Atmospheric pollutant-greenhouse gas collaborative traceability method and system based on receptor model

Abstract

The invention relates to the technical field of atmospheric environment monitoring, and particularly discloses an atmospheric pollutant-greenhouse gas collaborative traceability method and system based on a receptor model, and the method comprises the steps: quantifying the physical behavior difference of atmospheric pollutants and greenhouse gas in a diffusion process through synchronously monitoring the concentration and meteorological data of the atmospheric pollutants and the greenhouse gas; generating a multi-dimensional dynamic behavior difference index; and finally, utilizing the index and the collaborative monitoring data, executing iterative inversion and verification through an adaptive inversion system, generating a collaborative traceability result, and outputting a monitoring strategy optimization suggestion. According to the method, the defects of an independent analysis path and a preliminary cooperation path in a traditional traceability technology are overcome, the traceability accuracy and the result physical authenticity are improved by combining chemical constraint and physical verification, the method is particularly suitable for identification of unsteady and unregistered emission sources, the crossing of system intelligence is realized, and the system is suitable for popularization and application. The method has self-learning and evolutionary capabilities.

Description

Technical Field

[0001] This invention belongs to the field of atmospheric environment monitoring technology, and relates to a method and system for synergistic source tracing of atmospheric pollutants and greenhouse gases based on receptor models. Background Technology

[0002] Improving air quality and reducing greenhouse gas emissions are crucial global issues. Accurately identifying and quantifying emission sources—i.e., source apportionment—is a prerequisite for developing effective emission reduction strategies to achieve coordinated management of air pollutants and greenhouse gases. Receptor models are a key technique in atmospheric source apportionment; they analyze the chemical composition and concentration changes at atmospheric sampling points (receptors) to infer the contribution of various pollution sources to those points. With the increasing demand for synergistic pollution and carbon reduction, technologies for the coordinated source tracing of air pollutants and greenhouse gases have emerged.

[0003] Currently, two technical approaches are commonly used for source apportionment analysis of air pollutants and greenhouse gases. One is an independent analysis approach, which involves establishing separate receptor models for air pollutants such as particulate matter and source inventories or models for greenhouse gases, tracing the sources of the two types of substances separately, and then manually comparing the results. The other is a preliminary collaborative approach, which involves simultaneously observing pollutants and greenhouse gases at the monitoring end, and then, within the framework of a single receptor model, attempting to use greenhouse gases as a tracer or component, performing source apportionment calculations together with other pollutants.

[0004] However, the aforementioned existing technologies have significant shortcomings in practical applications. For independent analysis pathways, due to the fragmentation of monitoring data, model mechanisms, and assumptions, their respective source apportionment results often contradict each other in terms of spatial orientation and contribution estimation, making it difficult to form unified and consistent conclusions. For preliminary synergistic pathways, the traditional receptor models used are essentially static models based on the conservation of chemical components. These models treat greenhouse gases and air pollutants equally, failing to fully consider the inherent differences between them in physical diffusion behaviors such as deposition, chemical reactions, and long-distance transport. This results in the model lacking sufficient physical constraints during the inversion process, affecting the accuracy of source apportionment results. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a method and system for synergistic source tracing of atmospheric pollutants and greenhouse gases based on receptor model is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: The first aspect of this invention provides a method for synergistic tracing of atmospheric pollutants and greenhouse gases based on a receptor model, comprising: S1, acquiring a synergistic monitoring dataset of a target area, wherein the synergistic monitoring dataset includes atmospheric pollutant concentration data, greenhouse gas concentration data, and meteorological data obtained from synchronous monitoring.

[0007] S2. Based on the collaborative monitoring dataset, quantify the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process, and generate a multidimensional dynamic behavior difference index.

[0008] S3. Input the collaborative monitoring dataset and the multidimensional dynamic behavior difference index into the adaptive inversion system for collaborative tracing, and drive the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results.

[0009] S4. Based on the iterative inversion and verification process, generate monitoring strategy optimization suggestions and output collaborative tracing results and monitoring strategy optimization suggestions.

[0010] A second aspect of the present invention provides an atmospheric pollutant-greenhouse gas co-source tracing system based on a receptor model, comprising: a co-monitoring dataset acquisition module for acquiring a co-monitoring dataset of a target area, wherein the co-monitoring dataset includes atmospheric pollutant concentration data, greenhouse gas concentration data and meteorological data obtained from synchronous monitoring.

[0011] The dynamic behavior difference index generation module, based on the collaborative monitoring dataset, quantifies the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process, and generates a multidimensional dynamic behavior difference index.

[0012] The collaborative tracing result generation module takes the collaborative monitoring dataset and the multidimensional dynamic behavior difference index into the adaptive inversion system used for collaborative tracing, and drives the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results.

[0013] The optimization suggestion generation and output module generates monitoring strategy optimization suggestions based on the iterative inversion and verification process, and outputs collaborative tracing results and monitoring strategy optimization suggestions.

[0014] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention introduces and quantifies the difference in physical behavior between atmospheric pollutants and greenhouse gases during the diffusion process, and uses it as a verification dimension independent of chemical composition to construct a source tracing system that combines chemical constraints and physical verification. This system enables the source tracing conclusions to be verified by real physical processes, thereby improving the accuracy of identification of complex emission sources, especially non-steady-state and unregistered emission sources, and the physical authenticity of the results, which helps to reduce the ambiguity and uncertainty of traditional single-dimensional information source tracing methods.

[0015] (2) By recording and analyzing the iterative inversion process, this invention can assess the uncertainty of the tracing results and generate suggestions for optimizing the monitoring strategy based on this. This working mode from analysis to decision-making to optimization enables the entire monitoring and tracing system to have adaptive adjustment capabilities, and can guide the allocation of monitoring resources according to the actual tracing situation, thereby helping to improve the overall performance in long-term operation and improving the automation and adaptability of the system.

[0016] (3) The present invention adopts a two-stage inversion strategy of chemical constraint pre-screening and physical behavior verification, which helps to reduce the number of candidate hypotheses entering complex physical simulations and improves the computational efficiency of source tracing analysis. At the same time, by introducing a model adaptive correction mechanism, when all preset hypotheses cannot explain the observed facts, the system can generate new hypotheses and iterate, which makes it not only able to handle known emission sources, but also applicable to unknown and dynamically changing emission scenarios such as accidental emissions and illegal discharges, showing good robustness. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the method steps of the present invention.

[0019] Figure 2 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 The first aspect of the present invention provides a method for synergistic tracing of atmospheric pollutants and greenhouse gases based on a receptor model, comprising: S1, acquiring a synergistic monitoring dataset of a target area, wherein the synergistic monitoring dataset includes atmospheric pollutant concentration data, greenhouse gas concentration data and meteorological data obtained by synchronous monitoring.

[0022] S2. Based on the collaborative monitoring dataset, quantify the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process, and generate a multidimensional dynamic behavior difference index.

[0023] In a specific embodiment of the present invention, the specific steps for quantifying the differences in physical behavior between air pollutants and greenhouse gases during the diffusion process and generating a multidimensional dynamic behavior difference index include: identifying and tracking the same transported air mass within the target area based on meteorological data in the collaborative monitoring dataset.

[0024] It should be noted that, to generate a dynamic behavioral difference index that can be used to validate and drive the inversion model, the system performs a series of sophisticated engineering steps. The engineering objective is to extract and quantify behavioral differentiation signals of atmospheric pollutants and greenhouse gases in real atmospheric transport due to differences in their physicochemical properties, from synchronous monitoring data. The first step in this process is to identify and track the same transported air mass within the target area. The engineering goal is to pinpoint a consistent research object for subsequent difference analysis, namely, the spatial trajectory of gases within the same envelope. The system first calls upon meteorological data from the collaborative monitoring dataset, particularly three-dimensional wind field data from Doppler lidar or dense ground station networks. Using a trajectory tracking algorithm based on a Lagrange particle model, the system releases a virtual particle at a specific location and time of an upwind monitoring station and iteratively calculates the particle's trajectory in three-dimensional space at a fixed time step, such as 5-15 minutes, based on real-time wind field data. This trajectory represents the movement path of the same transported air mass. The system then searches along the trajectory to identify other monitoring stations and their corresponding time points that exist on the path or within a preset spatial tolerance, such as a radius of 100-500 meters, thereby establishing a spatiotemporally correlated monitoring point sequence.

[0025] Based on the collaborative monitoring dataset, atmospheric pollutant concentration sequences and greenhouse gas concentration sequences corresponding to the same transported air mass at different spatial locations were extracted.

[0026] It should be noted that the second step of this process is to extract the atmospheric pollutant concentration sequences and greenhouse gas concentration sequences corresponding to the air mass flowing through different spatial locations. The engineering goal is to construct a parallel dataset directly used for difference calculations. Based on the spatiotemporally correlated monitoring point sequence generated in the previous step, the system performs an index query in the collaborative monitoring dataset. For each point in the sequence, the system accurately extracts the atmospheric pollutant concentration data and greenhouse gas concentration data for the corresponding time point, for example... and The system generates two physically corresponding sequences of the same length by aggregating these concentration values ​​in sequential order: an air pollutant concentration sequence and a greenhouse gas concentration sequence.

[0027] The decay synchronization deviation, spatial distribution deformation vector, and temporal profile asymmetry of the atmospheric pollutant concentration series and the greenhouse gas concentration series are calculated, and the decay synchronization deviation, spatial distribution deformation vector, and temporal profile asymmetry are used together to form a multidimensional dynamic behavior difference index.

[0028] It should be noted that the third step in this process is to calculate the attenuation synchronization bias, spatial distribution deformation vector, and time profile asymmetry, and combine them into a final multidimensional dynamic behavior difference index. The engineering goal is to provide a more comprehensive and refined mathematical quantification of the behavioral differences between the two types of substances. To calculate the attenuation synchronization bias, the system first normalizes the atmospheric pollutant concentration series and the greenhouse gas concentration series, typically by dividing each value by the initial value of the series to eliminate the influence of absolute concentration differences and focus on comparing attenuation rates. Attenuation Synchronization Bias It can be calculated using the following formula: ,in, It is the first in the normalized atmospheric pollutant concentration sequence The value of each point, It is the first in the normalized greenhouse gas concentration series The value of each point, It is the length of the sequence, i.e., the total number of monitoring points. It is a dimensionless scalar. The larger its value, the more significant the difference in concentration decay behavior between pollutants and greenhouse gases during transport. This may indicate that there is a significant sedimentation or chemical transformation process of the pollutants.

[0029] To calculate the spatially distributed deformation vector, the system utilizes synchronized data from all stations in the collaborative monitoring dataset at a specific time snapshot. Using spatial interpolation algorithms such as Kriging interpolation, a continuous two-dimensional concentration field of atmospheric pollutants and greenhouse gases within the target area is generated, with a spatial grid resolution typically set to 50-200 meters. The system defines a characteristic contour line for each concentration field, such as the 50% peak concentration contour line. Subsequently, the geometric centroid coordinates of the area enclosed by these two contour lines are calculated. Spatial distribution deformation vector. This can be expressed by the following formula: ,in, These are the centroid coordinates of the characteristic contour lines of the atmospheric pollutant concentration field. These are the centroid coordinates of the characteristic contour lines of the greenhouse gas concentration field. and It is the unit vector in the Cartesian coordinate system. It is a vector with direction and magnitude. Its direction indicates the direction of the pollutant plume's offset relative to the more inert greenhouse gas plume, while its magnitude quantifies the degree of offset. This directly reflects the combined effects of local wind shear, asymmetric subsidence, or near-surface source influences.

[0030] To compute temporal profile asymmetry, the system extracts high temporal resolution (e.g., 1 minute) time series of atmospheric pollutant concentrations and greenhouse gas concentrations for each monitoring station during a pollution event (e.g., a full period of time when concentrations are above background levels). Due to pollutants (such as...) There is dry and wet deposition, and its concentration curve is usually higher than that of inert greenhouse gases (such as...) after the peak. The decay is faster, and the waveform is "sharper." The system uses continuous wavelet transform to analyze the two time series to quantify this morphological difference. Time profile asymmetry. It can be defined by the following formula: ,in, and These are the wavelet transform coefficients of the time series concentrations of air pollutants and greenhouse gases. For scale parameters, These are the translation parameters. It is a dimensionless scalar; the larger its value, the more significant the morphological difference between the two concentration time profiles, which directly reflects the intensity of local rapid sedimentation or chemical transformation of pollutants. When calculating time profile asymmetry, the system needs to obtain the wavelet transform coefficients. The coefficients are obtained by convolving the original concentration time series signal with a scaled and translated wavelet mother function. Specifically, the wavelet mother function is a waveform with oscillatory properties that decays rapidly to zero. Scale parameter The "width" of the wavelet mother function is determined. During analysis, the system starts with a small scale value and gradually increases it to a large scale value, performing multi-scale scanning to capture changes in the signal at different time scales. Translation parameters. This determines the position of the wavelet mother function on the time axis. The system will smoothly move the wavelet mother function along the entire time series from beginning to end, at each time point. All are calculated once. Therefore, the wavelet transform coefficients It is a two-dimensional matrix, where each element represents the value of the original signal at a given time point. Nearby, in The energy intensity at this specific scale (frequency) or its similarity to the wavelet mother function fully characterizes the time-frequency properties of the signal.

[0031] Finally, the system will calculate the scalar Vector and scalar These are combined into a structured data object, namely the multidimensional dynamic behavior difference index, which serves as a key input to the subsequent adaptive collaborative inversion model.

[0032] S3. Input the collaborative monitoring dataset and the multidimensional dynamic behavior difference index into the adaptive inversion system for collaborative tracing, and drive the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results.

[0033] In a specific embodiment of the present invention, the specific steps of driving the adaptive inversion system to perform iterative inversion and verification and generate collaborative source tracing results include: obtaining an initial source tracing hypothesis set containing multiple source emission characteristics, and establishing collaborative constraint conditions for emission ratios based on the physicochemical properties of source types.

[0034] It should be noted that the system first loads a preset set of initial source tracing hypotheses. This set is a database containing all known or potential emission sources within the region, such as factories and power plants, along with their corresponding emission composition spectra, geographical coordinates, and typical emission intensities. The establishment of emission ratio synergistic constraints based on the physicochemical properties of source types means that when generating the initial source tracing hypotheses, the emissions of various pollutants are not combined in isolation or randomly. Instead, the inherent and relatively stable physicochemical ratio relationships between the characteristic species of different pollution sources at the time of emission are used to construct a set of scientifically based mathematical constraint equations. This significantly reduces the invalid solution space and improves the efficiency and accuracy of source tracing. For example, when analyzing PM2.5 sources, it is known that typical coal-fired sources, while emitting PM2.5, also emit a large amount of other pollutants. ,That The mass ratio of NOx to PM2.5 typically falls within a relatively fixed range; similarly, when vehicle exhaust sources emit nitrogen oxides (NOx) and PM2.5, the NOx / PM2.5 ratio also exhibits significant source-type characteristics. Therefore, this method establishes a "source composition spectral database" beforehand, and then uses these emission ratios based on the physicochemical properties of the source type (such as...) (PM2.5, NOx / PM2.5, etc.) are used as co-constraints. When generating each candidate source tracing hypothesis (i.e., a set of potential pollution source emission inventories), the system automatically checks whether the emissions of multiple pollutants in the hypothesis simultaneously satisfy the characteristic proportional relationships of their corresponding source types. Any emission combination that does not satisfy these inherent proportional constraints will be directly eliminated or assigned a very low initial weight, ensuring the scientific rationality of the initial hypothesis set.

[0035] By combining the emission ratio synergistic constraint, the initial set of source tracing hypotheses is screened by chemical composition to obtain candidate source tracing hypotheses.

[0036] In a specific embodiment of the present invention, the step of combining emission ratio synergistic constraints to screen the initial source tracing hypothesis set for chemical composition and obtain candidate source tracing hypotheses includes: obtaining source composition spectrum information corresponding to each source tracing hypothesis, and calculating the theoretical ratio range corresponding to each source tracing hypothesis based on the emission ratio synergistic constraints.

[0037] Extract the actual concentration increment ratio from the collaborative monitoring dataset.

[0038] Each theoretical ratio range is compared with the actual concentration increment ratio, and the source tracing hypothesis that the theoretical ratio range does not cover the actual concentration increment ratio is eliminated to obtain candidate source tracing hypotheses.

[0039] It should be noted that the first step is to obtain the source composition spectrum information and initial emission intensity information corresponding to each source hypothesis in the initial source hypothesis set. The purpose is to prepare basic data for subsequent theoretical ratio calculations. The system traverses the loaded initial source hypothesis set. For each source hypothesis representing a potential pollution source, the system reads two core parameter sets from its attribute database: one is the source composition spectrum information, which is usually a list containing the relative emission ratios or emission factors of various air pollutants and greenhouse gases emitted by that source type. For example, the composition spectrum of a coal-fired power plant hypothesis would indicate the emission intensity per ton of coal burned. , , and The first step involves determining the mass ratio of the target pollutant to the target greenhouse gas. The second step is to calculate the theoretical range of the simulated pollutant concentration to simulated greenhouse gas concentration for each source hypothesis, based on the emission ratio coordination constraint. The engineering goal is to transform the inherent chemical properties of each source hypothesis into a quantifiable indicator that can be directly compared with external observations. The system uses the source composition spectrum information obtained in the previous step to directly calculate the emission mass ratio of the target pollutant to the target greenhouse gas. For example, if the composition spectrum indicates an emission of 1 kg... It will emit 50 kilograms at the same time. The theoretical emission mass ratio is 1:50. Considering actual combustion efficiency and operating condition fluctuations, the emission ratio coordination constraint is usually not a fixed value, but a reasonable fluctuation range, for example, allowing fluctuations of 15%-30% above and below the theoretical ratio. Therefore, the system calculates a theoretical ratio range for each source tracing hypothesis. The third step is to eliminate source tracing hypotheses that do not conform to the theoretical ratio range based on the actual concentration ratio, in order to generate candidate source tracing hypotheses. The purpose is to perform the core screening action and complete the chemical constraint filtering. The system first extracts atmospheric pollutant concentration and greenhouse gas concentration data from all monitoring points covering the entire target area and time window from the coordinated monitoring dataset, and calculates the incremental concentration ratio after enhancing the background value, which is more representative of the contribution of local source emissions. Then, the system compares this measured incremental concentration ratio one by one with the theoretical ratio range calculated for each source tracing hypothesis in the previous step. If the theoretical ratio range of a source tracing hypothesis does not include the measured incremental concentration ratio at all, meaning the measured value falls outside the theoretical ratio range, the system determines that the source tracing hypothesis is inconsistent with the observed facts in terms of chemical composition and removes it from the candidate list. After traversing and judging all hypotheses in the initial source tracing hypothesis set, all the source tracing hypotheses that were not removed, i.e., whose theoretical ratio range can encompass the measured concentration ratio, together constitute the candidate source tracing hypothesis set, which enters the subsequent more refined physical behavior verification stage.

[0040] For each candidate source tracing hypothesis, an atmospheric diffusion process simulation is performed to obtain the corresponding simulated concentration field, and the corresponding simulated multidimensional behavior difference index is calculated based on the simulated concentration field.

[0041] In a specific embodiment of the present invention, the step of simulating the atmospheric diffusion process for each candidate source tracing hypothesis to obtain the corresponding simulated concentration field includes: constructing a virtual source parameter set corresponding to each candidate source tracing hypothesis, wherein the virtual source parameter set includes source location, source strength and emission time characteristics.

[0042] It should be noted that, to construct a digital behavioral twin of each candidate source tracing hypothesis under real atmospheric conditions, the system needs to perform accurate atmospheric diffusion process simulations to generate its corresponding simulated concentration field. The purpose of this process is to transform abstract source emission parameters into concrete concentration distributions that can be directly compared with measured data in three-dimensional space and time, providing foundational data for subsequent physical behavior difference verification. The first step in this process is to construct a virtual source parameter set. The aim is to transform the information of a candidate source tracing hypothesis into a standardized input format recognizable by the atmospheric diffusion model. For a selected candidate source tracing hypothesis, the system first extracts key source emission information from its data structure. This includes the source's geographical coordinates, such as latitude and longitude, its altitude, pollutant emission intensity, greenhouse gas emission intensity, the physical height and outlet diameter of the chimney, and the temperature and rise rate of the emitted flue gas. Emission intensity is typically measured in grams per second or tons per hour. In addition, the system also extracts the temporal characteristics of the emissions, such as whether they are continuous emissions or follow a specific diurnal or weekly variation pattern. The system integrates these parameters into a structured data file or in-memory object, i.e., the virtual source parameter set.

[0043] A hybrid Lagrange-Euler coupled model was obtained to simulate atmospheric diffusion, and real-time three-dimensional wind field data and virtual source parameter set from the collaborative monitoring dataset were used as inputs to the hybrid Lagrange-Euler coupled model.

[0044] Drive a hybrid Lagrange-Euler coupled model to simulate the diffusion process of atmospheric pollutants and greenhouse gases emitted from virtual sources under the action of a real-time three-dimensional wind field, and output a simulated concentration field.

[0045] It should be noted that the second step in this process involves calling and configuring the input of a hybrid Lagrange-Euler coupled model. The aim is to utilize an advanced model that combines accuracy and efficiency to provide a high-quality simulation field for the subsequent accurate calculation of physical fingerprints. Traditional single models, such as Gaussian models, are poorly adapted to complex terrain, while pure Lagrange models are inefficient in calculating regional concentration fields. The hybrid Lagrange-Euler coupled model used in this invention works as follows: In the near-field region of the source (e.g., within 5 km downwind), the model employs a Lagrange particle model to accurately track the three-dimensional trajectories of thousands of virtual particles representing pollutant mass, thus finely characterizing the plume's rise and initial diffusion morphology. When these particles move to the far-field region, their carried mass is smoothly released and mapped onto a fixed Euler grid. This coupling method ensures both the accuracy of the source region simulation and the efficiency of regional-scale calculations, making it particularly suitable for the needs of this invention, which requires accurate simulation of spatial deformation and concentration profiles. The third step in this process is to run the model and output the simulated concentration field. The goal is to complete the core physical simulation calculations and obtain the final concentration distribution results. The system starts the selected atmospheric diffusion model, which internally calculates, based on the principles of fluid dynamics and atmospheric physics, how pollutants and greenhouse gases emitted from the virtual source undergo advection transport and turbulent diffusion under the influence of the input real-time three-dimensional wind field. For non-inert substances such as particulate matter, the model also considers parameters such as their settling rate and possible chemical transformation rate in the air. The model iterates at a fixed time step, such as 1-10 minutes, until the entire source tracing time window is covered. After the calculation is completed, the model outputs a four-dimensional data matrix, namely the simulated concentration field. The four dimensions of this matrix represent the x-axis, y-axis, z-axis of geographic coordinates, and the time axis, respectively. The value of each element in the matrix represents the simulated concentration of atmospheric pollutants and greenhouse gases contributed by the virtual source at a specific spatial point and time. This simulated concentration field provides a complete data foundation for subsequent calculations of the simulated behavior difference index.

[0046] It should also be noted that a simulated multidimensional behavior difference index is calculated for each simulated concentration field. The purpose is to reduce the dimensionality of the high-dimensional simulated concentration field data generated in the previous step and transform it into a quantitative index with the same structure and physical meaning as the measured dynamic behavior difference index, thus creating conditions for the final comparison. For each simulated concentration field, the system performs the same calculation steps as in step S2 above, namely, by tracking virtual air masses, extracting concentration sequences, and performing normalization processing, it calculates the simulated attenuation synchronization deviation, simulated spatial distribution deformation vector, and simulated time profile asymmetry under the simulated scenario, and combines them into the simulated multidimensional behavior difference index of the candidate source tracing hypothesis.

[0047] The physical behavior of each simulated multidimensional behavior difference index is compared with that of the multidimensional dynamic behavior difference index, and the candidate tracing hypothesis with the highest matching degree is selected as the collaborative tracing result.

[0048] It should be noted that the simulated multidimensional behavioral difference indices are compared with the measured multidimensional dynamic behavioral difference indices to select the candidate source hypothesis with the highest matching degree. The aim is to ultimately determine the emission source closest to reality from multiple candidates by evaluating the similarity of physical behavior. The system uses a cost function to quantify the matching degree between simulation and reality. Matching error score. It can be calculated using the following formula: ,in, , , It simulates attenuation synchronization deviation, spatial distribution deformation, and time profile asymmetry. , , These are dimensionless weighting factors, and their sum is 1. , , It is the normalization factor, usually taken as 1. , , The standard deviation in historical samples is used to eliminate the influence of different indicator units and numerical ranges. The system calculates the corresponding standard deviation for each candidate source hypothesis. Value. Ultimately, The candidate hypothesis with the smallest value, that is, the one whose simulated physical diffusion behavior best matches the actual observation, is determined by the system to be the final collaborative source tracing result.

[0049] In one specific embodiment of the present invention, , , The weighting factor is not fixed but can be adaptively adjusted according to meteorological conditions. For example, in a scenario of atmospheric stability and long-distance transport, the attenuation synchronization deviation... More importantly, the system can automatically improve. The weight of spatial distribution; while in the complex and variable wind field environment of urban canyons, the spatial distribution deformation and time profile asymmetry It better reflects local effects, and the system will improve accordingly. and The weights are determined by the specific adjustments made based on a pre-defined lookup table that takes atmospheric stability level and wind shear index as input, thus ensuring the scientific rigor and scenario adaptability of the evaluation system.

[0050] In a specific embodiment of the present invention, after comparing the physical behavior of each simulated multidimensional behavior difference index with the multidimensional dynamic behavior difference index, the method further includes: defining a credibility threshold for determining the matching degree, and determining whether the matching degree of all candidate source tracing hypotheses is lower than the credibility threshold.

[0051] If so, the deviation characteristics between the multidimensional dynamic behavior difference index and each simulated multidimensional behavior difference index are input into the source parameter intelligent inference engine to generate a probability distribution for the unknown source parameter.

[0052] Based on the probability distribution of the unknown source parameters, new source tracing hypothesis constraints are generated.

[0053] Update the initial set of source tracing hypotheses with the new source tracing hypothesis constraints, and return to perform the step of screening the chemical composition of the initial set of source tracing hypotheses.

[0054] It should be noted that, to ensure the robustness and adaptability of the collaborative source tracing results, a model correction and iteration mechanism based on intelligent reasoning was designed. The engineering purpose of this mechanism is that, when all preset source hypotheses fail to reasonably explain the observed physical fingerprints, the system can learn from these failures and directly predict the core parameters of the unknown source, thereby achieving efficient and accurate breakthroughs in unknown emission scenarios. The first step of this mechanism is to determine whether the matching degree of all candidate source tracing hypotheses is lower than a preset confidence threshold. The purpose is to establish a clear trigger condition to initiate model correction. This is achieved by completing the matching error scoring of all candidate source tracing hypotheses. After the calculation, the system will assign each The value is compared to a preset confidence threshold. This confidence threshold is typically set through historical data analysis or expert experience, reflecting the maximum acceptable model observation discrepancy for the system. If all calculated values... If all values ​​exceed the credibility threshold (e.g., scores generally exceed a normalized credibility threshold of 0.7), the system determines that none of the hypotheses in the initial source tracing hypothesis set can effectively explain the measured dynamic behavior difference index, and then triggers the model correction mechanism. The second step of this mechanism is to activate the source parameter intelligent inference engine to generate a probability distribution for the unknown source parameters. This engine is a pre-trained neural network model using deep learning. Its training data consists of tens of thousands of "physical fingerprint-source parameter" paired samples simulated by a high-precision diffusion model. During actual operation, the engine's input is the deviation vector between the measured multidimensional dynamic behavior difference index and all simulated indices when the current inversion fails. The engine's output is no longer a simple rule, but rather one or more probability density function graphs, including: 1) an unknown source location probability map: a two-dimensional geographic map where the color intensity of each pixel represents the probability of an unknown source existing at that location. 2) an unknown source intensity probability distribution: a one-dimensional curve indicating the range in which the emission intensity of the unknown source is most likely to fall. 3) an unknown source type suggestion: based on bias characteristics, it provides the most likely source types and their confidence levels. The third step of this mechanism is to generate new source tracing hypothesis constraints based on the probability distribution. The system analyzes the above probability distribution graph and generates a set of highly focused new source hypotheses within the regions with the highest probability (e.g., highlighted areas on the location probability map) and the most likely parameter ranges. These hypotheses constitute new constraints used to update the initial source tracing hypothesis set. For example, the system will generate new virtual source points at high resolution (e.g., 100 meters) within regions where the location probability is greater than 90%, and assign them inferred intensity and type. Finally, the initial source tracing hypothesis set is updated using the new source tracing hypothesis constraints, and the source tracing process is re-executed. The goal is to apply the newly generated knowledge to the actual inversion search, completing closed-loop optimization. The system applies the new source hypothesis constraints generated in the previous step to the generator of the initial source hypothesis set. This may mean generating a new batch of potential source hypothesis points in the existing spatial grid according to the new location constraints; or adding a new source component spectrum template to the source type library. After the update, the system forms a new, expanded, and optimized initial source hypothesis set. Subsequently, system control automatically returns to the preliminary screening step in step S3, thus starting a completely new, more likely to successful source iteration loop. This closed loop of "judgment-intelligent reasoning-correction-re-execution" endows the system with a powerful ability to directly generate new knowledge from data, approaching the intuition of human experts.

[0055] S4. Based on the iterative inversion and verification process, generate monitoring strategy optimization suggestions and output collaborative tracing results and monitoring strategy optimization suggestions.

[0056] In a specific embodiment of the present invention, the specific steps for generating monitoring strategy optimization suggestions based on the iterative inversion and verification process include: extracting the hypothesis matching degree data and model correction records that may occur during the iteration process.

[0057] It should be noted that, to ensure the source tracing method transforms from an open-loop analysis tool into an intelligent closed-loop system with self-evaluation and optimization capabilities, a rigorous backend processing flow was designed to output collaborative source tracing results and generate monitoring strategy optimization suggestions. The purpose of this flow is to transform complex inversion process information into intuitive and usable decision-making basis for managers, i.e., quantifying the credibility of the results and proactively proposing specific action plans to improve future source tracing accuracy. The first step of this flow is to extract hypothesis matching data and model correction records. The aim is to collect all key information reflecting the internal decision-making process from the operational logs of the adaptive collaborative inversion model. After the inversion task is completed, the system accesses the model's internal storage or log files to systematically extract the matching error scores corresponding to all candidate source tracing hypotheses in each iteration. Furthermore, if model corrections occur, the system will also fully extract the deviation feature analysis results that triggered the correction, as well as the new source hypothesis constraints generated accordingly. These data together constitute hypothesis matching data and model correction records, serving as the raw material for evaluating the quality of the entire inversion process.

[0058] Based on the hypothesis matching data and model correction records, the uncertainty quantification score of the collaborative tracing results is calculated.

[0059] It should be noted that the second step in this process is to calculate the uncertainty quantification score of the collaborative tracing results. The engineering goal is to attach a scientific and quantitative credibility index to the final tracing conclusions, avoiding simplistic black-and-white judgments. Based on the data extracted in the previous step, the system uses a comprehensive evaluation algorithm to calculate the final uncertainty quantification score. A feasible calculation formula is as follows: ,in, It is the matching error score of the hypothesis that is ultimately selected as the collaborative tracing result, i.e., the one with the smallest score. value. It is the largest among all candidate origination hypotheses. The value is used as a normalization benchmark. It refers to the number of model corrections the system undergoes before obtaining the final result. This is the preset maximum allowed number of corrections, used to normalize the impact of the number of corrections. The design logic of this formula is that the smaller the matching error of the final result, and the fewer corrections are required to reach that result, the higher the uncertainty quantification score. The higher the value, the more reliable the result. The value ranges from 0 to 1; for example, a score of 0.9 or higher represents a highly reliable result.

[0060] By combining the spatial distribution information and uncertainty quantification score of the collaborative tracing results, suggestions for optimizing the monitoring strategy are generated.

[0061] It should be noted that the third step of this process involves combining spatial distribution information and uncertainty quantification scores to generate monitoring strategy optimization suggestions. The aim is to combine abstract scores with specific geographic space to propose actionable monitoring optimization solutions that address the current system's shortcomings. The system first marks the collaborative source tracing results—that is, the locations of identified pollution sources—on a geographic information system. Subsequently, the system visualizes the locations of candidate source tracing hypotheses with low matching degrees during the inversion process but which cannot be completely ruled out, as well as "suspicious" areas indicated during model corrections, using different symbols or colors on the map. Specifically, for areas with low uncertainty quantification scores, the system analyzes the reasons for the low scores, such as whether it's due to sparse monitoring stations nearby leading to a low signal-to-noise ratio in the dynamic behavior difference index, or due to high uncertainty in wind field simulation. Based on this diagnosis, the system generates specific monitoring strategy optimization suggestions. For example, if the source tracing results for a certain area are highly uncertain and monitoring points are sparse, the system may suggest "adding a set of collaborative monitoring fixed stations on the key downwind transmission channel of the area"; if intermittent illegal discharge sources are suspected, it may suggest "using drones equipped with portable monitoring equipment to conduct high-density mobile patrols of the area during specific time periods, such as at night." These suggestions, in the form of text or instructions, constitute part of the final output.

[0062] In a specific embodiment of the present invention, before obtaining the collaborative monitoring dataset of the target area, the method further includes: receiving and parsing the source tracing task instruction input by the user, wherein the source tracing task instruction includes the target area range and the target time window.

[0063] Configure the parameters for acquiring the collaborative monitoring dataset according to the source tracing task instructions.

[0064] Based on the acquired parameters, the monitoring network is activated to complete the collection of the collaborative monitoring dataset.

[0065] It should be noted that this invention employs a pre-configuration task and data preparation process before executing the core collaborative source tracing algorithm. The purpose of this process is to precisely translate the user's macro-level requirements into specific, executable task parameters, ensuring that all necessary basic data is ready before the source tracing begins, thereby guaranteeing the relevance and efficiency of the entire subsequent analysis process. The first step of this process is to receive and parse the source tracing task instructions input by the user. This aims to provide initial, top-level guidance information for the entire source tracing task. The system receives the source tracing task instructions input by the operator through a graphical user interface or application programming interface. This instruction is a structured data object whose core fields must include: the target area, typically defined by a geographic coordinate polygon, such as the administrative boundary of an industrial park or city; the target time window, defining the start and end times of interest for the source tracing analysis, such as from 8:00 AM to 6:00 PM on a certain day; and the target pollutant and target greenhouse gas types to be traced, for example, specifying... , For the target pollutant, The system's internal parser performs format validation and legality checks on these instructions to ensure that the input parameters are complete and within the system's processing range. The second step of this process is to configure the acquisition parameters for the collaborative monitoring dataset based on the source tracing task instructions. The engineering goal is to translate high-level user instructions into specific parameters that the underlying data acquisition and processing modules can understand and execute. Based on the parsed source tracing task instructions, the system automatically generates a series of configuration parameters. For example, based on the target area range in the instructions, the system filters out all monitoring stations geographically located within that area or its surrounding buffer zone and generates a list of station IDs. Based on the target time window, the system sets the start and end timestamps for the data query. Based on the specified target pollutant and target greenhouse gas type, the system generates a list of specific monitoring indicator fields that need to be extracted from each station's database. These parameters are integrated into an acquisition parameter set as input for the next step of data acquisition. The third step of this process is to start the monitoring network for data acquisition. The engineering goal is to actually retrieve the required data from various data sources according to the configured parameters and aggregate them into a unified collaborative monitoring dataset. The system distributes the acquisition parameter set generated in the previous step to the data acquisition module. Based on the list of site IDs, this module initiates data requests to the local data servers or central cloud platform of each monitoring station. The requests include precise timestamp ranges and the required indicator fields. Upon receiving the data streams from each station, the data acquisition module performs preliminary quality control procedures, such as removing invalid values ​​or aligning timestamps, to ensure that data from all sources are synchronized in time. Finally, all processed atmospheric pollutant concentration data, greenhouse gas concentration data, and synchronized meteorological data are integrated and stored in a unified format data file or in-memory database—the collaborative monitoring dataset—providing a complete and reliable data foundation for subsequent quantitative analysis and inversion modeling.

[0066] In a specific embodiment of the present invention, the method further includes: obtaining historical emission inventories or historical source tracing results as benchmark data.

[0067] The results of collaborative source tracing are compared and analyzed in time and space with the baseline data, and changes in emission hotspots or new potential emission sources are identified from the results of the time and space comparison analysis.

[0068] Generate early warning reports that include information on changes in emission hotspots or new potential emission sources, and output the early warning reports together with the results of collaborative source tracing.

[0069] It should be noted that, in addition to outputting the final source tracing conclusion, this invention also integrates a change detection and early warning function based on historical data. The purpose of this function is to proactively identify potential environmental risk changes, such as abnormal fluctuations in emission behavior or newly emerging unknown pollution sources, by comparing the dynamic results of this source tracing with static or historical benchmark data, thereby providing more forward-looking decision support for environmental regulation. The first step of this function is to compare and analyze the collaborative source tracing results with a preset historical emission inventory or historical source tracing results. The aim is to establish a reference system to highlight the "new" and "different" aspects of the current source tracing results. The system first loads a benchmark database, which can be an official historical emission inventory published by government departments covering the target area, containing the pollutant and greenhouse gas emissions of known enterprises in a past year; or it can be historical source tracing results obtained by this system operating under stable meteorological conditions in the past. The system matches and compares the collaborative source tracing results generated in this calculation, i.e., the contribution and spatial location of each identified source, with the corresponding entries in this benchmark database one by one. The second step of this function is to identify changes in emission hotspots or new potential emission sources. The goal is to accurately pinpoint significant regulatory changes from comparative discrepancies. The system employs a change detection algorithm to perform this step. For changes in emission hotspots, the system calculates the percentage difference between the contribution of a known source in the current collaborative source tracing results and the registered emissions of that source in the historical emission inventory. If this difference exceeds a preset significance threshold, such as a positive or negative difference exceeding 30%-50%, the system marks the source as a change in emission hotspot. For newly added potential emission sources, the system performs a spatial difference operation at the geographic information system level, comparing the set of all source locations identified in the current collaborative source tracing results with the set of source locations in the historical emission inventory. If a source location exists in the current results where no registered source can be found within a certain neighborhood of the historical inventory (e.g., 500 meters to 1 kilometer), the system marks that location as a newly added potential emission source. The third step of this function is to generate an early warning report. The engineering objective is to present the detected change information to users in a clear and operable format. The system summarizes and formats the information on all emission hotspot changes and newly added potential emission sources identified in the previous step. For each warning event, the report will detail whether it is a "hotspot change" or a "new source," its geographical coordinates, the types of pollutants and greenhouse gases involved, and the estimated emission contribution from this source tracing. For hotspot change events, the report will also include specific differences from historical data. This structured warning report will ultimately be packaged together with the main body of the collaborative source tracing results, providing managers with a complete information package that includes current status analysis and dynamic risk warnings.

[0070] Reference Figure 2The second aspect of the present invention provides a collaborative source tracing system for atmospheric pollutants and greenhouse gases based on a receptor model, comprising: a collaborative monitoring dataset acquisition module, a dynamic behavior difference index generation module, a collaborative source tracing result generation module, and an optimization suggestion generation and output module.

[0071] The collaborative monitoring dataset acquisition module and the dynamic behavior difference index generation module are connected. Both the collaborative monitoring dataset acquisition module and the dynamic behavior difference index generation module are connected to the collaborative tracing result generation module. The collaborative tracing result generation module is connected to the optimization suggestion generation and output module.

[0072] The collaborative monitoring dataset acquisition module acquires the collaborative monitoring dataset of the target area, which includes atmospheric pollutant concentration data, greenhouse gas concentration data, and meteorological data obtained from synchronous monitoring.

[0073] The dynamic behavior difference index generation module, based on the collaborative monitoring dataset, quantifies the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process, and generates a multidimensional dynamic behavior difference index.

[0074] The collaborative tracing result generation module inputs the collaborative monitoring dataset and the multidimensional dynamic behavior difference index into the adaptive inversion system used for collaborative tracing, and drives the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results.

[0075] The optimization suggestion generation and output module generates monitoring strategy optimization suggestions based on the iterative inversion and verification process, and outputs collaborative tracing results and monitoring strategy optimization suggestions.

[0076] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model, characterized in that, include: S1. Obtain the collaborative monitoring dataset of the target area, wherein the collaborative monitoring dataset includes atmospheric pollutant concentration data, greenhouse gas concentration data and meteorological data obtained from synchronous monitoring; S2. Based on the collaborative monitoring dataset, quantify the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process, and generate a multidimensional dynamic behavior difference index. S3. Input the collaborative monitoring dataset and multidimensional dynamic behavior difference index into the adaptive inversion system for collaborative tracing, and drive the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results; S4. Based on the iterative inversion and verification process, generate monitoring strategy optimization suggestions and output collaborative tracing results and monitoring strategy optimization suggestions.

2. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 1, characterized in that, The specific steps for quantifying the differences in physical behavior between air pollutants and greenhouse gases during the diffusion process and generating a multidimensional dynamic behavior difference index include: Based on meteorological data in the collaborative monitoring dataset, identify and track the same transported air mass within the target area; Based on the collaborative monitoring dataset, atmospheric pollutant concentration sequences and greenhouse gas concentration sequences corresponding to the same transported air mass at different spatial locations were extracted. The decay synchronization deviation, spatial distribution deformation vector, and temporal profile asymmetry of the atmospheric pollutant concentration series and the greenhouse gas concentration series are calculated, and the decay synchronization deviation, spatial distribution deformation vector, and temporal profile asymmetry are used together to form a multidimensional dynamic behavior difference index.

3. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 1, characterized in that, The specific steps by which the adaptive inversion system performs iterative inversion and verification to generate collaborative tracing results include: Obtain an initial set of source tracing hypotheses containing emission characteristics from multiple sources, and establish coordinated constraints on emission ratios based on the physicochemical properties of source types; By combining the emission ratio synergistic constraint condition, the initial source tracing hypothesis set is screened by chemical composition to obtain candidate source tracing hypotheses; For each candidate source tracing hypothesis, an atmospheric diffusion process simulation is performed to obtain the corresponding simulated concentration field, and the corresponding simulated multidimensional behavior difference index is calculated based on the simulated concentration field. The physical behavior of each simulated multidimensional behavior difference index is compared with that of the multidimensional dynamic behavior difference index, and the candidate tracing hypothesis with the highest matching degree is selected as the collaborative tracing result.

4. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 3, characterized in that, After comparing the simulated multidimensional behavior difference index with the multidimensional dynamic behavior difference index using physical behavior, the process also includes: Define a confidence threshold for determining the matching degree, and determine whether the matching degree of all candidate source tracing hypotheses is lower than the confidence threshold; If so, the deviation characteristics of the multidimensional dynamic behavior difference index and each simulated multidimensional behavior difference index are input into the source parameter intelligent inference engine to generate a probability distribution for the unknown source parameter. Based on the probability distribution of the unknown source parameters, new source tracing hypothesis constraints are generated; Update the initial set of source tracing hypotheses with the new source tracing hypothesis constraints, and return to perform the step of screening the chemical composition of the initial set of source tracing hypotheses.

5. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 3, characterized in that, The combined emission ratio constraint condition is used to screen the initial source tracing hypothesis set by chemical composition to obtain candidate source tracing hypotheses, including: Obtain the source composition spectrum information corresponding to each source tracing hypothesis, and calculate the theoretical ratio range corresponding to each source tracing hypothesis based on the emission ratio coordination constraint condition; Extract the actual concentration increment ratio from the collaborative monitoring dataset; Each theoretical ratio range is compared with the actual concentration increment ratio, and the source tracing hypothesis that the theoretical ratio range does not cover the actual concentration increment ratio is eliminated to obtain candidate source tracing hypotheses.

6. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 3, characterized in that, The atmospheric diffusion process simulation for each candidate source tracing hypothesis, to obtain the corresponding simulated concentration field, includes: Construct a virtual source parameter set corresponding to each candidate source tracing hypothesis, wherein the virtual source parameter set includes source location, source strength and emission time characteristics; A hybrid Lagrange-Euler coupled model was obtained to simulate atmospheric diffusion, and real-time three-dimensional wind field data and virtual source parameter set from the collaborative monitoring dataset were used as inputs to the hybrid Lagrange-Euler coupled model. Drive a hybrid Lagrange-Euler coupled model to simulate the diffusion process of atmospheric pollutants and greenhouse gases emitted from virtual sources under the action of a real-time three-dimensional wind field, and output a simulated concentration field.

7. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 1, characterized in that, The specific steps for generating monitoring strategy optimization suggestions based on iterative inversion and verification include: Extract the hypothesis matching degree data and model correction records that may occur during the iteration process during the iterative inversion and verification process; Based on the hypothesis matching data and model correction records, calculate the uncertainty quantification score of the collaborative tracing results; By combining the spatial distribution information and uncertainty quantification score of the collaborative tracing results, suggestions for optimizing the monitoring strategy are generated.

8. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 1, characterized in that, Before obtaining the collaborative monitoring dataset for the target area, the following steps are also included: Receive and parse the source tracing task instruction input by the user, the source tracing task instruction including the target area range and the target time window; Configure the parameters for acquiring the collaborative monitoring dataset according to the source tracing task instructions; Based on the acquired parameters, the monitoring network is activated to complete the collection of the collaborative monitoring dataset.

9. The method for synergistic source tracing of air pollutants and greenhouse gases based on a receptor model according to claim 1, characterized in that, Also includes: Obtain historical emission inventories or historical source tracing results as baseline data; The results of collaborative source tracing are compared and analyzed in time and space with the baseline data, and changes in emission hotspots or new potential emission sources are identified from the results of the time and space comparison analysis. Generate early warning reports that include information on changes in emission hotspots or new potential emission sources, and output the early warning reports together with the results of collaborative source tracing.

10. A collaborative source tracing system for atmospheric pollutants and greenhouse gases based on a receptor model, characterized in that, include: The collaborative monitoring dataset acquisition module acquires the collaborative monitoring dataset of the target area, which includes atmospheric pollutant concentration data, greenhouse gas concentration data, and meteorological data obtained from synchronous monitoring. The dynamic behavior difference index generation module, based on the collaborative monitoring dataset, quantifies the differences in the physical behavior of air pollutants and greenhouse gases during the diffusion process and generates a multidimensional dynamic behavior difference index. The collaborative tracing result generation module takes the collaborative monitoring dataset and the multidimensional dynamic behavior difference index into the adaptive inversion system used for collaborative tracing, and drives the adaptive inversion system to perform iterative inversion and verification to generate collaborative tracing results. The optimization suggestion generation and output module generates monitoring strategy optimization suggestions based on the iterative inversion and verification process, and outputs collaborative tracing results and monitoring strategy optimization suggestions.

Images