of the m-th fixed source and the n-th mobile source at the j-th altitude layer of the i-th monitoring point. Let be the atmospheric chemical transformation coefficient of pollutant type k at time t. The time represents PM2.

5. The time represents ozone.

10. The PM2.5 and ozone co-source analysis system based on multi-source data coupling as described in claim 9, characterized in that, The pollution source contribution value determination unit corrects the preliminary estimate based on the dual-pollutant cross-sensitivity model, and obtains the corrected pollution source contribution value through the following sub-steps: Preliminary estimates of PM2.5 and ozone concentrations at each monitoring point were extracted from the multiphase pollution source tracing model. At the same time, the actual monitored concentrations of PM2.5 and ozone at the corresponding monitoring points in the same time series were extracted to establish a comparison dataset of preliminary estimates and actual monitored concentrations. Based on the comparative dataset, the Pearson correlation coefficient and partial correlation coefficient between PM2.5 and ozone concentrations were calculated to determine the degree of linear correlation between the concentration changes of the two pollutants. At the same time, the synergistic change patterns of the concentrations of the two pollutants under different temperature and humidity conditions were analyzed to identify the calibration meteorological threshold of chemical reaction coupling effect. The degree of linear correlation, the law of coordinated change, and the calibrated meteorological threshold are input into the dual-pollutant cross-sensitivity model. The cross-sensitivity coefficient of the dual-pollutant cross-sensitivity model is adjusted, and the correction amount of PM2.5 and ozone concentration is output.

Images