Simulation and prediction method for diffusion of sulfur dioxide pollution source

Abstract

The invention discloses a sulfur dioxide pollution source diffusion simulation and prediction method. The method comprises the following steps: analyzing and collecting real-time monitoring data of pollution source emission by using multispectral imaging, classifying pollution sources, identifying a concentrated area of the pollution sources, and defining the concentrated area of the pollution sources as pollution hot spots; estimating the emission intensity of a pollution source by using a reverse calculation technology, and modeling and predicting the time change of pollutant emission; integrating meteorological data, and considering the influence of terrain on a pollutant diffusion path; simulating a motion path of each unit pollutant in the air and a superimposed effect among a plurality of pollution sources; according to the method, the complete process of pollution source identification, diffusion modeling, dynamic prediction, health exposure assessment and mitigation measure optimization is realized, the innovative steps of virtual pollution cloud modeling, dynamic population exposure assessment and pollution mitigation path optimization expand the capability of traditional pollutant diffusion prediction, and it is ensured that the method is more comprehensive and accurate.

Description

Technical Field

[0001] The invention relates to the technical field of environmental prediction, and in particular to a method for simulating and predicting the diffusion of sulfur dioxide pollution sources. Background Art

[0002] The industrial distribution in the chemical park is complex. The atmospheric pollutants emitted by some industrial enterprises and the toxic and hazardous substances stored are potentially dangerous. Once an atmospheric pollution accident occurs, it will seriously affect the safety of the park and surrounding residents. Therefore, studying the distribution and variation characteristics of pollutant concentrations under different hazardous substance leakage conditions can provide important support for the assessment of the risk of sudden atmospheric pollution accidents in chemical parks, the formulation of actual management policies for enterprises, the safety protection of the atmospheric environment in chemical parks and surrounding areas, and the risk emergency response of accidents.

[0003] At present, environmental issues have attracted people's attention and have caused harm to human health, especially acid rain, which not only harms the human respiratory system, but also affects the surrounding environment, causing chain reactions and aggravating the deterioration of the environment. Acid rain is mainly caused by the emission of sulfur dioxide.

[0004] At present, the diffusion simulation of sulfur dioxide pollution sources usually uses ADMS-Urban Diffusion Model software for diffusion simulation. By applying the ADMS-Urban atmospheric diffusion model, the application of air quality simulation technology in ambient air is realized, and the transformation of urban environmental quality prediction from qualitative analysis to quantitative analysis is realized. However, the ADMS-Urban atmospheric diffusion model relies on the accuracy of input parameters (such as emission source characteristics, meteorological conditions, terrain data, etc.). The error of input data will lead to inaccurate simulation results; the simulation results are highly dependent on meteorological conditions, especially in areas with frequent weather changes. The model may not be able to accurately capture complex wind fields and changes in atmospheric conditions. In urban complex terrain and densely built areas, pollutant diffusion is significantly affected by turbulence, eddies and obstacles. Summary of the invention

[0005] The purpose of the present invention is to provide a method for simulating and predicting the diffusion of sulfur dioxide pollution sources to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above purpose, a method for simulating and predicting the diffusion of sulfur dioxide pollution sources includes the following steps:

[0007] S1. Use multispectral imaging analysis to collect real-time monitoring data of pollution source emissions, classify pollution sources, identify concentrated areas of pollution sources, and define concentrated areas of pollution sources as pollution "hot spots";

[0008] S2. Use reverse calculation techniques to estimate the emission intensity of pollution sources and model and predict the temporal changes of pollutant emissions;

[0009] S3. Integrate meteorological data and consider the impact of terrain on pollutant diffusion paths;

[0010] S4. Simulate the movement path of each unit of pollutant in the air and the superposition effect between multiple pollution sources;

[0011] S5. Refine the diffusion prediction in the target area and perform uncertainty analysis;

[0012] S6. Use dynamic population distribution data to calculate exposure to SO in different time periods 2 The regional population is analyzed and the risk map of respiratory system impact on residents in a specific area is generated through the toxic absorption coefficient regression model;

[0013] S7. Simulate the dynamic regulation scheme of pollution sources and optimize the deployment path of pollution control resources.

[0014] Preferably, in step S1, the pollution sources are classified by a convolutional neural network model to distinguish industrial emissions, traffic emissions and natural releases, specifically: input remote sensing or monitoring data, extract the characteristics of the pollution source, and the formula is:

[0015] Z = ReLU(W 1 ·X+b 1 )

[0016] Where X is the spectral feature; W 1 and b 1 are weight matrix and bias respectively; Z is the output activation value for further classification; ReLU is the activation function.

[0017] Preferably, the emission intensity based on the reverse calculation technology in step S2 is specifically: by measuring the SO 2 The concentration can be combined with the diffusion model to infer the emission intensity. The formula is:

[0018]

[0019] Where, Q is the emission intensity (kg / s); C is the measured SO 2 Concentration (μg / m 3 ); U is wind speed (m / s); H is chimney height (m); K y and K z are the horizontal and vertical diffusion coefficients, respectively.

[0020] Preferably, in step S3, wind speed, wind direction, humidity and temperature data are collected by using a high-resolution weather model, and the influence of terrain on the diffusion path of pollutants is considered using a DEM or digital elevation model. In the atmospheric boundary layer, the eddy diffusion equation is used to simulate the SO 2 The diffusion behavior of is:

[0021]

[0022] Among them, C is SO 2 Concentration (μg / m 3 );U x , U y and U z is the wind speed component; K x , K y and K z is the diffusion coefficient.

[0023] Preferably, in step S4, the movement path of each unit of sulfur dioxide pollutant in the air is simulated based on the Lagrangian particle trajectory model, and a pollution cloud superposition dynamic correction algorithm is used to simulate the superposition effect between multiple pollution sources, including the interaction and energy exchange between different pollution clouds;

[0024] Lagrangian particle trajectory model tracking SO 2 The formula for the movement of pollutant points in the air is:

[0025] x i (t+Δt)=x i (t)+U·Δt+R g

[0026] Among them, x i is the location of the polluted particle; U is the average wind speed; R g is the random eddy current disturbance component; Δt represents the time interval for updating the particle position;

[0027] The total pollution is calculated by the superposition effect of pollution cloud concentration, the formula is:

[0028]

[0029] Where N is the number of pollution sources, C i is the concentration of the ith pollution source, and (x, y, z) represents the spatial point where the pollution source is located.

[0030] Preferably, in step S5, a high-resolution grid interpolation method is used to refine the diffusion prediction in the target area, and a Bayesian prediction framework is introduced to integrate the uncertainty of meteorological, topographic and emission data into the diffusion results to generate a probability distribution;

[0031] Taking into account the uncertainty of meteorology and emissions, the probability distribution of concentration is generated, and the Bayesian prediction formula is:

[0032] P(C|Θ)=∫P(C|θ)P(θ|Θ)dθ

[0033] Among them, P(C|Θ) is the concentration distribution under given parameters; P(θ|Θ) is the uncertainty distribution of the parameters.

[0034] Preferably, in step S6, the SO of a specific population is calculated by a health exposure model. 2 Exposure dose, the formula is:

[0035]

[0036] Where E is the total exposure dose (mg); C(t) is the SO at a certain moment. 2 Concentration (μg / m 3 );IR is the inhalation rate (m 3 / h); T exp is the exposure time (h).

[0037] Preferably, in step S7, a mixed integer programming is used to simulate a dynamic control scheme of pollution sources, and a deployment path of pollution control resources is optimized based on a genetic algorithm;

[0038] Mixed integer programming formula: The objective function of minimizing emissions and governance costs,

[0039]

[0040] Constraints:

[0041] Among them, c i is the governance cost of the i-th pollution source; g j is the implementation cost of the jth policy; x i ,y j is the decision variable for whether to take action; L max The maximum deployment cost of governance resources.

[0042] Preferably, the prediction method further comprises step S8, verifying the simulation results through data from ground monitoring stations, evaluating the accuracy of the model using a root mean square error, and correcting the prediction deviation in the diffusion model in real time using a Kalman filter algorithm;

[0043] Root mean square error: Evaluates the error between the simulated concentration and the actual measured value,

[0044]

[0045] Among them, C sim，i is the simulated concentration; C obs，i is the observed concentration;

[0046] The Kalman filter update formula is:

[0047]

[0048] in, is the updated state estimate; K k is the Kalman gain; z k is the measurement value; H is the observation matrix.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] 1. Through hot spot analysis and emission modeling, the present invention can quickly identify and classify the main pollution sources, avoid resource waste, and simulate SO in real time. 2 diffusion paths and concentration distribution, assisting relevant departments in planning governance strategies efficiently; using high-resolution meteorological data, terrain impact modeling and uncertainty analysis to analyze the impact of SO 2 The key factors of diffusion are fully integrated to improve the accuracy of the model. The Lagrangian particle trajectory model and dynamic superposition effect algorithm are used to finely characterize the behavior of pollution clouds and capture the diffusion complexity of the real world.

[0051] 2. The present invention uses big data analysis methods such as mobile signals to accurately evaluate the exposure of people in different time periods and locations, providing a scientific basis for public health risk management. Based on mixed integer programming and multi-objective path optimization, it optimizes the allocation of governance equipment and human resources, reduces governance costs, and flexibly adjusts industrial and transportation emissions through time-sharing scheduling and other schemes to balance environmental governance and economic development.

[0052] 3. The present invention uses Kalman filtering and Bayesian analysis to dynamically correct the prediction deviation of the model, ensure that the simulation results are always consistent with the real-time monitoring data, and use probability distribution to predict SO 2 concentration, helping decision makers better understand and respond to possible diffusion scope and risks. Through comprehensive and dynamic diffusion simulation, it provides data-based decision-making basis and assists in formulating long-term pollution control policies. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 The module flow chart of the method of the present invention is shown in FIG. DETAILED DESCRIPTION

[0054] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0055] See also Figure 1 The present invention provides a technical solution: a method for simulating and predicting the diffusion of sulfur dioxide pollution sources, comprising the following steps:

[0056] S1. Use multispectral imaging analysis to collect real-time monitoring data of pollution source emissions, classify pollution sources through convolutional neural network models, identify concentrated areas of pollution sources, and define concentrated areas of pollution sources as pollution "hot spots". Specifically: input remote sensing or monitoring data and extract the characteristics of pollution sources. The formula is:

[0057] Z = ReLU(W 1 ·X+b 1 )

[0058] Where X is the spectral feature; W 1 and b 1 are weight matrix and bias respectively; Z is the output activation value for further classification; ReLU is the activation function;

[0059] S2. Use reverse calculation techniques to estimate the emission intensity of pollution sources, model and predict the temporal changes of pollutant emissions, and measure SO 2 The concentration can be combined with the diffusion model to infer the emission intensity. The formula is:

[0060]

[0061] Where, Q is the emission intensity (kg / s); C is the measured SO 2 Concentration (μg / m 3 ); U is wind speed (m / s); H is chimney height (m); K y and K z are the horizontal and vertical diffusion coefficients, respectively;

[0062] S3. Integrate meteorological data and consider the impact of terrain on the diffusion path of pollutants.

[0063] The wind speed, wind direction, humidity and temperature data are collected through high-resolution weather models. The DEM or digital elevation model is used to consider the impact of terrain on the diffusion path of pollutants. In the atmospheric boundary layer, the eddy diffusion equation is used to simulate SO 2 The diffusion behavior of is:

[0064]

[0065] Among them, C is SO 2 Concentration (μg / m 3 );U x , U y and U z is the wind speed component; K x , K y and K z is the diffusion coefficient;

[0066] S4, based on the Lagrangian particle trajectory model to simulate the movement path of each unit of pollutant in the air, using the pollution cloud superposition dynamic correction algorithm to simulate the superposition effect between multiple pollution sources, including the interaction and energy exchange between different pollution clouds, the Lagrangian particle trajectory model tracks SO 2 The formula for the movement of pollutant points in the air is:

[0067] x i (t+Δt)=x i (t)+U·Δt+R g

[0068] Among them, x i is the location of the polluted particle; U is the average wind speed; R g is the random eddy current disturbance component; Δt represents the time interval for updating the particle position;

[0069] The total pollution is calculated by the superposition effect of pollution cloud concentration, the formula is:

[0070]

[0071] Where N is the number of pollution sources, C i is the concentration of the i-th pollution source, (x, y, z) represents the spatial point where the pollution source is located;

[0072] S5. Use high-resolution grid interpolation to refine the diffusion prediction in the target area and perform uncertainty analysis. Introduce the Bayesian prediction framework to incorporate the uncertainty of meteorological, topographic and emission data into the diffusion results and generate a probability distribution. Considering the uncertainty of meteorology and emissions, generate a probability distribution of concentration. The Bayesian prediction formula is:

[0073] P(C|Θ)=∫P(C|θ)P(θ|Θ)dθ

[0074] Where P(C|Θ) is the concentration distribution under given parameters; P(θ|Θ) is the uncertainty distribution of the parameters;

[0075] S6. Use dynamic population distribution data to calculate exposure to SO in different time periods 2 Based on the regional population, a risk map of the respiratory system of residents in a specific area is generated through the toxic absorption coefficient regression model.

[0076] Calculation of SO for a specific population through health exposure modeling 2 Exposure dose, the formula is:

[0077]

[0078] Where E is the total exposure dose (mg); C(t) is the SO at a certain moment. 2Concentration (μg / m 3 );IR is the inhalation rate (m 3 / h); T exp is the exposure time (h);

[0079] S7. Use mixed integer programming to simulate the dynamic control scheme of pollution sources, optimize the deployment path of pollution control resources based on genetic algorithms, and the mixed integer programming formula: minimize the objective function of emissions and governance costs.

[0080]

[0081] Constraints:

[0082] Among them, c i is the governance cost of the i-th pollution source; g j is the implementation cost of the jth policy; x i ,y j is the decision variable for whether to take action, L max To manage the maximum deployment cost of resources;

[0083] S8. Verify the simulation results with data from ground monitoring stations, evaluate the accuracy of the model using the root mean square error, and use the Kalman filter algorithm to correct the prediction deviation in the diffusion model in real time;

[0084] Root mean square error: Evaluates the error between the simulated concentration and the actual measured value,

[0085]

[0086] Among them, C sim，i is the simulated concentration; C obs，i is the observed concentration;

[0087] The Kalman filter update formula is:

[0088]

[0089] in, is the updated state estimate; K k is the Kalman gain; z k is the measurement value; H is the observation matrix.

[0090] In summary: the present invention realizes the complete process of pollution source identification, diffusion modeling, dynamic prediction, health exposure assessment and optimization of mitigation measures. The innovative virtual pollution cloud modeling, dynamic population exposure assessment and pollution mitigation path optimization steps expand the capabilities of traditional pollutant diffusion prediction and ensure that it is more comprehensive and accurate.

[0091] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for simulating and predicting the diffusion of sulfur dioxide pollution sources, characterized in that: The steps include: S1. Use multispectral imaging analysis to collect real-time monitoring data of pollution source emissions, classify pollution sources, identify concentrated areas of pollution sources, and define concentrated areas of pollution sources as pollution "hot spots"; S2. Estimate the emission intensity of pollution sources and model and predict the temporal changes of pollutant emissions; S3. Integrate meteorological data and consider the impact of terrain on pollutant diffusion paths; S4. Simulate the movement path of each unit of pollutant in the air and the superposition effect between multiple pollution sources; S5. Refine the diffusion prediction in the target area and perform uncertainty analysis; S6. Use dynamic population distribution data to calculate the regional population exposed to SO2 in different time periods, and generate a risk map of the respiratory system of residents in a specific area through a toxic absorption coefficient regression model; S7. Simulate the dynamic regulation scheme of pollution sources and optimize the deployment path of pollution control resources.

2. A method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S1, the convolutional neural network model is used to classify pollution sources, distinguishing industrial emissions, traffic emissions and natural releases. Specifically, the spectral data of remote sensing or monitoring is input to extract the characteristics of the pollution source. The formula is: Z = ReLU(W1·X+b1) Among them, X is the spectral feature; W1 and b1 are the weight matrix and bias respectively; Z is the output activation value for further classification, and ReLU is the activation function.

3. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: The specific method for obtaining the emission intensity of the pollution source in step S2 is: by measuring the SO2 concentration in the downwind atmosphere, and combining the diffusion model to infer the emission intensity, the formula is: Where, Q is the emission intensity; C is the measured SO2 concentration; U is the wind speed; H is the chimney height; K y and K z are the horizontal and vertical diffusion coefficients, respectively.

4. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S3, wind speed, wind direction, humidity and temperature data are collected through a high-resolution weather model, and the influence of terrain on the diffusion path of pollutants is considered using a DEM or digital elevation model. In the atmospheric boundary layer, the diffusion behavior of SO2 is simulated using the eddy diffusion equation, and the formula is: Where, C is the SO2 concentration; U x , U y and U z is the wind speed component; K x , K y and K z is the diffusion coefficient.

5. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S4, the movement path of each unit of sulfur dioxide pollutant in the air is simulated based on the Lagrangian particle trajectory model, and a pollution cloud superposition dynamic correction algorithm is used to simulate the superposition effect between multiple pollution sources, including the interaction and energy exchange between different pollution clouds; The Lagrangian particle trajectory model tracks the movement of SO2 pollutant points in the air, and the formula is: x i (t+Δt)=x i (t)+U·Δt+R g Among them, x i is the location of the polluted particle; U is the average wind speed; R g is the random eddy current disturbance component, Δt represents the time interval for particle position update; Calculate the total pollution C through the superposition effect of pollution cloud concentration total (x,y,z), the formula is: Where N is the number of pollution sources, C i is the concentration of the ith pollution source, and (x, y, z) represents the spatial point where the pollution source is located.

6. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S5, a high-resolution grid interpolation method is used to refine the diffusion prediction in the target area, and a Bayesian prediction framework is introduced to integrate the uncertainty of meteorological, topographic and emission data into the diffusion results to generate a probability distribution; Taking into account the uncertainty of meteorology and emissions, the probability distribution of concentration is generated, and the Bayesian prediction formula is: P(C∣Θ)=∫P(C∣θ)P(θ∣Θ)dθ Among them, P(C|Θ) is the concentration distribution under given parameters; P(θ|Θ) is the uncertainty distribution of the parameters.

7. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S6, the SO2 exposure dose of a specific population is calculated using a health exposure model, and the formula is: Where E is the total exposure dose; C(t) is the SO2 concentration at a certain moment; IR is the inhalation rate; T exp For exposure time.

8. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: In step S7, a dynamic control scheme of the pollution source is simulated using mixed integer programming, and the deployment path of the pollution control resources is optimized based on the genetic algorithm; Mixed integer programming formula: The objective function of minimizing emissions and governance costs, Constraints: x i ,y j ∈{0, 1} Among them, c i is the governance cost of the i-th pollution source; g j is the implementation cost of the jth policy; x i ,y j is the decision variable for whether to take action, L max The maximum deployment cost of governance resources.

9. The method for simulating and predicting the diffusion of sulfur dioxide pollution sources according to claim 1, characterized in that: The method further includes S8, in which the simulation results are verified by data from a ground monitoring station, the accuracy of the model is evaluated by using a root mean square error, and the prediction deviation in the diffusion model is corrected in real time using a Kalman filter algorithm; Root mean square error: Evaluates the error between the simulated concentration and the actual measured value, Among them, C sim , i is the simulated concentration; C obs , i is the observed concentration; The Kalman filter update formula is: in, is the updated state estimate; K k is the Kalman gain; z k is the measurement value; H is the observation matrix.

Images