Simulation system for atmospheric pollution diffusion

Abstract

The invention relates to the technical field of environmental science, and discloses an atmospheric pollution diffusion simulation system, which comprises a data acquisition module, a data preprocessing module, a dynamic simulation model module, a simulation calculation module, a visualization module and a decision support module, and is characterized in that the data acquisition module is used for acquiring multi-source data; the data preprocessing module is used for cleaning, fusing and optimizing the collected data; the dynamic simulation model module adopts a dynamic mixed diffusion model and performs model parameter optimization in combination with deep learning; the simulation calculation module is used for completing efficient simulation of pollutant diffusion; the visualization module is used for dynamically displaying pollutant concentration distribution and diffusion paths in a three-dimensional manner; and the decision support module generates pollution control suggestions and risk early warning information based on the simulation result. According to the method, intelligent switching between the Gaussian plume model and the CFD model is realized by introducing the dynamic mixed diffusion model, and the method can adapt to various scenes from simple terrains to complex terrains.

Description

Technical Field

[0001] The invention relates to the field of environmental science and technology, and in particular to a simulation system for air pollution diffusion. Background Art

[0002] Air pollution has become one of the global environmental problems, posing a serious threat to ecosystems and human health. With the acceleration of urbanization and the increase in industrial activities, the emission of air pollutants such as PM2.5, SO2, and NOx has increased significantly, and their diffusion and distribution characteristics are affected by meteorological conditions, terrain factors, and pollution source characteristics. In order to assess the impact range, concentration distribution, and potential harm to the environment of air pollution, simulation technology has become an important means.

[0003] Although the Gaussian plume model commonly used in the prior art has the advantage of high computational efficiency, its simulation accuracy drops significantly under complex terrain and dynamic meteorological conditions, and it is difficult to reflect the impact of turbulence, terrain characteristics, and non-steady-state meteorology on pollutant diffusion. Although the CFD model can provide high-precision simulation in complex scenes, its computational complexity is extremely high and it is difficult to meet the needs of real-time or large-scale diffusion prediction. In addition, traditional models usually use a single simulation method, lack dynamic switching capabilities, and cannot balance computational efficiency and simulation accuracy. Summary of the invention

[0004] In order to make up for the above shortcomings, the present invention provides a simulation system for atmospheric pollution diffusion, aiming to improve the problem that traditional models usually use a single simulation method and lack dynamic switching capabilities.

[0005] In a first aspect, the present invention provides the following technical solution: a simulation system for air pollution diffusion, comprising a data acquisition module, a data preprocessing module, a dynamic simulation model module, a simulation calculation module, a visualization module and a decision support module.

[0006] The data acquisition module is used to collect multi-source data including pollutant concentration, meteorological parameters and geographic information;

[0007] The data preprocessing module is used to clean, fuse and optimize the collected data;

[0008] The dynamic simulation model module adopts a dynamic mixed diffusion model and combines deep learning to optimize model parameters;

[0009] The simulation calculation module is based on a distributed computing framework to complete efficient simulation of pollutant diffusion;

[0010] The visualization module is used to dynamically display the pollutant concentration distribution and diffusion path in three dimensions;

[0011] The decision support module generates pollution control suggestions and risk warning information based on the simulation results.

[0012] Preferably, the data acquisition module includes:

[0013] Ground collection units, which use electrochemical sensors to collect pollutant concentrations through fixed monitoring stations deployed in the target area;

[0014] Mobile collection unit, a laser spectrum sensor mounted on a drone, collects the concentration of trace pollutants in the air in real time and provides accurate spatial location data in combination with GPS;

[0015] The remote sensing acquisition unit, based on satellite remote sensing technology, uses multispectral image processing algorithms to extract atmospheric pollutant concentration distribution information.

[0016] Preferably, the data preprocessing module includes:

[0017] The data cleaning unit uses a statistically based anomaly detection algorithm to remove sensor error data and a sliding window filtering algorithm to smooth the time series data;

[0018] Missing value filling unit, which uses the spatial interpolation method based on Kriging interpolation and the LSTM deep learning model to predict missing values ​​of time series data;

[0019] The data fusion unit uses the weighted least squares method to perform spatiotemporal weighted fusion of multi-source data to generate a high-precision initial concentration field.

[0020] Preferably, the dynamic simulation model module includes:

[0021] The Gaussian model unit simulates pollution diffusion under large-scale simple terrain conditions based on the Gaussian plume model, which specifically includes the following steps:

[0022] S1. Establish Gaussian model formula

[0023]

[0024] in:

[0025] C(x,y,z) is the pollutant concentration;

[0026] Q is the emission rate of the pollution source;

[0027] u is the wind speed, ranging from 1 to 20 m / s;

[0028] σ y ,σ z are the lateral and vertical diffusion coefficients, calculated by correction formulas according to the atmospheric stability category and distance;

[0029] y is the lateral distance perpendicular to the wind direction;

[0030] z is the vertical height;

[0031] H is the emission height of the pollution source;

[0032] exp is an exponential function, which means e x , where e is the base of natural logarithms;

[0033] S2. Determination of diffusion coefficient σ based on Pasquill-Gifford stability category y and σ z The calculation formula is:

[0034] in:

[0035] x is the downwind distance;

[0036] a y ,a z are the initial coefficients of lateral and vertical diffusion;

[0037] b y ,b z are exponential parameters for lateral and vertical diffusion;

[0038] S3. Introduction of correction factors

[0039] According to the turbulence enhancement effect, correction factors α and β are added to correct the formula:

[0040] σ′ y =α·σ y ,σ′ z =β·σ z

[0041] The value range of α and β is 1.1-1.5, which is dynamically adjusted according to the changes in wind speed and temperature;

[0042] The CFD model unit simulates pollution diffusion under complex terrain and non-steady meteorological conditions based on computational fluid dynamics, which specifically includes the following steps:

[0043] S1. Establishment of control equations

[0044] The mass conservation equation:

[0045] in:

[0046] ρ is the atmospheric density;

[0047] t is time;

[0048] is the rate of change of mass inflow or outflow;

[0049] is the fluid velocity vector;

[0050] Momentum equation:

[0051] p is pressure;

[0052] μ is the dynamic viscosity coefficient;

[0053] is the gravitational acceleration vector;

[0054] Pollutant transport equation:

[0055] in:

[0056] C is the pollutant concentration;

[0057] is the fluid velocity vector;

[0058] D is the turbulent diffusion coefficient, which is calculated by the turbulent viscosity coefficient and has a value range of 10 -3 -10 -1 m 2 / s

[0059] S is the pollution source term, that is, the emission rate of the pollution source;

[0060] S2, using -k-∈ turbulence model

[0061] The parameters are:

[0062] Turbulent kinetic energy k = 1.5·(u′)2, where u′ is the turbulence intensity, ranging from 0.1 to 0.3·u;

[0063] Turbulence dissipation rate Where l is the turbulent mixing length, which ranges from 1 to 10 m, and C μ is the turbulence model constant, C μ =0.09;

[0064] S3. Grid division

[0065] Use unstructured triangular meshes to divide complex terrain;

[0066] The grid density is increased in the high concentration gradient area, and the grid size ranges from 0.1 to 10 m;

[0067] S4. Numerical solution

[0068] The finite volume method FVM is used to discretize the control equations;

[0069] Use SIMPLE algorithm to solve pressure-velocity coupling problems;

[0070] Model switching unit, set the terrain complexity threshold Slope max =15° and wind speed change rate threshold Δu / Δt=2m / s 2 ;

[0071] When the input data meets simple terrain and stable meteorological conditions, the Gaussian model unit is enabled;

[0072] When the input data exceeds the above threshold, it automatically switches to the CFD model unit;

[0073] The parameter optimization unit uses a deep neural network DNN to optimize the diffusion coefficient. The DNN architecture includes:

[0074] Input layer: 10 input features including wind speed, wind direction, temperature and turbulence intensity;

[0075] Hidden layer: 3 layers, each layer contains 64 neurons, and the activation function is ReLU;

[0076] Output layer: Output optimized diffusion coefficient σ y and σ z ;

[0077] Training data volume: no less than 100,000 pollution event data, using Adam optimizer with a learning rate of 0.001;

[0078] Multi-pollutant coordinated diffusion unit, establish multi-component pollutant chemical reaction model:

[0079]

[0080] in:

[0081] NO x For nitrogen oxides;

[0082] VOC stands for volatile organic compounds;

[0083] O3 is ozone;

[0084] k is the chemical reaction rate constant;

[0085] The chemical reaction rate k is dynamically adjusted according to experimental data;

[0086] Simulate the range of sedimentation rates for different pollutants:

[0087] PM2.5 (0.01-0.05m / s), SO2 (0.02-0.1m / s).

[0088] Preferably, the simulation calculation module includes:

[0089] Distributed computing unit, based on Hadoop or Spark computing framework, divides the simulation task into multiple small areas for parallel computing;

[0090] Real-time computing unit, combined with GPU acceleration technology, optimizes the real-time simulation capability of pollutant diffusion through CUDA programming;

[0091] The adaptive grid division unit uses an adaptive grid refinement algorithm to dynamically adjust the grid density based on the change in pollutant concentration gradient, concentrating computing resources on areas with high concentration changes.

[0092] Preferably, the visualization module includes:

[0093] 3D dynamic display unit, a visualization platform developed based on WebGL and GIS technology, which renders the 3D distribution of pollutant concentration in real time;

[0094] The time series animation unit uses a front-end framework based on D3.js to dynamically generate time series animations of the pollutant diffusion process;

[0095] The pollution source marking unit locates the pollution source based on the reverse tracing algorithm and marks the location and diffusion range of the pollution source on the map through the geographic information system GIS.

[0096] Preferably, the decision support module includes:

[0097] The pollution control suggestion unit analyzes the simulation results and generates targeted control measures based on the multi-objective optimization algorithm;

[0098] The risk warning unit uses a classification algorithm to classify the pollution risk level of the simulation area and issue warning information;

[0099] The report generation unit uses automated template generation technology to integrate pollution diffusion paths, risk assessment results and governance recommendations into a report with both pictures and text.

[0100] In a second aspect, the present invention provides the following technical solution, a simulation method for atmospheric pollution diffusion, comprising the following steps:

[0101] S1. Data collection

[0102] Air pollution related data is collected through multi-source collection equipment, including:

[0103] S101, collecting pollutant concentrations and meteorological parameters through electrochemical sensors at ground monitoring stations;

[0104] S102, collecting pollutant concentrations at spatial locations through a laser spectrum sensor carried by a drone;

[0105] S103. Extract regional pollutant concentration distribution in multispectral images through satellite remote sensing technology;

[0106] S2. Data Preprocessing

[0107] Process the collected data, including:

[0108] S201, data cleaning, smoothing time series data based on sliding window filtering algorithm and removing outliers;

[0109] S202, data filling, using Kriging interpolation method to handle missing spatial data, and using LSTM deep learning model to complete missing data of time series;

[0110] S203, data fusion, fusing ground, mobile and remote sensing data through weighted least squares method to generate high-precision initial pollutant concentration field;

[0111] S3. Dynamic simulation modeling

[0112] The dynamic mixed diffusion model is used to simulate the diffusion of pollutants, including:

[0113] S301, dynamically switching between the Gaussian plume model and the CFD model according to terrain complexity and meteorological conditions;

[0114] S302, Gaussian plume model is used for rapid diffusion calculation under simple terrain conditions;

[0115] S303, CFD model combines finite volume method and turbulence model to simulate diffusion under complex terrain conditions with high accuracy;

[0116] S304, introducing deep learning to optimize model parameters, including diffusion coefficient and turbulent mixing parameters;

[0117] S305. Establish a multi-pollutant co-diffusion model to simulate the chemical reaction and co-diffusion process between pollutants;

[0118] S4. Simulation calculation

[0119] The diffusion model is solved using a distributed parallel computing framework, including:

[0120] S401, dividing the simulation area into multiple sub-areas, and using Hadoop or Spark framework for parallel computing;

[0121] S402, applying adaptive meshing technology in high concentration gradient areas;

[0122] S403, with the help of GPU acceleration technology, the calculation speed is optimized through CUDA programming to perform real-time pollution diffusion prediction;

[0123] S5. Results Visualization

[0124] Dynamically visualize the simulation results, including:

[0125] S501. Use WebGL technology to develop a three-dimensional visualization platform to display the spatial distribution of pollutant concentrations in real time;

[0126] S502, generating time series animation based on D3.js to dynamically present the time evolution process of pollutant diffusion;

[0127] S503. Mark the location of pollution sources and their impact range in the visualization platform in combination with GIS;

[0128] S6. Decision support

[0129] Generate pollution control suggestions and risk warning information based on simulation results, including:

[0130] S601. Use multi-objective optimization algorithms to provide targeted governance suggestions, including pollution source closure and traffic restriction measures;

[0131] S602. Use the support vector machine (SVM) model to classify pollution risks and issue health protection recommendations;

[0132] S603. Automatically generate a comprehensive report including a pollutant diffusion path map, governance recommendations and health warning information.

[0133] In a third aspect, the invention provides the following technical solution: a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned simulation method for atmospheric pollution diffusion when executing the computer program.

[0134] In a fourth aspect, the present invention provides the following technical solution: a readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned simulation method for atmospheric pollution diffusion when executed by a processor.

[0135] The present invention has the following beneficial effects:

[0136] 1. In the present invention, by introducing a dynamic mixed diffusion model, the intelligent switching between the Gaussian plume model and the CFD model is realized, which can adapt to a variety of scenarios from simple terrain to complex terrain. At the same time, the diffusion coefficient and turbulent mixing parameters are optimized by deep learning technology, so that the model has high accuracy and high adaptability under dynamic meteorological conditions and emergencies, which is significantly better than the limitations of the traditional single model.

[0137] 2. In the present invention, the system adopts distributed computing architecture and GPU acceleration technology to greatly improve computing efficiency through task decomposition and parallel processing. Combined with adaptive meshing technology, the computing grid is encrypted in high-concentration change areas to achieve refined simulation while reducing the computing burden in low-concentration areas. This method takes into account both simulation speed and accuracy, meeting the needs of real-time prediction and large-scale regional simulation.

[0138] 3. In the present invention, through the multi-pollutant synergistic diffusion model, the system can dynamically simulate the interaction between multiple pollutants (such as chemical reactions and deposition), such as the reaction process of NOx and VOC to generate ozone. This capability enables the system to provide more realistic simulation results in complex pollution scenarios, providing a scientific basis for multi-source complex pollution control.

[0139] 4. In the present invention, the system uses GIS and WebGL technology to develop a three-dimensional dynamic visualization platform, which can intuitively display the distribution of pollutant concentrations, diffusion paths and time evolution processes. Users can adjust simulation parameters through the interactive interface and view updated results in real time. This function not only improves the comprehensibility of simulation results, but also provides a clear decision-making reference for environmental management departments. BRIEF DESCRIPTION OF THE DRAWINGS

[0140] Figure 1 This is a general architecture diagram of a simulation system for atmospheric pollution diffusion proposed by the present invention;

[0141] Figure 2 This is a data acquisition module architecture diagram of a simulation system for air pollution diffusion proposed by the present invention;

[0142] Figure 3 This is a data preprocessing module architecture diagram of an atmospheric pollution diffusion simulation system proposed by the present invention;

[0143] Figure 4 This is a dynamic simulation model module architecture diagram of an atmospheric pollution diffusion simulation system proposed by the present invention;

[0144] Figure 5 This is a simulation calculation module architecture diagram of an atmospheric pollution diffusion simulation system proposed by the present invention;

[0145] Figure 6 A visualization module architecture diagram of an atmospheric pollution diffusion simulation system proposed by the present invention;

[0146] Figure 7 This is a decision support module architecture diagram of an atmospheric pollution diffusion simulation system proposed by the present invention;

[0147] Figure 8 The present invention provides a method flow chart for simulating the diffusion of atmospheric pollution. DETAILED DESCRIPTION

[0148] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0149] Embodiment 1

[0150] Reference Figure 1-Figure 7 In the first embodiment of the present invention, the present invention provides a simulation system for air pollution diffusion, including a data acquisition module, a data preprocessing module, a dynamic simulation model module, a simulation calculation module, a visualization module and a decision support module.

[0151] The data acquisition module is used to collect multi-source data including pollutant concentrations, meteorological parameters and geographic information;

[0152] The data preprocessing module is used to clean, fuse and optimize the collected data;

[0153] The dynamic simulation model module uses a dynamic mixed diffusion model and combines deep learning to optimize model parameters;

[0154] The simulation calculation module is based on a distributed computing framework to complete efficient simulation of pollutant diffusion;

[0155] The visualization module is used to dynamically display the pollutant concentration distribution and diffusion path in three dimensions;

[0156] The decision support module generates pollution control recommendations and risk warning information based on simulation results.

[0157] Specifically, this system realizes the intelligent switching between Gaussian plume model and CFD model through the innovative dynamic mixed diffusion model, which makes it have excellent adaptability under simple terrain and complex terrain, static and dynamic meteorological conditions. The diffusion coefficient and turbulent mixing parameters are optimized in real time using deep learning technology, which further improves the accuracy of the model. At the same time, the system adopts a distributed computing architecture combined with GPU acceleration technology to greatly improve the computing efficiency, and optimizes resource allocation through adaptive meshing technology to fine-tune the simulation of high-concentration change areas. Combined with the multi-pollutant collaborative diffusion model, the system can dynamically simulate the chemical reactions and interactive diffusion processes between multiple pollutants, such as the chemical conversion of NOx and VOC to ozone. This comprehensive innovative design significantly improves the accuracy, dynamic adaptability and computing efficiency of pollution diffusion simulation, and provides reliable technical support for pollution prediction and governance in complex scenarios.

[0158] The data acquisition module includes:

[0159] Ground collection units, which use electrochemical sensors to collect pollutant concentrations through fixed monitoring stations deployed in the target area;

[0160] Mobile collection unit, a laser spectrum sensor mounted on a drone, collects the concentration of trace pollutants in the air in real time and provides accurate spatial location data in combination with GPS;

[0161] The remote sensing acquisition unit, based on satellite remote sensing technology, uses multispectral image processing algorithms to extract atmospheric pollutant concentration distribution information.

[0162] Specifically, the module ensures the comprehensiveness and real-time nature of data through collaborative collection and transmission of multi-source equipment. The ground monitoring station is equipped with a variety of electrochemical sensors to collect the concentration of pollutants such as PM2.5, SO2, NOx, and record meteorological conditions (such as wind speed, wind direction, temperature and humidity). These data are uploaded to the central server via LoRa or 5G networks to ensure the stability and low latency of transmission.

[0163] In addition, the laser spectrometer carried by the drone can obtain real-time high spatial resolution pollutant concentration data, and combined with its built-in GPS module to record precise geographic location information. Remote sensing data comes from multi-spectral satellite images. Through custom image processing algorithms, the pollutant concentration distribution in a specific area is extracted, and the collected data is unified in time and space.

[0164] The data preprocessing module includes:

[0165] The data cleaning unit uses a statistically based anomaly detection algorithm to remove sensor error data and a sliding window filtering algorithm to smooth the time series data;

[0166] Missing value filling unit, which uses the spatial interpolation method based on Kriging interpolation and the LSTM deep learning model to predict missing values ​​of time series data;

[0167] The data fusion unit uses the weighted least squares method to perform spatiotemporal weighted fusion of multi-source data to generate a high-precision initial concentration field.

[0168] Specifically, this module aims to improve data quality and provide high-precision input for subsequent simulation modeling. During the data cleaning process, a sliding window filtering algorithm is used to smooth the noise data, and an outlier removal method based on box plot analysis is designed to reduce measurement errors.

[0169] For missing data, the Kriging interpolation algorithm is used to process incomplete information in space, and the time series data is predicted and completed based on the long short-term memory network (LSTM) to ensure the continuity and integrity of the data. Finally, combined with multi-source data fusion technology, the weighted least squares method is used to fuse data from different sources, and the weight coefficient is corrected to make the fusion result consistent in temporal and spatial resolution.

[0170] The dynamic simulation model module includes:

[0171] The Gaussian model unit simulates pollution diffusion under large-scale simple terrain conditions based on the Gaussian plume model, which specifically includes the following steps:

[0172] S1. Establish Gaussian model formula

[0173]

[0174] in:

[0175] C(x,y,z) is the pollutant concentration;

[0176] Q is the emission rate of the pollution source;

[0177] u is the wind speed, ranging from 1 to 20 m / s;

[0178] σ y ,σ z are the lateral and vertical diffusion coefficients, calculated by correction formulas according to the atmospheric stability category and distance;

[0179] y is the lateral distance perpendicular to the wind direction;

[0180] z is the vertical height;

[0181] H is the emission height of the pollution source;

[0182] exp is an exponential function, which means e x , where e is the base of natural logarithms;

[0183] S2. Determination of diffusion coefficient σ based on Pasquill-Gifford stability category y and σ z The calculation formula is:

[0184] in:

[0185] x is the downwind distance;

[0186] a y ,a z are the initial coefficients of lateral and vertical diffusion;

[0187] by ,b z are exponential parameters for lateral and vertical diffusion;

[0188] S3. Introduction of correction factors

[0189] According to the turbulence enhancement effect, correction factors α and β are added to correct the formula:

[0190] σ′ y =α·σ y ,σ′ z =β·σ z

[0191] The value range of α and β is 1.1-1.5, which is dynamically adjusted according to the changes in wind speed and temperature;

[0192] The CFD model unit simulates pollution diffusion under complex terrain and non-steady meteorological conditions based on computational fluid dynamics, which specifically includes the following steps:

[0193] S1. Establishment of control equations

[0194] The mass conservation equation:

[0195] in:

[0196] ρ is the atmospheric density;

[0197] t is time;

[0198] is the rate of change of mass inflow or outflow;

[0199] is the fluid velocity vector;

[0200] Momentum equation:

[0201] p is pressure;

[0202] μ is the dynamic viscosity coefficient;

[0203] is the gravitational acceleration vector;

[0204] Pollutant transport equation:

[0205] in:

[0206] C is the pollutant concentration;

[0207] is the fluid velocity vector;

[0208] D is the turbulent diffusion coefficient, which is calculated by the turbulent viscosity coefficient and has a value range of 10 -3-10 -1 m 2 / s

[0209] S is the pollution source term, that is, the emission rate of the pollution source;

[0210] S2, using -k-∈ turbulence model

[0211] The parameters are:

[0212] Turbulent kinetic energy k = 1.5·(u′)2, where u′ is the turbulence intensity, ranging from 0.1 to 0.3·u;

[0213] Turbulence dissipation rate Where l is the turbulent mixing length, which ranges from 1 to 10 m, and C μ is the turbulence model constant, C μ =0.09;

[0214] S3. Grid division

[0215] Use unstructured triangular meshes to divide complex terrain;

[0216] The grid density is increased in the high concentration gradient area, and the grid size ranges from 0.1 to 10 m;

[0217] S4. Numerical solution

[0218] The finite volume method FVM is used to discretize the control equations;

[0219] Use SIMPLE algorithm to solve pressure-velocity coupling problems;

[0220] Model switching unit, set the terrain complexity threshold Slope max =15° and wind speed change rate threshold Δu / Δt=2m / s 2 ;

[0221] When the input data meets simple terrain and stable meteorological conditions, the Gaussian model unit is enabled;

[0222] When the input data exceeds the above threshold, it automatically switches to the CFD model unit;

[0223] The parameter optimization unit uses a deep neural network DNN to optimize the diffusion coefficient. The DNN architecture includes:

[0224] Input layer: 10 input features including wind speed, wind direction, temperature and turbulence intensity;

[0225] Hidden layer: 3 layers, each layer contains 64 neurons, and the activation function is ReLU;

[0226] Output layer: Output optimized diffusion coefficient σ yand σ z ;

[0227] Training data volume: no less than 100,000 pollution event data, using Adam optimizer with a learning rate of 0.001;

[0228] Multi-pollutant coordinated diffusion unit, establish multi-component pollutant chemical reaction model:

[0229]

[0230] in:

[0231] NO x For nitrogen oxides;

[0232] VOC stands for volatile organic compounds;

[0233] O3 is ozone;

[0234] k is the chemical reaction rate constant;

[0235] The chemical reaction rate k is dynamically adjusted according to experimental data;

[0236] Simulate the range of sedimentation rates for different pollutants:

[0237] PM2.5 (0.01-0.05m / s), SO2 (0.02-0.1m / s).

[0238] Specifically, the core of this module is to dynamically select an adaptive diffusion model and improve the adaptability of the model through intelligent parameter optimization. In the Gaussian plume model, a parameter table based on the Pasquill-Gifford stability classification is constructed to correct the diffusion coefficient in real time and adapt to the diffusion calculation under different meteorological conditions.

[0239] In the CFD model part, a three-dimensional turbulent flow field model was constructed and the finite volume method (FVM) was used to discretize the control equations of pollutant diffusion. In complex terrain, terrain meshing technology was used to generate unstructured grids, and the grid density was dynamically adjusted to improve the simulation accuracy of pollutant concentration distribution. In the deep learning optimization part, a deep neural network (DNN) was designed, which input meteorological data and pollutant concentration characteristics and output optimized model parameters, significantly improving the efficiency of dynamic modeling.

[0240] The simulation calculation modules include:

[0241] Distributed computing unit, based on Hadoop or Spark computing framework, divides the simulation task into multiple small areas for parallel computing;

[0242] Real-time computing unit, combined with GPU acceleration technology, optimizes the real-time simulation capability of pollutant diffusion through CUDA programming;

[0243] The adaptive grid division unit uses an adaptive grid refinement algorithm to dynamically adjust the grid density based on the change in pollutant concentration gradient, concentrating computing resources on areas with high concentration changes.

[0244] Specifically, this module adopts a distributed computing framework in the computing architecture, performs sharding processing of large-scale data based on Hadoop, and schedules real-time simulation tasks through Spark. In terms of computing resource optimization, GPU acceleration technology is introduced, and the simulation of pollutant diffusion with high computational complexity is optimized based on CUDA programming.

[0245] In addition, to ensure the accuracy of calculation, an adaptive meshing algorithm was developed. This algorithm automatically refines the mesh in areas where the pollutant concentration changes dramatically, and reduces the mesh density in areas where the concentration distribution is flat, thereby achieving dynamic allocation and optimization of computing resources. This module can effectively support real-time pollutant diffusion simulation in a large range of complex environments

[0246] The visualization modules include:

[0247] 3D dynamic display unit, a visualization platform developed based on WebGL and GIS technology, which renders the 3D distribution of pollutant concentration in real time;

[0248] The time series animation unit uses a front-end framework based on D3.js to dynamically generate time series animations of the pollutant diffusion process;

[0249] The pollution source marking unit locates the pollution source based on the reverse tracing algorithm and marks the location and diffusion range of the pollution source on the map through the geographic information system GIS.

[0250] Specifically, the visualization module is built based on GIS technology, which can present the concentration distribution and diffusion trend of pollutants in real time. By developing a three-dimensional dynamic visualization platform with WebGL technology, users can observe the three-dimensional distribution structure of pollutant diffusion from multiple angles.

[0251] To enhance the user experience, a time series animation tool based on D3.js was developed to dynamically display the evolution of the pollutant diffusion process. In addition, the platform integrates the pollution source location function, determines the pollution source through the reverse tracking algorithm, and marks its specific location and impact range on the map. Users can also input parameters through the interactive interface, update the simulation results in real time and observe their changes.

[0252] The decision support modules include:

[0253] The pollution control suggestion unit analyzes the simulation results and generates targeted control measures based on the multi-objective optimization algorithm;

[0254] The risk warning unit uses a classification algorithm to classify the pollution risk level of the simulation area and issue warning information;

[0255] The report generation unit uses automated template generation technology to integrate pollution diffusion paths, risk assessment results and governance recommendations into a report with both pictures and text.

[0256] Specifically, this module generates governance plans and risk warning information through comprehensive analysis of simulation results. In the pollution control recommendation part, a multi-objective optimization algorithm is used to calculate the effects of different governance plans, and the optimal plan is recommended based on indicators such as cost and timeliness.

[0257] In terms of risk warning, a risk classification model based on machine learning is designed to classify polluted areas through support vector machines (SVM) and generate personalized protection recommendations based on regional health standards. The module also has an automatic report generation function that can integrate simulation results and recommendations into a document with pictures and texts, including pollution diffusion path maps, pollution source impact ranges, and health protection guidelines.

[0258] Embodiment 2:

[0259] Reference Figure 8 In a second embodiment of the present invention, the present invention provides a simulation method for atmospheric pollution diffusion, comprising the following steps:

[0260] S1. Data collection

[0261] Air pollution related data is collected through multi-source collection equipment, including:

[0262] S101, collecting pollutant concentrations and meteorological parameters through electrochemical sensors at ground monitoring stations;

[0263] S102, collecting pollutant concentrations at spatial locations through a laser spectrum sensor carried by a drone;

[0264] S103. Extract regional pollutant concentration distribution in multispectral images through satellite remote sensing technology;

[0265] S2. Data Preprocessing

[0266] Process the collected data, including:

[0267] S201, data cleaning, smoothing time series data based on sliding window filtering algorithm and removing outliers;

[0268] S202, data filling, using Kriging interpolation method to handle missing spatial data, and using LSTM deep learning model to complete missing data of time series;

[0269] S203, data fusion, fusing ground, mobile and remote sensing data through weighted least squares method to generate high-precision initial pollutant concentration field;

[0270] S3. Dynamic simulation modeling

[0271] The dynamic mixed diffusion model is used to simulate the diffusion of pollutants, including:

[0272] S301, dynamically switching between the Gaussian plume model and the CFD model according to terrain complexity and meteorological conditions;

[0273] S302, Gaussian plume model is used for rapid diffusion calculation under simple terrain conditions;

[0274] S303, CFD model combines finite volume method and turbulence model to simulate diffusion under complex terrain conditions with high accuracy;

[0275] S304, introducing deep learning to optimize model parameters, including diffusion coefficient and turbulent mixing parameters;

[0276] S305. Establish a multi-pollutant co-diffusion model to simulate the chemical reaction and co-diffusion process between pollutants;

[0277] S4. Simulation calculation

[0278] The diffusion model is solved using a distributed parallel computing framework, including:

[0279] S401, dividing the simulation area into multiple sub-areas, and using Hadoop or Spark framework for parallel computing;

[0280] S402, applying adaptive meshing technology in high concentration gradient areas;

[0281] S403, with the help of GPU acceleration technology, the calculation speed is optimized through CUDA programming to perform real-time pollution diffusion prediction;

[0282] S5. Results Visualization

[0283] Dynamically visualize the simulation results, including:

[0284] S501. Use WebGL technology to develop a three-dimensional visualization platform to display the spatial distribution of pollutant concentrations in real time;

[0285] S502, generating time series animation based on D3.js to dynamically present the time evolution process of pollutant diffusion;

[0286] S503. Mark the location of pollution sources and their impact range in the visualization platform in combination with GIS;

[0287] S6. Decision support

[0288] Generate pollution control suggestions and risk warning information based on simulation results, including:

[0289] S601. Use multi-objective optimization algorithms to provide targeted governance suggestions, including pollution source closure and traffic restriction measures;

[0290] S602. Use the support vector machine (SVM) model to classify pollution risks and issue health protection recommendations;

[0291] S603. Automatically generate a comprehensive report including a pollutant diffusion path map, governance recommendations and health warning information.

[0292] Embodiment 3

[0293] The third embodiment of the present invention is based on the same inventive concept and proposes a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the simulation method of atmospheric pollution diffusion of the above embodiment.

[0294] Embodiment 4

[0295] The fourth embodiment of the present invention is based on the same inventive concept. The present invention proposes a computer device, the terminal includes: a processor, a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory, and execute the simulation method of atmospheric pollution diffusion of the above embodiment.

[0296] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0297] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A simulation system for air pollution diffusion, comprising a data acquisition module, a data preprocessing module, a dynamic simulation model module, a simulation calculation module, a visualization module and a decision support module, characterized in that: The data acquisition module is used to collect multi-source data including pollutant concentration, meteorological parameters and geographic information; The data preprocessing module is used to clean, fuse and optimize the collected data; The dynamic simulation model module adopts a dynamic mixed diffusion model and combines deep learning to optimize model parameters; The simulation calculation module is based on a distributed computing framework to complete efficient simulation of pollutant diffusion; The visualization module is used to dynamically display the pollutant concentration distribution and diffusion path in three dimensions; The decision support module generates pollution control suggestions and risk warning information based on the simulation results.

2. The simulation system for air pollution diffusion according to claim 1, characterized in that: The data acquisition module comprises: Ground collection units, which use electrochemical sensors to collect pollutant concentrations through fixed monitoring stations deployed in the target area; Mobile collection unit, a laser spectrum sensor mounted on a drone, collects the concentration of trace pollutants in the air in real time and provides accurate spatial location data in combination with GPS; The remote sensing acquisition unit, based on satellite remote sensing technology, uses multispectral image processing algorithms to extract atmospheric pollutant concentration distribution information.

3. The simulation system for air pollution diffusion according to claim 1, characterized in that: The data preprocessing module comprises: The data cleaning unit uses a statistically based anomaly detection algorithm to remove sensor error data and a sliding window filtering algorithm to smooth the time series data; Missing value filling unit, which uses the spatial interpolation method based on Kriging interpolation and the LSTM deep learning model to predict missing values ​​of time series data; The data fusion unit uses the weighted least squares method to perform spatiotemporal weighted fusion of multi-source data to generate a high-precision initial concentration field.

4. The simulation system for air pollution diffusion according to claim 1, characterized in that: The dynamic simulation model module comprises: The Gaussian model unit simulates pollution diffusion under large-scale simple terrain conditions based on the Gaussian plume model, which specifically includes the following steps: S1. Establish Gaussian model formula in: C(x,y,z) is the pollutant concentration; Q is the emission rate of the pollution source; u is the wind speed, ranging from 1 to 20 m / s; σ y ,σ z are the lateral and vertical diffusion coefficients, calculated by correction formulas according to the atmospheric stability category and distance; y is the lateral distance perpendicular to the wind direction; z is the vertical height; H is the emission height of the pollution source; exp is an exponential function, which means e x , where e is the base of natural logarithms; S2. Determination of diffusion coefficient σ based on Pasquill-Gifford stability category y and σ z The calculation formula is: in: x is the downwind distance; a y ,a z are the initial coefficients of lateral and vertical diffusion; b y ,b z are exponential parameters for lateral and vertical diffusion; S3. Introduction of correction factors According to the turbulence enhancement effect, correction factors α and β are added to correct the formula: in y =a·s y ,s′ z =b·s z The value range of α and β is 1.1-1.5, which is dynamically adjusted according to the changes in wind speed and temperature; The CFD model unit simulates pollution diffusion under complex terrain and non-steady meteorological conditions based on computational fluid dynamics, which specifically includes the following steps: S1. Establishment of control equations The mass conservation equation: in: ρ is the atmospheric density; t is time; is the rate of change of mass inflow or outflow; is the fluid velocity vector; Momentum equation: p is pressure; μ is the dynamic viscosity coefficient; is the gravitational acceleration vector; Pollutant transport equation: in: C is the pollutant concentration; is the fluid velocity vector; D is the turbulent diffusion coefficient, which is calculated by the turbulent viscosity coefficient and has a value range of 10 -3 -10 -1 m 2 / s S is the pollution source term, that is, the emission rate of the pollution source; S2, using -k-∈ turbulence model The parameters are: Turbulent kinetic energy k = 1.5·(u′)2, where u′ is the turbulence intensity, ranging from 0.1 to 0.3·u; Turbulence dissipation rate Where l is the turbulent mixing length, which ranges from 1 to 10 m, and C μ is the turbulence model constant, C μ =0.09; S3. Grid division Use unstructured triangular meshes to divide complex terrain; The grid density is increased in the high concentration gradient area, and the grid size ranges from 0.1 to 10 m; S4. Numerical solution The finite volume method FVM is used to discretize the control equations; Use SIMPLE algorithm to solve pressure-velocity coupling problems; Model switching unit, set the terrain complexity threshold Slope max =15° and wind speed change rate threshold Δu / Δt=2m / s 2 ; When the input data meets simple terrain and stable meteorological conditions, the Gaussian model unit is enabled; When the input data exceeds the above threshold, it automatically switches to the CFD model unit; The parameter optimization unit uses a deep neural network DNN to optimize the diffusion coefficient. The DNN architecture includes: Input layer: 10 input features including wind speed, wind direction, temperature and turbulence intensity; Hidden layer: 3 layers, each layer contains 64 neurons, and the activation function is ReLU; Output layer: Output optimized diffusion coefficient σ y and σ z ; Training data volume: no less than 100,000 pollution event data, using Adam optimizer with a learning rate of 0.001; Multi-pollutant coordinated diffusion unit, establish multi-component pollutant chemical reaction model: in: NO x For nitrogen oxides; VOC stands for volatile organic compounds; O3 is ozone; k is the chemical reaction rate constant; The chemical reaction rate k is dynamically adjusted according to experimental data; Simulate the range of sedimentation rates for different pollutants: PM2.5 (0.01-0.05m / s), SO2 (0.02-0.1m / s).

5. The simulation system for air pollution diffusion according to claim 1, characterized in that: The simulation calculation module comprises: Distributed computing unit, based on Hadoop or Spark computing framework, divides the simulation task into multiple small areas for parallel computing; Real-time computing unit, combined with GPU acceleration technology, optimizes the real-time simulation capability of pollutant diffusion through CUDA programming; The adaptive grid division unit uses an adaptive grid refinement algorithm to dynamically adjust the grid density based on the change in pollutant concentration gradient, concentrating computing resources on areas with high concentration changes.

6. The simulation system for air pollution diffusion according to claim 1, characterized in that: The visualization module comprises: 3D dynamic display unit, a visualization platform developed based on WebGL and GIS technology, which renders the 3D distribution of pollutant concentration in real time; The time series animation unit uses a front-end framework based on D3.js to dynamically generate time series animations of the pollutant diffusion process; The pollution source marking unit locates the pollution source based on the reverse tracing algorithm and marks the location and diffusion range of the pollution source on the map through the geographic information system GIS.

7. The simulation system for air pollution diffusion according to claim 1, characterized in that: The decision support module includes: The pollution control suggestion unit analyzes the simulation results and generates targeted control measures based on the multi-objective optimization algorithm; The risk warning unit uses a classification algorithm to classify the pollution risk level of the simulation area and issue warning information; The report generation unit uses automated template generation technology to integrate pollution diffusion paths, risk assessment results and governance recommendations into a report with both pictures and text.

8. A simulation method for atmospheric pollution diffusion, characterized in that: A simulation system for air pollution diffusion according to any one of claims 1 to 7, comprising the following steps: S1. Data collection Air pollution related data is collected through multi-source collection equipment, including: S101, collecting pollutant concentrations and meteorological parameters through electrochemical sensors at ground monitoring stations; S102, collecting pollutant concentrations at spatial locations through a laser spectrum sensor carried by a drone; S103. Extract regional pollutant concentration distribution in multispectral images through satellite remote sensing technology; S2. Data Preprocessing Process the collected data, including: S201, data cleaning, smoothing time series data based on sliding window filtering algorithm and removing outliers; S202, data filling, using Kriging interpolation method to handle missing spatial data, and using LSTM deep learning model to complete missing data of time series; S203, data fusion, fusing ground, mobile and remote sensing data through weighted least squares method to generate high-precision initial pollutant concentration field; S3. Dynamic simulation modeling The dynamic mixed diffusion model is used to simulate the diffusion of pollutants, including: S301, dynamically switching between the Gaussian plume model and the CFD model according to terrain complexity and meteorological conditions; S302, Gaussian plume model is used for rapid diffusion calculation under simple terrain conditions; S303, CFD model combines finite volume method and turbulence model to simulate diffusion under complex terrain conditions with high accuracy; S304, introducing deep learning to optimize model parameters, including diffusion coefficient and turbulent mixing parameters; S305. Establish a multi-pollutant co-diffusion model to simulate the chemical reaction and co-diffusion process between pollutants; S4. Simulation calculation The diffusion model is solved using a distributed parallel computing framework, including: S401, dividing the simulation area into multiple sub-areas, and using Hadoop or Spark framework for parallel computing; S402, applying adaptive meshing technology in high concentration gradient areas; S403, with the help of GPU acceleration technology, the calculation speed is optimized through CUDA programming to perform real-time pollution diffusion prediction; S5. Results Visualization Dynamically visualize the simulation results, including: S501. Use WebGL technology to develop a three-dimensional visualization platform to display the spatial distribution of pollutant concentrations in real time; S502, generating time series animation based on D3.js to dynamically present the time evolution process of pollutant diffusion; S503. Mark the location of pollution sources and their impact range in the visualization platform in combination with GIS; S6. Decision support Generate pollution control suggestions and risk warning information based on simulation results, including: S601. Use multi-objective optimization algorithms to provide targeted governance suggestions, including pollution source closure and traffic restriction measures; S602. Use the support vector machine (SVM) model to classify pollution risks and issue health protection recommendations; S603. Automatically generate a comprehensive report including a pollutant diffusion path map, governance recommendations and health warning information.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method for simulating atmospheric pollution diffusion as claimed in claim 8 is implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for simulating the diffusion of atmospheric pollution as claimed in claim 8 is implemented.

Images