Accurate traceability method, device and equipment based on high-resolution air quality monitoring data and storage medium

Abstract

The invention provides an accurate traceability method, device and equipment based on high-resolution air quality monitoring data and a storage medium, and relates to the technical field of atmospheric pollution source tracking. The method comprises the steps of simulating a meteorological phenomenon from a weather scale to a regional scale in a real meteorological environment based on a meteorological field simulation model, and determining high-resolution meteorological field data; inputting the high-resolution meteorological field data into a Lagrange particle tracking random transmission model so as to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, and obtaining movement tracks of the virtual particles; and determining pollution source information in the real meteorological environment based on pollutant monitoring data of the monitoring point, preset emission data and the motion trail of the virtual particles. According to the embodiment of the invention, the defect of low accuracy of pollutant traceability in the prior art is overcome, and a more accurate pollution source positioning and traceability result can be provided through combination of high-resolution meteorological simulation and a random transmission model.

Description

Technical Field

[0001] The present invention relates to the field of atmospheric pollution source tracing technology, and in particular to a precise tracing method, device, equipment and storage medium based on high-resolution air quality monitoring data. Background Art

[0002] In recent years, air quality monitoring has become a critical concern in urban areas. Consequently, small sensor networks have been deployed across cities across my country. By establishing a multi-scale monitoring network, my country has developed a comprehensive air quality monitoring system. However, currently, the role of micro-stations is primarily focused on identifying high-concentration locations. Due to the large number of micro-stations and the diverse surrounding pollution sources, identifying pollution sources is extremely difficult.

[0003] Existing air pollution source tracing technologies suffer from several key flaws, limiting their effectiveness in practical applications. First, most existing technologies rely on low-spatial-resolution monitoring methods, typically only providing results over a range of a few to tens of kilometers. This is particularly insufficient for accurately tracking pollution sources. In urban environments, pollution sources are often concentrated and highly localized, making traditional monitoring equipment and models unable to provide accurate source tracing analysis at small scales.

[0004] Secondly, existing technologies often rely on a single data source for analysis, failing to effectively integrate multiple data sources, such as meteorological, emissions, and monitoring data. This results in low accuracy and reliability in tracing results. Furthermore, many existing tracing methods are weak at identifying small-scale pollution sources, such as low-emission sources like construction and the catering industry, making it impossible to fully identify the impact of pollution sources on air quality.

[0005] Therefore, how to accurately monitor air quality and accurately trace the source of pollutants is a technical problem that needs to be solved urgently. Summary of the Invention

[0006] The present invention provides a precise tracing method, device, equipment and storage medium based on high-resolution air quality monitoring data, which is used to address the low accuracy of pollutant tracing in the existing technology. By combining high-resolution meteorological simulation with random transmission models, it can provide more accurate pollution source positioning and tracing results.

[0007] In a first aspect, the present invention provides a precise tracing method based on high-resolution air quality monitoring data, comprising the following steps: Based on the meteorological field simulation model, simulate meteorological phenomena from the synoptic scale to the regional scale in the real meteorological environment and determine high-resolution meteorological field data; Inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; Based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particles, the pollution source information in the real meteorological environment is determined; wherein, the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0008] Preferably, according to a precise traceability method based on high-resolution air quality monitoring data provided by the present invention, the meteorological field simulation model simulates meteorological phenomena from weather scale to regional scale in a real meteorological environment to determine high-resolution meteorological field data, including: Simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to obtain meteorological background data at at least two spatial scales; the at least two spatial scales are characterized by spatial scales at multiple different levels from a first height to a second height; Inputting the meteorological background data of the at least two spatial scales into a trained downscaling neural network for processing, and outputting the high-resolution meteorological field data; Among them, the trained downscaling neural network is obtained by performing multiple rounds of training processing through the nested results of multiple convolutional layers and pooling layers, using the multi-level meteorological elements output by the meteorological field simulation model as training samples and the measured data of the ground observation station as supervision labels.

[0009] Preferably, according to a precise traceability method based on high-resolution air quality monitoring data provided by the present invention, the high-resolution meteorological field data is input into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining the motion trajectory of the virtual particles, including: When it is detected that the pollutant concentration in the space where the monitoring point is located is abnormal, the high-resolution meteorological field data is input into the Lagrangian particle tracking random transport model to release virtual particles in the space where the monitoring point with abnormal pollutant concentration is located, and the virtual particles are driven to move in the corresponding meteorological field based on the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein the pollutant concentration is abnormal, indicating that the pollutant concentration of at least one type of pollutant is greater than or equal to a preset concentration threshold.

[0010] Preferably, according to a precise traceability method based on high-resolution air quality monitoring data provided by the present invention, the motion trajectory dX(t) of the virtual particle is expressed by the following formula: Where, is the particle position, is the wind speed vector, is the turbulent diffusion coefficient, is the random term of the Wiener process, and t is the time.

[0011] Preferably, according to a precise source tracing method based on high-resolution air quality monitoring data provided by the present invention, the method determines the pollution source information in the real meteorological environment based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle, including: Performing spatial clustering processing on the motion trajectory of the virtual particles to determine the dominant transmission path; Determining a trajectory weight coefficient of the motion trajectory of the virtual particle based on concentration time series data indicated by pollutant monitoring data at the monitoring point; wherein a larger value of the concentration time series data indicates a larger value of the corresponding trajectory weight coefficient; Based on the dominant transmission path and the trajectory weight coefficient, the residence time of the pollutants in the potential source area is quantified according to a preset kernel density estimation algorithm to generate a residence time distribution matrix; wherein the area with a high residence time density in the residence time distribution matrix is ​​characterized as the potential source location of the corresponding pollutant; The residence time distribution matrix and the emission data are analyzed and processed to determine the pollution source information of the pollutants in the real meteorological environment.

[0012] Preferably, according to a precise source tracing method based on high-resolution air quality monitoring data provided by the present invention, the analysis and processing based on the residence time distribution matrix and the emission data to determine the pollution source information of the pollutants in the real meteorological environment includes: Performing spatiotemporal matching of the potential source location of the pollutant and the high-resolution meteorological field corresponding to the high-resolution meteorological field data, and extracting the mixing layer height parameter of the corresponding time period; Constructing a sensitivity matrix based on the emission data when verifying, based on the mixing layer height parameter, that the trajectory direction of the virtual particle's motion trajectory is the same as the wind direction of the local wind field in the space where the corresponding monitoring point is located; wherein the sensitivity matrix is ​​used to quantify the potential contribution of each grid to the pollution at the monitoring point; Screening out significant emission sources based on the sensitivity matrix, the potential source locations, and the pollutant concentrations at the monitoring points, constructing a regularized objective function, and screening out significant emission sources through regularized regression; Based on the significant emission source, the pollutant monitoring data of the monitoring point, the emission data and the data source weight coefficient, the pollution source coordinates and / or pollutant emission intensity in the real meteorological environment are determined; wherein, the data source weight coefficient is determined by dynamic allocation based on the random forest model.

[0013] In a second aspect, the present invention further provides a precise traceability device based on high-resolution air quality monitoring data, comprising: Determine a high-resolution meteorological field data module, which is used to simulate meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model, and determine high-resolution meteorological field data; a motion trajectory determination module, configured to input the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; A pollution source information determination module is used to determine the pollution source information in a real meteorological environment based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0014] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements any of the above-described precise tracing methods based on high-resolution air quality monitoring data.

[0015] In a fourth aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described precise tracing methods based on high-resolution air quality monitoring data.

[0016] In a fifth aspect, the present invention also provides a computer program product, comprising a computer program, which, when executed by a processor, implements any of the above-described precise tracing methods based on high-resolution air quality monitoring data.

[0017] The present invention provides a precise tracing method, device, equipment, and storage medium based on high-resolution air quality monitoring data. The method determines high-resolution meteorological field data by simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model. The method inputs the high-resolution meteorological field data into a Lagrangian particle tracking random transmission model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles. The motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment. The method determines pollution source information in a real meteorological environment based on pollutant monitoring data at monitoring points, preset emission data, and the motion trajectories of the virtual particles. The method further comprises: determining the pollution source information in a real meteorological environment based on the pollutant monitoring data at monitoring points, preset emission data, and the emission data of different types of pollutants. The method further comprises: determining the pollution source coordinates and / or the pollutant emission intensity of pollutants in a real meteorological environment. The method solves the problem of low accuracy in pollutant tracing in the prior art, and realizing the combination of high-resolution meteorological simulation and random transmission model to provide more accurate pollution source positioning and tracing results. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is one of the flow charts of the precise traceability method based on high-resolution air quality monitoring data provided by the present invention.

[0020] Figure 2 It is a structural schematic diagram of the precise tracing device based on high-resolution air quality monitoring data provided by the present invention.

[0021] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0022] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0023] In the related art, there are at least the following technical problems: Existing air pollution source tracing technologies suffer from several key flaws, limiting their effectiveness in practical applications. First, most existing technologies rely on low-spatial-resolution monitoring methods, typically only providing results over a range of a few to tens of kilometers. This is particularly insufficient for accurately tracking pollution sources. In urban environments, pollution sources are often concentrated and highly localized, making traditional monitoring equipment and models unable to provide accurate source tracing analysis at small scales.

[0024] Second, existing technologies struggle to handle complex meteorological conditions. Meteorological factors such as wind speed, humidity, and air pressure significantly influence the dispersion of pollutants, but existing meteorological models are unable to accurately capture small-scale meteorological variations. This is particularly true in complex terrain and urban environments, where model predictions are inaccurate, leading to large errors in tracing pollution sources. Furthermore, existing methods suffer from poor real-time performance, with many systems unable to quickly respond to sudden changes in air quality, hindering the timely response of environmental regulators to pollution incidents.

[0025] Furthermore, existing technologies generally lack effective data fusion and comprehensive analysis capabilities. Most methods rely solely on a single data source for analysis, failing to effectively integrate multiple data sources, including meteorological, emissions, and monitoring data. This results in low accuracy and reliability in tracing results. Furthermore, many existing tracing methods are weak at identifying small-scale pollution sources, such as low-emission sources like construction and the catering industry, making it impossible to fully identify the impact of pollution sources on air quality.

[0026] The following combination Figure 1-Figure 3 The present invention describes a precise tracing method, device, equipment and storage medium based on high-resolution air quality monitoring data, which is used to address the low accuracy of pollutant tracing in the existing technology. By combining high-resolution meteorological simulation with a random transmission model, it can provide more accurate pollution source positioning and tracing results.

[0027] Figure 1 This is one of the flow charts of a precise traceability method based on high-resolution air quality monitoring data provided by the present invention, such as Figure 1 As shown, the method may include but is not limited to steps S100 to S300: S100, simulating meteorological phenomena at synoptic scales to regional scales in a real meteorological environment based on a meteorological field simulation model, and determining high-resolution meteorological field data; S200, inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; S300, based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle, determine the pollution source information in the real meteorological environment; wherein, the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0028] In step S100 of some embodiments, meteorological phenomena from weather scale to regional scale in a real meteorological environment are simulated based on a meteorological field simulation model to determine high-resolution meteorological field data.

[0029] It should be noted that the meteorological field simulation model can be the WRF (Weather Research and Forecasting) meteorological model. The WRF meteorological model is a numerical weather forecast and atmospheric simulation system widely used to simulate meteorological phenomena at scales from synoptic to regional. It is highly flexible and scalable, capable of simulating various meteorological processes, including changes in temperature, humidity, wind speed, and air pressure. The WRF model utilizes advanced numerical algorithms and physical schemes to provide highly accurate weather forecasts and simulation results.

[0030] Furthermore, based on the meteorological field simulation model, meteorological phenomena at the synoptic to regional scales in the real meteorological environment are simulated to determine high-resolution meteorological field data, including: Simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to obtain meteorological background data at at least two spatial scales; the at least two spatial scales are characterized by spatial scales at multiple different levels from a first height to a second height; Inputting the meteorological background data of the at least two spatial scales into a trained downscaling neural network for processing, and outputting the high-resolution meteorological field data; Among them, the trained downscaling neural network is obtained by performing multiple rounds of training processing through the nested results of multiple convolutional layers and pooling layers, using the multi-level meteorological elements output by the meteorological field simulation model as training samples and the measured data of the ground observation station as supervision labels.

[0031] It is understood that at least two of the spatial scales can be multi-level spatial scales: the WRF model output typically contains data from multiple vertical levels, from low layers (near the ground) to high layers (e.g., the tropopause). For example, multiple levels can be generated from a first altitude (e.g., 3 km) to a second altitude (e.g., 27 km), each corresponding to a different spatial resolution.

[0032] 27km scale: used to capture large-scale weather systems (such as fronts and cyclones). 3km scale: used to describe regional-scale meteorological characteristics (such as topographic influences and urban heat island effects).

[0033] Meteorological elements: The meteorological elements output by the WRF model include temperature, humidity, wind speed, wind direction, air pressure, etc. These elements have different distribution characteristics at different spatial scales.

[0034] The steps of the downscaling neural network process are as follows: Input data: The multi-level meteorological background data output by the WRF model is used as input. This data contains meteorological information from large scale to regional scale, but with low spatial resolution.

[0035] Downscaling neural network: Through the trained downscaling neural network, low-resolution meteorological data is converted into high-resolution refined meteorological field data.

[0036] Training process: Training samples: Multi-level meteorological elements (such as temperature, humidity, wind speed, etc.) output by the WRF model are used as input.

[0037] Supervision labels: Measured data from ground observation stations (such as temperature, precipitation, wind speed, etc.) are used as true values ​​for supervised training.

[0038] Network structure: Through the nested structure of multiple convolutional layers and pooling layers, the spatial correlation features of meteorological elements are extracted and the transformation rules between spatial scales are learned.

[0039] Downscaling neural network function: Downscaling neural network can capture the complex relationships between meteorological elements at different spatial scales and convert low-resolution data into high-resolution data based on these relationships.

[0040] Use downscaling neural network to output high-resolution meteorological field data.

[0041] Refined meteorological field: After processing by the downscaling neural network, high-resolution meteorological field data is output. This data has a higher spatial resolution and can more finely describe regional-scale meteorological phenomena.

[0042] Application scenarios of high-resolution meteorological field data: providing high-precision meteorological field data for air quality models, pollution source tracing, etc.

[0043] By combining the WRF meteorological model with a downscaling neural network, the team simulated meteorological phenomena from synoptic to regional scales and generated high-resolution, refined meteorological field data. This approach not only improved the spatial resolution of meteorological data but also enhanced the ability to describe local meteorological characteristics.

[0044] Inputting the meteorological background data of the at least two spatial scales into a trained downscaling neural network for processing and outputting the high-resolution meteorological field data can further reduce the 3km resolution meteorological data output by the WRF model to 1km resolution to capture more detailed local meteorological features (such as urban canopy effects, terrain undulations, etc.).

[0045] The trained downscaling neural network constructs a nonlinear mapping relationship to achieve the conversion from coarse resolution (3km) to high resolution (1km).

[0046] Steps in the process of training a downscaling neural network: Loss function: The mean square error (MSE) loss function is used to measure the difference between the model's predicted value and the true value.

[0047] Optimizer: Use the Adam optimizer to accelerate model convergence and improve training efficiency through adaptive learning rate adjustment.

[0048] Training goal: To enable the model to learn the mapping relationship from macroscopic meteorological patterns to microscopic local features, and ultimately generate a refined meteorological field with a spatial resolution of 1 km.

[0049] This embodiment uses statistical downscaling technology to capture the nonlinear relationship between meteorological elements through the CNN model, which can generate high-resolution refined meteorological fields and significantly improve the degree of detail restoration. Compared with traditional dynamic downscaling methods, statistical downscaling technology does not require complex numerical calculations, has higher computational efficiency, and is suitable for large-scale data processing. The multi-level simulation of the WRF model is combined with the downscaling processing of the CNN model to meet the meteorological simulation needs of different regions and scales.

[0050] In step S200 of some embodiments, the high-resolution meteorological field data is input into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data to obtain motion trajectories of virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment.

[0051] The Stochastic Time-Inverted Lagrangian Transport (STILT) model is a stochastic transport model based on Lagrangian particle tracking. It is primarily used to simulate the transport and diffusion of pollutants in the atmosphere. It simulates the transport paths and diffusion ranges of pollutants by releasing a large number of virtual particles and tracking their motion in the atmosphere. The STILT model accounts for the temporal and spatial variations in meteorological fields, as well as the impact of factors such as topography and landforms on pollutant transport.

[0052] The high-resolution meteorological data provided by the WRF model, such as wind speed, temperature, and air pressure, can be used as input to the STILT model to drive the movement of virtual particles. Furthermore, information on pollutant concentration distribution and transmission paths simulated by the STILT model can be fed back into the WRF model to further optimize weather forecasts and simulations.

[0053] By combining the meteorological simulations of the WRF model with the pollutant transport simulations of the STILT model, a comprehensive simulation of the atmospheric environment and pollutant transport processes can be achieved. This collaborative simulation can more accurately describe the interactions and impacts between atmospheric phenomena and pollutant transport.

[0054] Leveraging the high-precision weather forecasts of the WRF model, combined with the particle tracking technology of the STILT model, can improve the accuracy of simulations of pollutant transport and diffusion processes. This integration also reduces the errors and uncertainties that can arise when using either model alone, improving overall simulation efficiency.

[0055] In some embodiments of the present invention, when it is detected that the pollutant concentration in the space where the monitoring point is located is abnormal, the high-resolution meteorological field data is input into the Lagrangian particle tracking random transport model to release virtual particles in the space where the monitoring point with abnormal pollutant concentration is located, and the virtual particles are driven to move in the corresponding meteorological field based on the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein, the pollutant concentration is abnormal, indicating that the pollutant concentration of at least one type of pollutant is greater than or equal to a preset concentration threshold.

[0056] It should be noted that high-resolution meteorological data (1 km resolution) is generated using the WRF model and CNN downscaling technology, and includes meteorological elements such as temperature, wind speed, and air pressure. It is used to provide the driving field for the STILT model, simulating the movement of virtual particles in a real meteorological environment.

[0057] The STILT model is based on the Lagrangian particle tracking method and simulates the transmission and diffusion of pollutants in the atmosphere by releasing a large number of virtual particles.

[0058] The model takes into account the impact of temporal and spatial changes in meteorological fields (such as wind speed, wind direction, turbulence, etc.) as well as factors such as topography and landform on pollutant transmission.

[0059] A random algorithm is used to simulate the diffusion behavior of pollutants in a turbulent environment.

[0060] Furthermore, when the pollutant concentration at the monitoring point is abnormal (greater than or equal to the preset concentration threshold), the STILT model is started for simulation.

[0061] The simulation process includes the following steps: Virtual particle release: release virtual particles in the space where the monitoring point with abnormal pollutant concentration is located.

[0062] Meteorological field drive: Based on high-resolution meteorological field data, drive virtual particles to move in the corresponding meteorological field.

[0063] Motion trajectory generation: The motion trajectory of virtual particles is calculated through the STILT model to simulate the transmission path of pollutants in a real meteorological environment.

[0064] Output results: motion trajectory of virtual particles, representing the motion trajectory of pollutants in the real meteorological environment.

[0065] This example utilizes meteorological data with a 1km resolution to improve simulation accuracy and detail. The STILT model's stochastic algorithm more realistically simulates the diffusion behavior of pollutants in turbulent environments. It can respond to abnormal pollutant concentrations in real time and rapidly generate pollutant transmission paths.

[0066] By combining high-resolution meteorological data with the STILT model, a refined simulation of pollutant movement trajectories is achieved. In the event of abnormal pollutant concentrations, virtual particles can be rapidly released and their movement trajectories generated, providing strong support for pollution source tracing, emergency response, and environmental assessment. This approach not only improves simulation accuracy but also enhances adaptability to complex meteorological environments and pollutant diffusion processes.

[0067] In some embodiments of the present invention, the motion trajectory dX(t) of the virtual particle can be expressed by the following stochastic differential equation (SDE): Where, is the particle position, is the wind speed vector, is the turbulent diffusion coefficient, is the random term of the Wiener process, and t represents time.

[0068] X(t): The position vector of the particle at time t (i.e., coordinates (x, y, z) in three-dimensional space).

[0069] U(X(t), t): wind speed vector, representing the wind speed at position X(t) and time t.

[0070] : Turbulent diffusion coefficient matrix, representing the turbulent diffusion intensity at position X(t) and time t.

[0071] dW(t): The random term of the Wiener process, representing the random displacement caused by turbulence.

[0072] Wind speed driven term: U(X(t),t)dt represents the deterministic displacement of particles driven by the wind speed field U. The wind speed vector U is provided by high-resolution meteorological field data and reflects the overall trend of atmospheric flow.

[0073] Turbulent diffusion term: represents the random displacement caused by turbulence. Turbulent diffusion coefficient matrix It describes the intensity and anisotropy of the turbulence, and dW(t) is the random increment of the Wiener process, satisfying dW(t)∼N(0,dt) (normal distribution with mean 0 and variance dt).

[0074] This formula describes the trajectory of virtual particles in the meteorological field by combining a wind speed-driven term and a turbulent diffusion term. The wind speed vector u provides deterministic driving force, while the turbulent diffusion coefficient B and the Wiener process random term dW(t) introduce randomness, enabling the model to more realistically simulate the diffusion behavior of pollutants in a turbulent environment. By numerically solving this formula, the trajectory of virtual particles can be generated, supporting pollutant transport analysis and pollution source tracing.

[0075] The reverse trajectory calculation of pollutants is achieved through the STILT model, and its core process begins with the spatial and temporal coordinate positioning of the monitoring point (receptor point).

[0076] When pollutant concentrations are abnormal, the system releases tens to hundreds of thousands of virtual particles at the monitoring point. Based on the 1km resolution three-dimensional meteorological field generated by statistical downscaling (including parameters such as wind speed, turbulent diffusion coefficient, and boundary layer height), the system simulates the particle motion trajectory in the atmosphere using a reverse time integration method. The motion of each particle is driven by the Lagrangian stochastic differential equation, which is: in is the particle position, is the wind speed vector, is the turbulent diffusion coefficient, is the random term of the Wiener process. During this process, the model dynamically couples refined meteorological parameters such as vertical wind shear and atmospheric stability level at a minute-level time step, and introduces an urban building canopy model to correct for terrain friction effects, ultimately generating a high-confidence reverse trajectory cluster.

[0077] To improve the spatiotemporal accuracy of trajectory inversion, the system adopts a multi-scale computing strategy: for short-term near-field diffusion (such as within a few hours), the rapid transmission of pollutants is captured with a high temporal resolution of 1 minute; for long-distance transmission (transmission across regions for several days), the step size is dynamically adjusted to 10-30 minutes to balance computing efficiency and accuracy.

[0078] In some embodiments of the present invention, the step of determining the motion trajectory of the virtual particle further includes a core process of calculating the reverse trajectory of the pollutant, including the following steps: 1. Determine the spatial and temporal coordinates of the monitoring points: Determine the spatial and temporal coordinates (latitude, longitude, altitude, time) of the monitoring point (receptor point) where the pollutant concentration is abnormal. This point serves as the starting point for the reverse trajectory calculation.

[0079] 2. Virtual particle release: Tens of thousands to hundreds of thousands of virtual particles are released at the monitoring point.

[0080] These particles represent the possible locations of pollution sources, and by tracing their movement trajectories in reverse, the source of the pollutants can be inferred.

[0081] 3. Driven by high-resolution meteorological fields: The 1 km resolution three-dimensional meteorological field data (including wind speed, turbulent diffusion coefficient, boundary layer height and other parameters) generated by statistical downscaling are used as the driving field.

[0082] Meteorological field data provides the background environment for particle movement, including information such as wind speed vector, turbulence intensity, and atmospheric stability.

[0083] 4. Reverse time integration: The motion of each particle is calculated using a Lagrangian stochastic differential equation driving the equation.

[0084] Taking the monitoring point as the starting point, reverse integration is performed along time to simulate the movement trajectory of particles in the atmosphere.

[0085] 5. Refined meteorological parameter coupling: Dynamically couple refined meteorological parameters such as vertical wind shear and atmospheric stability level with a time step of minutes.

[0086] The urban building canopy model is introduced to correct the terrain friction effect and consider the influence of complex urban terrain on particle motion.

[0087] 6. Multi-scale computing strategy: Short-term near-field diffusion: Captures the rapid movement of pollutants with a high time resolution of 1 minute for rapid transmission within a few hours. Long-distance transmission: For particles that travel across regions for several days, dynamically adjusts the step size to 10-30 minutes to balance computational efficiency and accuracy.

[0088] 7. Generate reverse trajectory clusters: By reverse tracking a large number of virtual particles, high-confidence reverse trajectory clusters are generated.

[0089] These trajectory clusters reflect the possible source areas and transmission pathways of pollutants.

[0090] This embodiment uses 1km resolution meteorological field data and a minute-level time step to ensure the precision of trajectory calculation.

[0091] A multi-scale computing strategy is adopted to take into account the needs of short-time near-field diffusion and long-distance transmission.

[0092] The Lagrangian stochastic differential equation is combined with refined meteorological parameters to realistically simulate the diffusion behavior of pollutants in a turbulent environment.

[0093] The urban building canopy model corrects the terrain friction effect and adapts to the complex urban environment.

[0094] In step S300 of some embodiments, the pollution source information in the real meteorological environment is determined based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0095] The pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of pollutants in a real meteorological environment.

[0096] By combining the pollutant monitoring data of the monitoring points, the preset emission data and the movement trajectory of the virtual particles, the pollution source information of the pollutants in the real meteorological environment is analyzed. The specific process is as follows: 1.1 Data input steps include: Input pollutant monitoring data at monitoring points: including pollutant concentration, time, spatial coordinates (latitude, longitude, altitude), etc. This is used to locate areas and times with abnormal pollutant concentrations.

[0097] Preset emission data: includes data on the types of pollutants emitted, as well as information such as the location, emission intensity, and emission time of known pollution sources. This data is used to assist in the identification and verification of pollution sources.

[0098] Virtual particle motion trajectories: The reverse trajectory clusters generated by the STILT model reflect the possible source areas of pollutants.

[0099] 1.2 Trajectory cluster analysis steps include: Cluster analysis is performed on the reverse trajectories of virtual particles to identify possible sources of pollutants. Clustering algorithms (such as K-means and DBSCAN) are used to divide the trajectories into multiple clusters, each corresponding to a possible pollution source area.

[0100] 1.3 The steps for determining the pollution source coordinates include: The coordinates of the pollution source are determined by trajectory clustering results. The pollution source coordinates are usually the center of the trajectory cluster or the high-density area.

[0101] 1.4 The steps for calculating pollutant emission intensity include: combining the pollutant concentration at the monitoring point and the particle density of the trajectory cluster to calculate the emission intensity of the pollution source.

[0102] Estimate emissions from pollution sources using the principle of conservation of mass or the back-diffusion model.

[0103] Pollution source information includes the following: Pollution source coordinates: the spatial location (latitude, longitude, and altitude) of the pollution source in the real meteorological environment.

[0104] Pollutant emission intensity: The amount of emissions from a pollution source (e.g., mass or volume of emissions per unit time). This reflects the contribution of the pollution source to the pollutant concentration at the monitoring point.

[0105] Clustering algorithms: Use K-means, DBSCAN, and other clustering algorithms to group reverse trajectories into multiple clusters. Each cluster corresponds to a possible pollution source. Trajectories are clustered based on metrics such as spatial density and temporal consistency. High-density areas typically correspond to pollution sources.

[0106] Principle of conservation of mass: Calculate the emission of pollution sources through the pollutant concentration at the monitoring point and the particle density of the trajectory cluster.

[0107] The formula is: Q=C V / t Where: Q is the emission intensity of the pollution source, C is the pollutant concentration at the monitoring point, V is the air volume corresponding to the trajectory cluster, and t is time.

[0108] Backward Diffusion Model: Use a backward diffusion model (such as the Gaussian diffusion model) to estimate the emission intensity of pollution sources. Combined with meteorological data (wind speed, turbulent diffusion coefficient, etc.) and monitoring data, the emission volume of pollution sources can be inferred.

[0109] The embodiment of the present invention further includes: the pollution source verification step includes: Comparison with preset emission data: Compare the calculated pollution source information with preset emission data to verify the accuracy of the pollution source. If the pollution source is consistent with a known emission source, its reliability is further confirmed. Multi-monitoring point joint analysis: Combine pollutant monitoring data and trajectory analysis results from multiple monitoring points to improve the accuracy of pollution source positioning.

[0110] The present invention uses inverse trajectory clustering and emission intensity calculation to accurately locate the location and emission intensity of pollution sources. It also combines monitoring data, preset emission data, and virtual particle trajectories to improve the reliability of pollution source identification.

[0111] In some embodiments, the steps further include the following processing flow: monitoring point pollutant monitoring data + preset emission data + virtual particle trajectory → trajectory cluster analysis → pollution source coordinate determination → emission intensity calculation → pollution source information output.

[0112] In some embodiments of the present invention, determining pollution source information in a real meteorological environment based on pollutant monitoring data at monitoring points, preset emission data, and motion trajectories of the virtual particles includes: Performing spatial clustering processing on the motion trajectory of the virtual particles to determine the dominant transmission path; Determining a trajectory weight coefficient of the motion trajectory of the virtual particle based on concentration time series data indicated by pollutant monitoring data at the monitoring point; wherein a larger value of the concentration time series data indicates a larger value of the corresponding trajectory weight coefficient; Based on the dominant transmission path and the trajectory weight coefficient, the residence time of pollutants in the potential source area is quantified according to a preset kernel density estimation algorithm to generate a residence time distribution matrix; The region with a high residence time density in the residence time distribution matrix is ​​characterized as a potential source location of the corresponding pollutant; the kernel density estimation algorithm represents a probability density function for estimating the trajectory weight coefficient; The residence time distribution matrix and the emission data are analyzed and processed to determine the pollution source information of the pollutants in the real meteorological environment.

[0113] It can be understood that the spatial clustering of the motion trajectories of virtual particles: the reverse trajectories of virtual particles are spatially clustered to determine the dominant transmission path.

[0114] Trajectory weight coefficient calculation: Based on the pollutant concentration time series data of the monitoring point, the weight coefficient of each trajectory is calculated. The higher the concentration of the pollutant concentration time series data, the greater the trajectory weight.

[0115] Steps for combining kernel density estimation algorithm with residence time quantification: Combine the dominant transport path and trajectory weight coefficient to quantify the residence time of pollutants in potential source areas through kernel density estimation.

[0116] Generate a residence time distribution matrix and identify areas with high residence time density as potential pollution source locations.

[0117] Pollution source information determination: Combined with the residence time distribution matrix and preset emission data, the location and emission intensity of the pollution source are analyzed and determined.

[0118] Furthermore, the virtual particle motion trajectories are spatially clustered. The purpose of clustering is to identify the dominant paths of pollutant transmission and reduce the complexity of trajectory data.

[0119] Clustering method: Use spatial clustering algorithms (such as DBSCAN, K-means, etc.) to cluster reverse trajectories. Clustering is based on the spatial similarity and temporal consistency of trajectories.

[0120] Output results: Dominant transmission path: The main trajectory cluster obtained after clustering represents the transmission path of pollutants from the monitoring point to the pollution source.

[0121] Weight coefficient definition: The weight coefficient of each trajectory reflects its contribution to the pollutant concentration. The larger the concentration time series data value, the larger the trajectory weight coefficient.

[0122] Based on the pollutant concentration time series data of the monitoring point, the weight coefficient of each trajectory is calculated. The formula is as follows: Where, is the weight coefficient of the i-th trajectory, is the pollutant concentration at the corresponding time of the i-th trajectory, is the pollutant concentration at the time corresponding to the jth trajectory, and N is the total number of trajectories.

[0123] The kernel density estimation steps are as follows: Use a kernel density estimation method (such as the Gaussian kernel function) to calculate the spatial density of potential source areas. Combined with the trajectory weight coefficient, the residence time of pollutants in the potential source area is quantified.

[0124] The steps to determine the residence time distribution matrix are as follows: Generate a two-dimensional matrix representing the residence time density of pollutants at different spatial locations.

[0125] The area with a larger value in the matrix indicates that the pollutant stays longer in the area, which may be the location of the pollution source.

[0126] Furthermore, through the residence time distribution matrix, areas with high residence time density are identified as potential pollution source locations.

[0127] Combine the preset emission data and residence time distribution matrix to calculate the emission intensity of the pollution source. The formula is as follows: Where Q is the emission intensity of the pollution source, D(x) is the residence time density at location x, V is the air volume, and t is time.

[0128] The role of dominant transmission path identification: Through spatial clustering, the main paths of pollutant transmission can be identified to reduce the complexity of data processing.

[0129] The role of the trajectory weight coefficient: Combined with the time series data of pollutant concentration, the trajectory weight is dynamically adjusted to improve the accuracy of pollution source positioning.

[0130] The role of the kernel density estimation algorithm is to quantify the residence time of pollutants in potential source areas and identify areas with high residence time density as potential pollution sources.

[0131] The role of multi-source data fusion: combining virtual particle trajectories, monitoring data and emission data to improve the reliability of pollution source information.

[0132] By combining spatial clustering of virtual particle trajectories, trajectory weight coefficients, kernel density estimation, and residence time distribution matrices, this method can efficiently identify the locations and emission intensities of potential pollutant sources. This method leverages multi-source data fusion and advanced algorithms to improve the accuracy and reliability of pollution source location, providing strong support for pollution source tracing, emission inventory verification, and emergency response.

[0133] In some embodiments of the present invention, the analyzing and processing based on the residence time distribution matrix and the emission data to determine the pollution source information of the pollutants in the real meteorological environment includes: Performing spatiotemporal matching of the potential source location of the pollutant with the high-resolution meteorological field corresponding to the high-resolution meteorological field data, and extracting the mixing layer height parameter for the corresponding time period; verifying, based on the mixing layer height parameter, that the trajectory direction of the virtual particle's motion trajectory is the same as the wind direction of the local wind field in the space where the corresponding monitoring point is located, and constructing a sensitivity matrix based on the emission data; wherein the sensitivity matrix is ​​used to quantify the potential contribution of each grid to the pollution at the monitoring point; Screening out significant emission sources based on the sensitivity matrix, the potential source locations, and the pollutant concentrations at the monitoring points, constructing a regularized objective function, and screening out significant emission sources through regularized regression; Based on the significant emission source, the pollutant monitoring data of the monitoring point, the emission data and the data source weight coefficient, the pollution source coordinates and / or pollutant emission intensity in the real meteorological environment are determined; wherein, the data source weight coefficient is determined by dynamic allocation based on the random forest model.

[0134] It is understood that the steps of spatiotemporal matching and mixing layer height extraction are: spatiotemporal matching of potential source locations with high-resolution meteorological field data, extraction of mixing layer height parameters for the corresponding time period, and verification of whether the direction of the virtual particle trajectory is consistent with the local wind field direction.

[0135] Sensitivity matrix construction: Based on the emission data, a sensitivity matrix is ​​constructed to quantify the potential contribution of each grid to the pollution at the monitoring point.

[0136] Regularized objective function and significant emission source screening: A regularized objective function is constructed by combining the sensitivity matrix, potential source locations, and real-time monitoring data. Significant emission sources are screened through regularized regression. Pollution source information determination: Pollution source coordinates and emission intensities are determined based on significant emission sources, monitoring data, emission data, and data source weight coefficients. Data source weight coefficients are dynamically assigned by a random forest model.

[0137] Specifically, the spatiotemporal matching step involves spatiotemporally matching the potential source location with high-resolution meteorological data (e.g., 1 km resolution), ensuring that the timing of the meteorological data is consistent with the pollutant transmission period.

[0138] Mixing layer height extraction steps: Extract the mixing layer height parameter from the matched meteorological field data. The mixing layer height reflects the vertical diffusion range of pollutants.

[0139] Steps for trajectory direction verification: Compare the virtual particle trajectory direction with the local wind direction at the monitoring point.

[0140] If the two are consistent, the trajectory simulation is reasonable and the subsequent analysis continues; otherwise, the unreasonable trajectory is excluded.

[0141] Furthermore, the sensitivity matrix is ​​defined as follows: The sensitivity matrix S is a two-dimensional matrix that represents the potential contribution of each grid to the pollution of the monitoring point.

[0142] The matrix element Si,j represents the pollution contribution of grid i to monitoring point j.

[0143] Construction method: Based on emission data and meteorological field data (such as wind speed, mixing layer height, etc.), calculate the pollution contribution of each grid to the monitoring point.

[0144] Use the Lagrangian particle diffusion model or Gaussian diffusion model to simulate the transport process of pollutants from the grid to the monitoring points.

[0145] Output: Sensitivity matrix S, which is used to quantify the pollution contribution potential of each grid.

[0146] Construct an objective function that combines the sensitivity matrix, potential source locations, and real-time monitoring data to optimize the identification of emission sources.

[0147] The formula of regularized objective function K is as follows: K Where y is the pollutant concentration data at the monitoring point, S is the sensitivity matrix, and x is the emission source intensity vector. is a regularization parameter that controls sparsity. Regularized regression methods such as LASSO (Least Absolute Shrinkage and Selection Operator) are used to screen out significant emission sources.

[0148] If the value corresponding to a significant emission source is non-zero, other emission sources are eliminated. The location and intensity of the significant emission source are obtained. Pollution source coordinate determination: The coordinates of the pollution source are determined by combining the location of the significant emission source and the location of the potential source.

[0149] Emission intensity determination: Calculate the emission intensity of pollution sources based on the intensity of significant emission sources and monitoring data.

[0150] Data source weighting: A random forest model is used to dynamically assign data source weights, comprehensively considering the contributions of monitoring data, emissions data, and significant emission sources. The random forest model assigns weights to different data sources based on feature importance assessments.

[0151] Output: coordinates of pollution sources and emission intensity.

[0152] Dynamically assign data source weight coefficients, comprehensively consider the contributions of multi-source data, and improve the reliability of results.

[0153] In some embodiments of the present invention, in order to improve the spatiotemporal accuracy of trajectory inversion, the system adopts a multi-scale computing strategy: for short-term near-field diffusion (such as within a few hours), the rapid transmission of pollutants is captured with a high temporal resolution of 1 minute; for long-distance transmission (transmission across regions for several days), the step size is dynamically adjusted to 10-30 minutes to balance computing efficiency and accuracy.

[0154] Trajectory clusters were spatially clustered using the DBSCAN algorithm to identify dominant transmission pathways. Trajectory weight coefficients were then assigned based on time-series data from monitoring point concentrations to further identify highly contributing pathways. Finally, a residence time distribution (RTD) matrix was generated using kernel density estimation to identify potential source areas of pollutant transmission, with a spatial resolution of up to 100 meters.

[0155] During the pollution source location and emission characteristic analysis stage, the system matches the potential source area with the 1km high-resolution meteorological field in time and space, extracts key parameters such as the wind rose diagram and mixing layer height of the corresponding time period, and verifies the consistency of the trajectory direction with the local wind field.

[0156] By constructing a sensitivity matrix, we quantify the potential contribution of each grid to pollution at the monitoring point. Based on a Bayesian inversion framework, using trajectory dwell time distribution as a priori probability, we integrate real-time monitoring concentration data to construct an L1 regularized objective function. Lasso regression is used to screen significant emission sources, and non-negative matrix factorization (NMF) is used to analyze the emission time spectrum of pollution sources and identify characteristics such as diurnal fluctuations in industrial emissions.

[0157] To ensure the reliability of the results, the system cross-validates the inversion results with multi-source data (emission inventories, enterprise online monitoring), dynamically assigns weight coefficients to each data source through a random forest model, and ultimately outputs the pollution source coordinates and emission intensity.

[0158] The present invention combines meteorological data, pollutant monitoring data, and emissions data, using machine learning algorithms to quantify the contributions of different pollution sources. By integrating real-time monitoring data (such as PM2.5 and PM10 concentrations) with emissions inventory data and meteorological simulation results, this technology can identify the pollution sources that contribute most to micro-station monitoring data and further analyze the specific impacts of different pollution sources on air quality. Unlike traditional single-source data analysis methods, this invention improves the accuracy of source contribution analysis through multi-source data fusion and the application of deep learning algorithms, enabling detailed analysis of complex pollution sources and environments.

[0159] This invention has the ability to respond to pollution incidents in real time. When pollutant concentrations exceed standards or a sudden pollution incident occurs, it can quickly analyze the pollution source using real-time monitoring data and meteorological data. Through real-time data updates and model calculations, environmental monitoring departments can promptly locate pollution sources and predict the spread of pollutants, providing decision support. This real-time capability gives this technology a significant advantage in emergency pollution incident response.

[0160] After tracing pollution sources, this invention uses visualization tools to present results such as the location of pollution sources, the spatial distribution of pollutant concentrations, and source contribution analysis. These visualizations can help environmental regulators clearly understand the distribution of pollution sources, the diffusion paths of pollutants, and the impact of different sources on the environment, thereby providing a scientific basis for subsequent pollution control and environmental management. Through this visualization application, regulators can take timely measures to optimize policy formulation and pollution prevention and control strategies.

[0161] The present invention provides a precise tracing method, device, equipment, and storage medium based on high-resolution air quality monitoring data. The method determines high-resolution meteorological field data by simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model. The method inputs the high-resolution meteorological field data into a Lagrangian particle tracking random transmission model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles. The motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment. The method determines pollution source information in a real meteorological environment based on pollutant monitoring data at monitoring points, preset emission data, and the motion trajectories of the virtual particles. The method further comprises: determining the pollution source information in a real meteorological environment based on the pollutant monitoring data at monitoring points, preset emission data, and the emission data of different types of pollutants. The method further comprises: determining the pollution source coordinates and / or the pollutant emission intensity of pollutants in a real meteorological environment. The method solves the problem of low accuracy in pollutant tracing in the prior art, and realizing the combination of high-resolution meteorological simulation and random transmission model to provide more accurate pollution source positioning and tracing results.

[0162] The precise tracing device based on high-resolution air quality monitoring data provided by the present invention is described below. The precise tracing device based on high-resolution air quality monitoring data described below and the precise tracing method based on high-resolution air quality monitoring data described above can be referenced to each other.

[0163] like Figure 2 The figure shows a schematic diagram of the structure of a precise traceability device based on high-resolution air quality monitoring data provided by the present invention. The precise traceability device based on high-resolution air quality monitoring data includes the following modules: A high-resolution meteorological field data determination module 210 is configured to simulate meteorological phenomena at a synoptic scale to a regional scale in a real meteorological environment based on a meteorological field simulation model to determine high-resolution meteorological field data; a motion trajectory determination module 220 for inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; The pollution source information determination module 230 is used to determine the pollution source information in the real meteorological environment based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0164] Preferably, the precise tracing device based on high-resolution air quality monitoring data provided by the present invention is specifically used to simulate meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to obtain meteorological background data at at least two spatial scales; the at least two spatial scales are characterized by multiple different levels of spatial scales between a first height and a second height; Inputting the meteorological background data of the at least two spatial scales into a trained downscaling neural network for processing, and outputting the high-resolution meteorological field data; Among them, the trained downscaling neural network is obtained by performing multiple rounds of training processing through the nested results of multiple convolutional layers and pooling layers, using the multi-level meteorological elements output by the meteorological field simulation model as training samples and the measured data of the ground observation station as supervision labels.

[0165] Preferably, the precise tracing device based on high-resolution air quality monitoring data provided by the present invention is specifically used to input the high-resolution meteorological field data into the Lagrangian particle tracking random transport model when it is monitored that the pollutant concentration in the space where the monitoring point is located is abnormal, so as to release virtual particles in the space where the monitoring point with abnormal pollutant concentration is located, and drive the virtual particles to move in the corresponding meteorological field based on the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein, the pollutant concentration is abnormal, indicating that the pollutant concentration of at least one type of pollutant is greater than or equal to a preset concentration threshold.

[0166] Preferably, the precise traceability device based on high-resolution air quality monitoring data provided by the present invention is specifically used for the motion trajectory dX(t) of the virtual particle, which is expressed by the following formula: Where, is the particle position, is the wind speed vector, is the turbulent diffusion coefficient, is the random term of the Wiener process.

[0167] Preferably, the precise tracing device based on high-resolution air quality monitoring data provided by the present invention is specifically used to perform spatial clustering processing on the motion trajectory of the virtual particles to determine the dominant transmission path; Determining a trajectory weight coefficient of the motion trajectory of the virtual particle based on concentration time series data indicated by pollutant monitoring data at the monitoring point; wherein a larger value of the concentration time series data indicates a larger value of the corresponding trajectory weight coefficient; Based on the dominant transmission path and the trajectory weight coefficient, the residence time of the pollutants in the potential source area is quantified according to a preset kernel density estimation algorithm to generate a residence time distribution matrix; wherein the areas with high residence time density in the residence time distribution matrix are characterized as potential source locations of the corresponding pollutants; the kernel density estimation algorithm represents the probability density function used to estimate the trajectory weight coefficient; The residence time distribution matrix and the emission data are analyzed and processed to determine the pollution source information of the pollutants in the real meteorological environment.

[0168] Preferably, the precise source tracing device based on high-resolution air quality monitoring data provided by the present invention is specifically used to perform spatiotemporal matching between the potential source location of the pollutant and the high-resolution meteorological field corresponding to the high-resolution meteorological field data, and extract the mixing layer height parameter of the corresponding time period; and construct a sensitivity matrix based on the emission data when verifying that the trajectory direction of the virtual particle's motion trajectory is the same as the wind direction of the local wind field in the space where the corresponding monitoring point is located based on the mixing layer height parameter; wherein the sensitivity matrix is ​​used to quantify the potential contribution of each grid to the pollution at the monitoring point; Screening out significant emission sources based on the sensitivity matrix, the potential source locations, and the pollutant concentrations at the monitoring points, constructing a regularized objective function, and screening out significant emission sources through regularized regression; Based on the significant emission source, the pollutant monitoring data of the monitoring point, the emission data and the data source weight coefficient, the pollution source coordinates and / or pollutant emission intensity in the real meteorological environment are determined; wherein, the data source weight coefficient is determined by dynamic allocation based on the random forest model.

[0169] The present invention provides a precise tracing method, device, equipment, and storage medium based on high-resolution air quality monitoring data. The method determines high-resolution meteorological field data by simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model. The method inputs the high-resolution meteorological field data into a Lagrangian particle tracking random transmission model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles. The motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment. The method determines pollution source information in a real meteorological environment based on pollutant monitoring data at monitoring points, preset emission data, and the motion trajectories of the virtual particles. The method further comprises: determining the pollution source information in a real meteorological environment based on the pollutant monitoring data at monitoring points, preset emission data, and the emission data of different types of pollutants. The method further comprises: determining the pollution source coordinates and / or the pollutant emission intensity of pollutants in a real meteorological environment. The method solves the problem of low accuracy in pollutant tracing in the prior art, and realizing the combination of high-resolution meteorological simulation and random transmission model to provide more accurate pollution source positioning and tracing results.

[0170] Figure 3 An example of a physical structure diagram of an electronic device is shown below. Figure 3 As shown, the electronic device may include: a processor 310 , a communications interface 320 , a memory 330 and a communication bus 340 , wherein the processor 310 , the communications interface 320 and the memory 330 communicate with each other via the communication bus 340 . The processor 310 can call the logic instructions in the memory 330 to execute a precise tracing method based on high-resolution air quality monitoring data, which includes: simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to determine high-resolution meteorological field data; inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein the motion trajectory of the virtual particles represents the motion trajectory of pollutants moving in a real meteorological environment; based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particles, determining the pollution source information in the real meteorological environment; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0171] Furthermore, the logic instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0172] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the precise tracing method based on high-resolution air quality monitoring data provided by the above methods, the method including: simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to determine high-resolution meteorological field data; inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein the motion trajectory of the virtual particles represents the motion trajectory of pollutants moving in the real meteorological environment; based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particles, determining the pollution source information in the real meteorological environment; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

[0173] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the precise tracing method based on high-resolution air quality monitoring data provided by the above-mentioned methods, the method comprising: simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to determine high-resolution meteorological field data; inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data to obtain motion trajectories of virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; determining pollution source information in a real meteorological environment based on the pollutant monitoring data of the monitoring point, preset emission data and the motion trajectories of the virtual particles; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of pollutants in a real meteorological environment.

[0174] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0175] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0176] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A precise traceability method based on high-resolution air quality monitoring data, characterized in that: include: Based on the meteorological field simulation model, simulate meteorological phenomena from the synoptic scale to the regional scale in the real meteorological environment and determine high-resolution meteorological field data; Inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; Based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particles, the pollution source information in the real meteorological environment is determined; wherein, the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

2. The precise source tracing method based on high-resolution air quality monitoring data according to claim 1, characterized in that: The meteorological field simulation model is used to simulate meteorological phenomena at the weather scale to the regional scale in a real meteorological environment, and determine high-resolution meteorological field data, including: Simulating meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model to obtain meteorological background data at at least two spatial scales; the at least two spatial scales are characterized by spatial scales at multiple different levels from a first height to a second height; Inputting the meteorological background data of the at least two spatial scales into a trained downscaling neural network for processing, and outputting the high-resolution meteorological field data; Among them, the trained downscaling neural network is obtained by training and processing the multi-level meteorological elements output by the meteorological field simulation model as training samples and the measured data of the ground observation station as supervision labels through the nested results of multiple convolutional layers and pooling layers.

3. The precise source tracing method based on high-resolution air quality monitoring data according to claim 1, characterized in that: Inputting the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data to obtain motion trajectories of the virtual particles includes: When it is detected that the pollutant concentration in the space where the monitoring point is located is abnormal, the high-resolution meteorological field data is input into the Lagrangian particle tracking random transport model to release virtual particles in the space where the monitoring point with abnormal pollutant concentration is located, and the virtual particles are driven to move in the corresponding meteorological field based on the high-resolution meteorological field data to obtain the motion trajectory of the virtual particles; wherein the pollutant concentration is abnormal, indicating that the pollutant concentration of at least one type of pollutant is greater than or equal to a preset concentration threshold.

4. The precise source tracing method based on high-resolution air quality monitoring data according to claim 3 is characterized in that: The motion trajectory dX(t) of the virtual particle is expressed by the following formula: Where, is the particle position, is the wind speed vector, is the turbulent diffusion coefficient, is the random term of the Wiener process, and t is the time.

5. The precise source tracing method based on high-resolution air quality monitoring data according to claim 1, characterized in that: The determining of pollution source information of pollutants in a real meteorological environment based on pollutant monitoring data of monitoring points, preset emission data and motion trajectories of the virtual particles includes: Performing spatial clustering processing on the motion trajectory of the virtual particles to determine the dominant transmission path; Determining a trajectory weight coefficient of the motion trajectory of the virtual particle based on concentration time series data indicated by pollutant monitoring data at the monitoring point; wherein a larger value of the concentration time series data indicates a larger value of the corresponding trajectory weight coefficient; Based on the dominant transmission path and the trajectory weight coefficient, the residence time of the pollutants in the potential source area is quantified according to a preset kernel density estimation algorithm to generate a residence time distribution matrix; wherein the area with a high residence time density in the residence time distribution matrix is ​​characterized as the potential source location of the corresponding pollutant; The residence time distribution matrix and the emission data are analyzed and processed to determine the pollution source information of the pollutants in the real meteorological environment.

6. The precise source tracing method based on high-resolution air quality monitoring data according to claim 5 is characterized in that: The analyzing and processing based on the residence time distribution matrix and the emission data to determine the pollution source information of the pollutants in the real meteorological environment includes: Performing spatiotemporal matching of the potential source location of the pollutant and the high-resolution meteorological field corresponding to the high-resolution meteorological field data, and extracting the mixing layer height parameter of the corresponding time period; Constructing a sensitivity matrix based on the emission data when verifying, based on the mixing layer height parameter, that the trajectory direction of the virtual particle's motion trajectory is the same as the wind direction of the local wind field in the space where the corresponding monitoring point is located; wherein the sensitivity matrix is ​​used to quantify the potential contribution of each grid to the pollution at the monitoring point; Screening out significant emission sources based on the sensitivity matrix, the potential source locations, and the pollutant concentrations at monitoring points; Based on the significant emission source, the pollutant monitoring data of the monitoring point, the emission data and the data source weight coefficient, the pollution source coordinates and / or pollutant emission intensity in the real meteorological environment are determined; wherein, the data source weight coefficient is determined by dynamic allocation based on the random forest model.

7. A precise traceability device based on high-resolution air quality monitoring data, characterized in that: include: Determine a high-resolution meteorological field data module, which is used to simulate meteorological phenomena from weather scale to regional scale in a real meteorological environment based on a meteorological field simulation model, and determine high-resolution meteorological field data; a motion trajectory determination module, configured to input the high-resolution meteorological field data into a Lagrangian particle tracking random transport model to drive virtual particles to move in a meteorological field corresponding to the high-resolution meteorological field data, thereby obtaining motion trajectories of the virtual particles; wherein the motion trajectories of the virtual particles represent the motion trajectories of pollutants moving in a real meteorological environment; A pollution source information determination module is used to determine the pollution source information in a real meteorological environment based on the pollutant monitoring data of the monitoring point, the preset emission data and the motion trajectory of the virtual particle; wherein the pollutant monitoring data represents the pollutant concentration corresponding to each type of pollutant, and the emission data represents the type data of different types of pollutants emitted; the pollution source information represents the pollution source coordinates and / or pollutant emission intensity of the pollutants in the real meteorological environment.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the program, the precise tracing method based on high-resolution air quality monitoring data as described in any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the precise tracing method based on high-resolution air quality monitoring data as described in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the precise tracing method based on high-resolution air quality monitoring data as described in any one of claims 1 to 6 is implemented.

Images