Environmental pollution risk assessment method and system, computing device and storage medium

Abstract

The invention provides an environmental pollution risk assessment method and system, computing equipment and a storage medium. The method comprises the following steps: acquiring geographic space data and environmental pollution data of a target area; determining a first pollution risk distribution diagram and a pollutant diffusion path by using a target pollutant diffusion model according to the environmental pollution data; generating a second pollution risk distribution map by using a pre-trained graph convolutional neural network in combination with the geographic space data and the pollutant diffusion path; and performing fusion analysis according to the first pollution risk distribution diagram and the second pollution risk distribution diagram to obtain an environmental pollution risk assessment result of the target area. According to the technical scheme provided by the invention, the accuracy of an environmental pollution risk assessment result is improved.

Description

Technical Field

[0001] This application relates to the technical field of environmental pollution risk assessment, and particularly to an environmental pollution risk assessment method, system, computing device, and storage medium. Background Art

[0002] In complex environmental scenarios such as industrial parks and urban agglomerations, environmental pollution risk assessment faces multiple challenges. With the intensification of industrial activities and the acceleration of urbanization, the types of pollutants are numerous and the migration paths are complex, and their diffusion is strongly coupled with meteorological conditions and geographical features. For example, pollutants may form local enrichments due to terrain barriers, or diffuse downstream along the water system, and sudden meteorological changes will significantly change the diffusion direction and range. However, existing technologies are difficult to efficiently integrate multi-source dynamic data, resulting in biases in the spatio-temporal correlation of risk assessment results.

[0003] To solve the above problems, the prior art adopts a hybrid method of diffusion model and Kriging interpolation. The specific process of this method includes: filling in the missing pollutant concentration data at monitoring points through Kriging interpolation to enhance data coverage, using a diffusion model (such as the Gaussian plume model) to simulate the diffusion path of pollutants, and generating a risk distribution map. This prior art has certain advantages in dealing with data missing and simulating pollutant diffusion.

[0004] However, the diffusion model is based on certain idealized assumptions, such as stable atmospheric conditions, uniform terrain, etc., which are often difficult to meet in actual complex environmental scenarios. Therefore, factors such as complex changes in terrain, height differences of buildings, and rapid changes in local meteorological conditions may all affect the actual diffusion path of pollutants, and ultimately there is a gap between the predicted environmental pollution risk assessment results and the actual situation. Summary of the Invention

[0005] This application provides an environmental pollution risk assessment method, system, computing device, and storage medium to solve the problem of poor accuracy of environmental pollution risk assessment results in the prior art.

[0006] In a first aspect, an embodiment of this application provides an environmental pollution risk assessment method, including: Obtaining the geospatial data and environmental pollution data of the target area; According to the environmental pollution data, using the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path; Combining the geospatial data and the pollutant diffusion path, and using a pre-trained graph convolutional neural network to generate a second pollution risk distribution map; Performing fusion analysis based on the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

[0007] Optionally, the determining the first pollution risk distribution map and the pollutant diffusion path by using the target pollutant diffusion model according to the environmental pollution data includes: Preprocessing the environmental pollution data to obtain preprocessed environmental pollution data; Selecting an initial pollutant diffusion model that matches both the geographical environment information and the pollutant type of the target area from a preset diffusion model library, and adjusting the parameters of the initial pollutant diffusion model according to the meteorological data in the preprocessed environmental pollution data to obtain the target pollutant diffusion model; According to the pollutant concentration distribution data in the preprocessed environmental pollution data, using the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area, and generating the first pollution risk distribution map and the pollutant diffusion path.

[0008] Optionally, the generating the first pollution risk distribution map and the pollutant diffusion path by using the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area according to the pollutant concentration distribution data in the preprocessed environmental pollution data includes: Based on the geospatial data, dividing the target area into multiple spatial grid units, and obtaining the initial pollutant concentration values corresponding to each spatial grid unit from the pollutant concentration distribution data; According to the initial pollutant concentration values, iteratively solving the target pollutant diffusion model at a preset time step, calculating the pollutant concentration change amount of each spatial grid unit at multiple time nodes, and updating the pollutant concentration values according to the pollutant concentration change amount; Mapping the updated pollutant concentration values to the corresponding preset risk level color scales to generate the first pollution risk distribution map; Based on the spatial gradient distribution of the updated pollutant concentration values, using the reverse particle tracking algorithm to determine the pollutant diffusion path, and the diffusion path includes the pollution source location, dynamic diffusion direction identifier, concentration contour line and diffusion range boundary.

[0009] Optionally, the adjusting the parameters of the initial pollutant diffusion model according to the meteorological data in the preprocessed environmental pollution data to obtain the target pollutant diffusion model includes: Combining the pollutant diffusion equation and the sedimentation equation into the initial pollutant diffusion model, and the parameters of the initial pollutant diffusion model include the reference diffusion coefficient and the reference sedimentation rate; Generate an adjusted diffusion coefficient according to the ratio of the wind speed data in the meteorological data to the reference wind speed, in combination with the wind speed influence factor and the reference diffusion coefficient; Generate an adjusted sedimentation rate according to the ratio of the humidity data to the temperature data in the meteorological data, in combination with the temperature-humidity coupling coefficient and the reference sedimentation rate; Combine the adjusted diffusion coefficient with the pollutant diffusion equation, and combine the adjusted sedimentation rate with the sedimentation equation to obtain a target pollutant diffusion model.

[0010] Optionally, combining the geospatial data and the pollutant diffusion path, and using a pre-trained graph convolutional neural network to generate a second pollution risk distribution map, including: Construct a graph structure including nodes and edges according to the geospatial data and the pollutant diffusion path. A node represents a spatial grid unit of the target area, an edge is a pollutant diffusion path, the node feature of the node includes the pollutant concentration value of the corresponding grid unit, and the edge feature of the edge includes a path weight coefficient; Input the graph structure into the pre-trained graph convolutional neural network, and iteratively update the node features through multiple graph convolutional layers of the pre-trained graph convolutional neural network. The weight matrix of the convolutional kernel corresponding to each graph convolutional layer is dynamically adjusted according to the dynamic diffusion direction identifier of the pollutant diffusion path; Map the node features output by the last graph convolutional layer to a preset pollution risk probability space to generate a second pollution risk distribution map including a spatio-temporal correlation risk heat map and the confidence of the pollutant diffusion path.

[0011] Optionally, dynamically adjusting the weight matrix of the convolutional kernel corresponding to multiple graph convolutional layers according to the dynamic diffusion direction identifier of the pollutant diffusion path includes: According to the dynamic diffusion direction identifier of the pollutant diffusion path, decompose the corresponding diffusion direction into a horizontal direction angle and a vertical direction angle, and generate a direction vector through sine position encoding; Calculate the attention coefficient of the edge according to the direction vector and the node feature; Combine the attention coefficient with the reference weight matrix term to generate a dynamic weight matrix; Update the weight matrix of the convolutional kernel corresponding to each graph convolutional layer based on the dynamic weight matrix.

[0012] Optionally, the fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area includes: Based on the fluctuation value of the pollutant concentration, dynamically adjust the fusion weights of the first pollution risk distribution map and the second pollution risk distribution map to obtain the comprehensive risk assessment distribution data of the target area; According to the comprehensive risk assessment distribution data and the corresponding relationship between the preset risk levels and risk assessment intervals, divide the target area into multiple sub-areas with different risk levels; Generate a dynamic report including the geographical boundaries, risk levels and corresponding treatment plans of each sub-area, and perform trustworthy storage and traceability of the dynamic report through blockchain technology.

[0013] In a second aspect, an embodiment of the present application provides an environmental pollution risk assessment system, including: An acquisition module for acquiring the geographical spatial data and environmental pollution data of the target area; A determination module for determining the first pollution risk distribution map and the pollutant diffusion path by using the target pollutant diffusion model according to the environmental pollution data; A generation module for generating a second pollution risk distribution map by using the pre-trained graph convolutional neural network in combination with the geographical spatial data and the pollutant diffusion path; A fusion analysis module for performing fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

[0014] In a third aspect, an embodiment of the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an environmental pollution risk assessment method according to any one of the first aspect.

[0015] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it implements an environmental pollution risk assessment method according to any one of the first aspect.

[0016] In the present application, an environmental pollution risk assessment method is provided, including: acquiring the geographical spatial data and environmental pollution data of the target area; according to the environmental pollution data, using the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path; combining the geographical spatial data and the pollutant diffusion path, and using the pre-trained graph convolutional neural network to generate the second pollution risk distribution map; performing fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

[0017] The technical solution of this application has the following beneficial effects: efficiently integrating multi-source dynamic data to achieve accurate deduction of the pollutant diffusion path and risk grading. By using the target pollutant diffusion model and graph convolutional neural network, it can not only accurately simulate the pollutant diffusion process, but also capture the spatio-temporal correlation in a complex environment, thereby improving the accuracy and reliability of the environmental pollution risk assessment results.

[0018] Furthermore, through the refined preprocessing of environmental pollution data, targeted selection and adjustment of diffusion model parameters, and comprehensive analysis using graph convolutional neural network, this application not only improves the prediction accuracy, but also enhances the adaptability to complex conditions (such as instantaneous strong winds or terrain obstacles) in the actual environment. This method effectively solves the dynamic change problem that is difficult to handle by traditional static models and provides a more scientific and reasonable environmental pollution risk assessment result. In addition, by introducing the reverse particle tracking algorithm to determine the pollutant diffusion path, the pollution source location and its diffusion characteristics are further refined, which helps to formulate more effective pollution control measures.

[0019] These aspects or other aspects of this application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a flowchart of an environmental pollution risk assessment method provided by an embodiment of this application; Figure 2 It is a schematic structural diagram of an environmental pollution risk assessment system provided by an embodiment of this application; Figure 3 It is a schematic structural diagram of a computing device provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application.

[0023] In some of the processes described in the specification, claims, and the above-mentioned drawings of this application, a number of operations appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The serial numbers of the operations, such as 11, 12, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any order of execution. Additionally, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this application are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0024] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0025] To address the problem that factors such as complex terrain changes, height differences in buildings, and rapid changes in local meteorological conditions may all affect the actual diffusion path of pollutants, ultimately resulting in a gap between the predicted environmental pollution risk assessment results and the actual situation, the embodiments of the present application provide an environmental pollution risk assessment method with the following concept: Integrate geospatial data and environmental pollution data to construct a multi-dimensional spatio-temporal database to solve the data heterogeneity problem in traditional methods; secondly, establish a dynamic parameter adaptation mechanism based on the pollutant diffusion model, correct the diffusion coefficient and sedimentation rate through real-time wind speed, temperature, and humidity data, and generate the first pollution risk distribution map and diffusion path; then encode the geospatial topology and diffusion path features into a graph structure, use a graph convolutional neural network to capture non-Euclidean spatial relationships, and generate the second risk distribution map; finally, fuse the outputs of the diffusion model and the graph convolutional neural network, and adjust the weight allocation in combination with the pollutant concentration fluctuation value to improve the accuracy of the environmental pollution risk assessment results for the target area.

[0026] Figure 1 The flowchart of an environmental pollution risk assessment method provided by the embodiments of the present application is as Figure 1 shown. The method includes: S11. Obtain the geospatial data and environmental pollution data of the target area.

[0027] Among them, the target area refers to the geographical space selected for analysis, evaluation, or operation based on specific research purposes, project requirements, or management needs. It usually has clear geographical boundaries, and a series of activities such as data collection, model simulation, and risk assessment are carried out within this range. Among them, the geographical boundaries can be natural boundaries (such as rivers, mountains) or artificial boundaries (such as administrative divisions, industrial park boundaries). The target area has variable scale. Specifically, it ranges from very small local areas (such as the area around a certain chemical plant) to larger areas (such as all or part of a city). The target area also has functional characteristics, which may be defined based on factors such as the integrity of the ecosystem, the location of pollution sources, and population density. Exemplarily, the target area is a chemical industrial park, a community, or a city.

[0028] And, the geospatial data includes information such as topography, landform, hydrogeology, etc. These data are used to describe the spatial structure of the target area. The environmental pollution data includes pollutant concentration distribution data and meteorological data. The pollutant concentration distribution data and meteorological data together constitute multi-source data. Among them, the pollutants can be particulate matter (PM) 2.5, volatile organic compounds (VOCs), heavy metal ions, etc. The unit of pollutant concentration is μg / m 3 , PM2.5 refers to particulate matter with an aerodynamic diameter less than 2.5 micrometers. VOCs are a class of organic chemical substances that are volatile at normal temperature, including formaldehyde, benzene, toluene, xylene, etc., and are widely present in sources such as industrial emissions, vehicle exhaust, paint, solvents, and cleaners. The pollutant concentration distribution data reflects the distribution of different pollutants in the environment, and the meteorological data includes key factors such as wind speed, humidity, and temperature that affect pollutant diffusion.

[0029] In the embodiment of the present application, geospatial data is collected through means such as remote sensing technology and ground monitoring stations; then, an environmental pollution data is obtained by using a sensor network, including the real-time concentration of various pollutants in the atmosphere and the meteorological conditions at that time. Then, these raw data are cleaned and corrected to ensure their accuracy and consistency. Finally, the processed data is integrated into a unified format as the input for subsequent steps.

[0030] Exemplarily, in a pollution assessment project of an industrial park, first, a high-resolution camera carried by a drone is used to take topographic maps of the park and its surrounding areas, and detailed geospatial data is generated in combination with existing map materials. At the same time, multiple air quality monitoring points are arranged in the park to collect in real time the concentrations of pollutants such as PM2.5 and sulfur dioxide SO2, as well as meteorological parameters such as temperature, humidity, and wind speed.

[0031] S12. According to the environmental pollution data, use the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path.

[0032] Among them, the target pollutant diffusion model can simulate the migration and transformation law of pollutants in the environmental medium through physical equations. The parameters of the target pollutant diffusion model are dynamically adjusted with meteorological data. The first pollution risk distribution map is a map that uses spatial grids as units and visualizes the pollutant concentration distribution and risk level through color gradients (such as red to yellow to green), representing the migration and transformation law of pollutants in environmental media such as air and water. The pollutant diffusion path is the pollutant source and migration trajectory determined by the reverse tracking algorithm, including elements such as the source location, the dominant diffusion direction, and the concentration gradient contour line.

[0033] S13. Combine the geospatial data and the pollutant diffusion path, and use the pre-trained graph convolutional neural network to generate the second pollution risk distribution map.

[0034] Among them, the geospatial data and the diffusion path can be integrated into a graph structure in this step and input into the graph convolutional neural network. The graph structure uses spatial grids as nodes and the diffusion path as edges. The node features include pollutant concentration and elevation data, and the edge features can be composed of the direction of the diffusion path and the concentration gradient weight. The graph convolutional neural network is a deep learning model for processing graph structure data. By aggregating adjacent node features to capture spatial dependence relationships, it can learn the non-linear association features of multi-source data. The second pollution risk distribution map can be a spatio-temporal associated risk heat map and path confidence after fusing the non-linear association features of geospatial and diffusion paths.

[0035] S14. Conduct a fusion analysis based on the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

[0036] It should be understood that compared with the independent analysis of the first pollution risk distribution map, the fusion analysis can improve the accuracy of the environmental pollution risk assessment result.

[0037] The following is a specific example. In the scenario of a VOCs leakage in a chemical industrial park and the need to evaluate the risk of the surrounding 10 km 2 area, the embodiments of the present application execute the following process: In S11, the UAV aerial photography generates a DEM with a resolution of 0.5 m, and identifies the elevation difference in the plant area (20 - 30 m); 50 monitoring points collect the VOCs concentration (0 - 500 ppb) and the southeast wind speed of 6 m / s in real time; Kriging interpolation fills in the data of 5 faulty points to generate a 100×100 grid data set. In S12, the Gaussian plume model is selected, and the initial diffusion coefficient D0 = 1.2 m 2 / s, the corrected diffusion coefficient D = 1.8 m 2 / s; The simulation shows that after 12 hours, VOCs spread to a certain community (concentration 120 ppb, 2.4 times exceeding the standard); Reverse tracking locates the leakage source as the No. 3 reactor (confidence level 95%). In S13, a graph structure is constructed. This graph structure involves 10,000 nodes, and the vector corresponding to the encoded diffusion direction (horizontal angle θ = 150°) is [0.48, -0.87, 0.09]; The pre-trained graph convolutional neural network identifies that the adsorption in the coastal wetland leads to the accumulation of local pollutant concentrations. The comprehensive risk value C final = 0.72, and 0.72 is greater than the risk threshold 0.6 corresponding to the high risk. Therefore, it is set as a red warning, triggering a level I response, and the corresponding treatment method is to stop work and guide the residents to evacuate.

[0038] By executing steps S11~S14, the embodiment of the present application constructs a standardized dataset with high resolution and high coverage by integrating geospatial data and multi-dimensional environmental pollution data, providing accurate input for the subsequent model. The target pollutant diffusion model adapts to changes in meteorological conditions by dynamically adjusting parameters (diffusion coefficient, sedimentation rate), improving the accuracy of pollutant migration simulation. The pre-trained graph convolutional neural network constructs a non-linear association between the geospatial and diffusion paths, identifying hidden risks that are difficult to detect by traditional physical models, such as underground leakage paths. Comprehensive analysis of the first pollution risk distribution map and the second pollution risk distribution map can make up for the possible limitations of a single model, thereby obtaining a more accurate and reliable environmental pollution risk assessment result. Moreover, the embodiment of the present application can implement corresponding warning and emergency response strategies for different levels of risk areas, effectively reducing the impact of environmental pollution incidents on human health.

[0039] In a possible embodiment, S12. According to the environmental pollution data, use the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path, including: Step 121. Preprocess the environmental pollution data to obtain the preprocessed environmental pollution data.

[0040] Among them, the preprocessing includes noise removal, missing value filling, and standardization processing. Therefore, the preprocessed environmental pollution data refers to the structured dataset after noise removal, missing value filling, and standardization processing, including fields such as pollutant concentrations (such as PM2.5, VOCs), meteorological parameters (wind speed, temperature, humidity), etc. Among them, noise removal is to eliminate outliers caused by abnormal fluctuations of sensors or transmission interference; Missing value filling is to complete the data lost due to equipment failures or network interruptions through interpolation or prediction methods; Standardization processing is to process data with different dimensions (such as concentration μg / m 3Map the wind speed (m / s) to a unified dimension (such as the interval [0, 1]) to eliminate the impact of numerical differences on the model.

[0041] As a possible implementation, this embodiment can adopt a sliding window mean filtering algorithm to calculate the moving average of the pollutant concentration in units of a time window (such as 10 minutes), and replace the outliers exceeding the threshold (such as ±3 times the standard deviation) within the window. For example, the PM2.5 concentration at a certain monitoring point suddenly increases to 500 μg / m 3 (normal range 0 - 200 μg / m 3 ) within 10 minutes, then the mean value of 150 μg / m 3 of other data within the window is used to replace 500 μg / m 3 . Use the Kriging interpolation algorithm to complement the missing data based on spatial correlation. Specifically, assume that the PM2.5 data at a certain monitoring point is missing due to a fault, then the weights are calculated according to the variogram (such as the spherical model) of the concentration and distance of adjacent monitoring points to generate the interpolation result. The maximum-minimum normalization method is adopted to map the pollutant concentration and meteorological parameters to the interval [0, 1] respectively. For example, the PM2.5 concentration range is 0 - 200 μg / m 3 . After calculation, 150 μg / m 3 is mapped to 0.75.

[0042] Exemplarily, among the 20 monitoring points in a chemical industrial park, the PM2.5 data at point 3 is missing due to equipment failure. The preprocessing process is as follows: The data recorded at point 2 due to instantaneous interference is 300 μg / m3 (normal value 80 - 120 μg / m 3 ), and it is corrected to 110 μg / m 3 by using the sliding window mean filtering (window = 10 minutes). For the missing PM2.5 data at point 3, based on the concentrations (100, 105, 95 μg / m 3 ) and distance weights of points 1, 4, and 5, the concentration at point 3 is obtained as 98 μg / m 3 through Kriging interpolation. The PM2.5 concentration (range 0 - 200 μg / m 3 ) is normalized to [0, 1], for example, 150 μg / m 3 is converted to 0.75.

[0043] Step 122: Select an initial pollutant diffusion model that matches both the geographical environment information and the pollutant type of the target area from the preset diffusion model library, and adjust the parameters of the initial pollutant diffusion model according to the meteorological data in the preprocessed environmental pollution data to obtain the target pollutant diffusion model.

[0044] Among them, the initial pollutant diffusion model matches the geographical environment (such as plain, mountain) and pollutant type (such as gas, particulate matter) with the benchmark model (such as Gaussian plume model, Lagrangian particle model) from the preset model library. The diffusion coefficient reflects the diffusion ability of pollutants in the medium, and the sedimentation rate characterizes the sedimentation speed of particulate matter due to gravity or adsorption.

[0045] In this step, the initial pollutant diffusion model is matched from the preset model library containing models such as the Gaussian model and the Lagrangian model, according to the geographical environment of the target area (such as selecting a model considering terrain obstruction for mountainous areas) and the pollutant type (such as selecting a particulate matter diffusion model for PM2.5).

[0046] Exemplarily, since the target area is a flat industrial park and the pollutant is particulate matter, the Gaussian plume model is selected as the initial pollutant diffusion model. Based on the ratio of the real-time wind speed (v) to the reference wind speed (v0 = 2 m / s), the diffusion coefficient is adjusted according to the formula D = D0×(v / v0). 1.2 For example, the initial diffusion coefficient D0 = 0.8 m 2 / s. When the wind speed v = 3 m / s, the corrected diffusion coefficient D = 0.8×(3 / 2) 1.2 ≈1.12 m 2 / s. According to the ratio of the temperature (T) to the humidity (RH), the sedimentation rate is adjusted according to the formula S = S0×e 0.05×(RH / T) For example, the initial S0 = 0.1 m / s. When RH = 70% and T = 25 °C, the corrected sedimentation rate S = 0.1×e (0.05×70 / 25) ≈0.13 m / s.

[0047] Step 123: According to the pollutant concentration distribution data in the preprocessed environmental pollution data, use the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area, and generate the first pollution risk distribution map and the pollutant diffusion path.

[0048] Among them, the first pollution risk distribution map is a risk level visualization map with grid cells as units, and the color gradient (such as red, yellow, green, etc.) corresponds to different concentration intervals. For example, red corresponds to the concentration interval > 100 μg / m 3 The pollutant diffusion path is the contour line of the pollution source location, the dominant diffusion direction and the concentration gradient generated by the reverse particle tracking algorithm.

[0049] In this step, the target area is divided into 100×100 grids, and the preprocessed pollutant concentration data is loaded (such as PM2.5 at point 3 = 98 μg / m 3 ). The diffusion equation ∂C / ∂t = D∇ is iteratively solved using the finite difference method. 2C - ∇×(v×C), with a time step Δt = 1 minute, simulating the spatio-temporal variation of concentration over 24 hours. In the above diffusion equation, ∂C / ∂t represents the rate of change of pollutant concentration C with respect to time t. If ∂C / ∂t > 0, it means the pollutant concentration at a certain location is increasing. D is the diffusion coefficient, characterizing the diffusion ability of a substance in a medium. The diffusion coefficient of PM2.5 in air is approximately 10 −5 m 2 / s, and the diffusion coefficient of ions in water is approximately 10 −9 m 2 / s. ∇ 2 C represents the Laplacian operator of concentration, that is, the second-order spatial derivative of concentration, used to describe the diffusion flux caused by the non-uniformity of concentration spatial distribution. Exemplarily, the diffusion rate from a high-concentration region to a low-concentration region is determined by ∇ 2 C. v is the velocity field vector, representing the flow velocity of a fluid (such as air or water). Exemplarily, a wind speed v = 3 m / s means the pollutant migrates with the air flow at a speed of 3 meters per second. ∇×(v×C) is the divergence of the product of the velocity field and concentration, representing the mass transport caused by fluid flow, and, releasing 10 3 virtual particles from a high-concentration region (such as C > 100 μg / m 5 ), calculating the particle movement trajectories inversely according to the concentration gradient to determine the pollution source location (such as the No. 3 reactor). Mapping the final concentration to a preset color scale (such as 0 - 50 μg / m 3 is green, 50 - 100 is yellow, > 100 is red), generating the first pollution risk distribution map. Gaussian model simulation shows that PM2.5 diffuses to a residential area 1.2 km downwind within 6 hours, with a concentration reaching 85 μg / m 3 (exceeding the standard by 35%). Backtracking locks the leakage source as the No. 3 reactor (confidence level 92%).

[0050] By performing steps 121 to 123, the embodiments of the present application eliminate noise and missing interference through data preprocessing, dynamically adjust model parameters to adapt to real-time meteorological conditions, and accurately simulate pollutant diffusion based on physical simulation and backtracking, finally generating a high-precision risk map and the pollution source path. In the case of PM2.5 leakage in a chemical industrial park, the preprocessing reduces the data error by 15%, the dynamically adjusted target pollutant diffusion model reduces the concentration prediction error from 25% to 8%, the backtracking accurately locates the leakage source, and supports a 40% improvement in the emergency response efficiency. This method improves the reliability and timeliness of risk assessment in complex environments.

[0051] In the above embodiment, in step 122, according to the meteorological data in the preprocessed environmental pollution data, adjusting the parameters of the initial pollutant diffusion model to obtain the target pollutant diffusion model, including: Step a1: Combine the pollutant diffusion equation and the sedimentation equation into an initial pollutant diffusion model. The parameters of the initial pollutant diffusion model include the reference diffusion coefficient and the reference sedimentation rate.

[0052] It should be understood that the pollutant diffusion equation can be ∂C / ∂t = D∇ 2 C - ∇×(v×C), or it can be other expressions other than this expression. The embodiments of the present application do not specifically limit the form of the expression of the pollutant diffusion equation. The sedimentation equation is used to describe the sedimentation behavior of particulate matter due to gravity or adsorption, and its form is ∂C / ∂t = -S×C, where S is the adjusted sedimentation rate.

[0053] It should also be understood that the reference diffusion coefficient can be understood as the initial diffusion coefficient. The reference diffusion coefficient D0 is the diffusion ability under ideal steady-state conditions. For example, the diffusion coefficient D0 of PM2.5 in a windless environment is 0.8m 2 / s. The reference sedimentation rate is the sedimentation velocity under standard temperature and humidity conditions. For example, S0 of PM2.5 at a humidity of 60% and a temperature of 20°C is 0.1m / s.

[0054] Exemplarily, the geographical environment of a chemical industrial park is a flat industrial area, and the pollutant type is PM2.5 particulate matter. Select the Gaussian plume model from the model library as the initial model, and set the reference diffusion coefficient D0 = 0.8m 2 / s and the reference sedimentation rate S0 = 0.1m / s.

[0055] Step a2: Generate an adjusted diffusion coefficient according to the ratio of the wind speed data in the meteorological data to the reference wind speed, in combination with the wind speed influence factor and the reference diffusion coefficient.

[0056] Among them, the adjusted diffusion coefficient D = D0×(v / v0) α , the wind speed influence factor α = 1.2, v0 is the reference wind speed, and v is the wind speed data. For example, the initial diffusion coefficient D0 = 0.8m 2 / s. When the wind speed v = 3m / s, the corrected diffusion coefficient D = 0.8×(3 / 2) 1.2 ≈1.12m 2 / s.

[0057] Step a3: Generate an adjusted sedimentation rate according to the ratio of the humidity data to the temperature data in the meteorological data, in combination with the temperature and humidity coupling coefficient and the reference sedimentation rate.

[0058] The adjusted sedimentation rate S = S0×e β×(RH / T) , β The temperature and humidity coupling coefficient β reflecting the synergistic effect of humidity and temperature on sedimentation is usually 0.05. Exemplarily, the corrected sedimentation rate S = 0.1×e (0.05×70 / 25) ≈0.13m / s.

[0059] Step a4: Combine the adjusted diffusion coefficient with the pollutant diffusion equation, and combine the adjusted sedimentation rate with the sedimentation equation to obtain the target pollutant diffusion model.

[0060] In this step, the parameters of the target pollutant diffusion model include the diffusion coefficient, sedimentation rate, etc. Among them, the diffusion coefficient is corrected according to the ratio of the real-time wind speed to the reference wind speed. For example, when the wind speed is 3 m / s, D = 0.8×(3 / 2) 1.2 ≈1.12 m 2 / s, and the diffusion coefficient may be adjusted from the reference value of 0.8 m 2 / s to 1.12 m 2 / s. The sedimentation rate is corrected by an exponential function in combination with the temperature and humidity changes. The target pollutant diffusion model iteratively solves the diffusion equation by the finite difference method to generate the spatio-temporal concentration distribution of the pollutant, that is, the first pollution risk distribution map. At the same time, in this step, the inverse particle tracking algorithm traces the pollution source backward from the high-concentration area to determine the diffusion path, such as the location of the leakage source, the identification of the dominant wind direction arrow, etc. Among them, the finite difference method and the inverse particle tracking algorithm are both conventional techniques, and will not be elaborated in the embodiments of the present application.

[0061] For example, in a PM2.5 leakage incident of a chemical plant, the simulation of the standard pollutant diffusion model shows that the pollutant diffuses to a residential area 1.2 km downwind within 6 hours, and the concentration is 85 μg / m 3 , the concentration exceeds the standard by 35%, a red high-risk area is generated, and the leakage source is locked as the No. 3 reactor through backward tracking.

[0062] By performing Step a1 to Step a4, the embodiments of the present application dynamically adjust the diffusion coefficient and sedimentation rate, so that the model can adapt to the changes in meteorological conditions. In the case of PM2.5 leakage in a chemical industrial park, the diffusion coefficient is corrected from the reference value of 0.8 to 1.12 m 2 / s, the sedimentation rate is corrected from 0.1 to 0.12 m / s, the prediction accuracy of the model is improved by 60%, the positioning error of the leakage source is less than 50 meters, and the efficiency of supporting emergency response is increased by 40%. This method solves the problem of insufficient adaptability of traditional static models in complex meteorological environments and provides a reliable technical basis for dynamic risk assessment.

[0063] In the above embodiment, Step 123: According to the pollutant concentration distribution data in the preprocessed environmental pollution data, use the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area, and generate the first pollution risk distribution map and the pollutant diffusion path, including: Step b1: Based on the geospatial data, divide the target area into multiple spatial grid cells, and obtain the initial pollutant concentration values corresponding to each spatial grid cell from the pollutant concentration distribution data.

[0064] In this step, this embodiment can divide the target area into equal-sized rectangular or hexagonal grids based on geospatial data (such as latitude and longitude or a plane coordinate system), with each grid representing an independent computational unit. For example, a chemical park can be divided into 100×100 grids, with each grid 10 meters long. The initial pollutant concentration value can refer to the pollutant concentration (such as PM2.5 concentration) at the center of each grid point extracted from the preprocessed data.

[0065] Specifically, based on geospatial data, a geographic information system tool is used to divide the target area into grids of specified accuracy. For example, a 1km×1km area is divided into 100×100 grids, with each grid size being 10m×10m. From the pollutant concentration distribution data, the discrete monitoring point data is mapped to the center point of each grid using bilinear interpolation. For example, the PM2.5 concentration at a certain monitoring point is 150μg / m 3 The concentration of the grid where it is located and the adjacent grids are generated by distance-weighted interpolation.

[0066] For example, the chemical park (1km×1km) is divided into 100×100 grids (each grid is 10m×10m). According to the pre-processed monitoring data, the initial PM2.5 concentration in the grid where the No. 3 reactor is located is 200μg / m 3 , the neighboring grid is interpolated according to the distance to 180μg / m 3 (east side), 160 μg / m 3 (South side) etc.

[0067] Step b2: Based on the initial pollutant concentration value, the target pollutant diffusion model is iteratively solved according to the preset time step, the pollutant concentration change of each spatial grid unit at multiple time nodes is calculated, and the pollutant concentration value is updated according to the pollutant concentration change.

[0068] The preset time step refers to the time interval between each iteration in the numerical simulation, which must meet the stability condition. For example, if Δt = 1 minute, 1440 iterations are required for a 24-hour simulation. The concentration change (ΔC) is the concentration change of each grid due to diffusion, convection, and sedimentation during the Δt time. The concentration change can be calculated as: ΔC = (D∇ 2 C − ∇ × (v × C) − S × C) Δt. The explanation of each symbol in this formula is as described above and will not be repeated here.

[0069] Specifically, for each grid, calculate the Laplace term ∇2C and the convection term ∇×(v×C), calculate ΔC according to the formula, and update the concentration value Cnew = Cold + ΔC. Repeat the above process until the preset simulation duration (such as 24 hours) is reached. Exemplarily, in the PM2.5 diffusion simulation in a chemical industrial park, the initial concentration of the grid of reactor No. 3 is 200 μg / m 3 , after iterative calculation with Δt = 1 minute, the concentration change ΔC = -5 μg / m 3 (diffusing to adjacent grids), and the updated concentration is 195 μg / m 3 . After 24 hours, the concentration of the grid 1.2 km downwind reaches 85 μg / m 3 .

[0070] Step b3: Map the updated pollutant concentration values to the corresponding preset risk level color scales to generate the first pollution risk distribution map.

[0071] It should be understood that the preset risk level color scales can map the concentration intervals to color gradients, such as red: >100 μg / m³ (high risk), yellow: 50 - 100 μg / m³ (medium risk), green: <50 μg / m³ (low risk). Convert the final concentration values of each grid to red, green, and blue color values according to the preset color scales. For example, 85 μg / m³ corresponds to yellow (255, 255, 0). Use software with rendering functions to render the color mapping results into a heat map. Exemplarily, the simulation results of the chemical industrial park show that the grids around reactor No. 3 are red (concentration >100 μg / m 3 ), and the grid of the residential area 1.2 km away is yellow (85 μg / m 3 ), and the generated first risk distribution map is released in real time through a preset platform.

[0072] Step b4: Based on the spatial gradient distribution of the updated pollutant concentration values, use the inverse particle tracking algorithm to determine the pollutant diffusion path, and the diffusion path includes the source location, dynamic diffusion direction identifier, concentration contour lines, and diffusion range boundaries.

[0073] Among them, the inverse particle tracking algorithm releases virtual particles from high-concentration regions, calculates the particle movement trajectories in reverse according to the concentration gradient, and determines the pollution source and diffusion path. The concentration contour lines are closed curves connecting grids with the same concentration values, used to identify the pollution range.

[0074] Exemplarily, release 10 in the high-risk area (red grids) of the risk distribution map 5Particles. According to the concentration gradient field, calculate the position of the particles at the previous moment according to the formula corresponding to the reverse particle tracking algorithm until tracing back to the initial moment. Mark the grids passed by more than 80% of the particles as pollution sources, and generate diffusion path arrows and concentration contour lines. Exemplarily, particles are released from the yellow grid (85 μg / m³) in the residential area. Reverse tracking shows that 80% of the particles pass through the No. 3 reactor. The diffusion path is marked as a southeast arrow, and the concentration contour line shows that the radius of the pollution range is 1.2 km.

[0075] By executing step b1 to step b4, the embodiments of the present application realize accurate prediction and traceability of pollutant diffusion through grid simulation, dynamic iteration, visualization mapping and reverse tracking. In the case of a chemical industrial park, the grid division accuracy is 10 m × 10 m, the simulation concentration error ≤ 8%, the error of reverse tracking to locate the pollution source < 50 meters, and the generated risk distribution map and diffusion path support the improvement of the emergency response efficiency.

[0076] In a possible embodiment, S13. Combine the geospatial data and the pollutant diffusion path, and use the pre-trained graph convolutional neural network to generate a second pollution risk distribution map, including: Step 131. According to the geospatial data and the pollutant diffusion path, construct a graph structure including nodes and edges. A node represents a spatial grid unit of the target area, and an edge is a pollutant diffusion path. The node features of a node include the pollutant concentration value of the corresponding grid unit, and the edge features of an edge include the path weight coefficient.

[0077] Among them, the node features include the pollutant concentration value (such as PM2.5 concentration), elevation data (in meters), land use type (classification code), etc. of the grid. The edge is used to connect the pollutant diffusion paths of two adjacent grids, and the edge features include the path weight coefficient (reflecting the diffusion intensity) and the dynamic diffusion direction identifier (such as the southeast wind direction angle of 135°). The path weight coefficient can be calculated based on the concentration gradient and distance of the diffusion path. The formula is w ij =ΔC ij / d ij , where ΔC ij is the concentration difference between adjacent grids i and j, and d ij is the grid spacing.

[0078] In this step, the 100×100 grid (each 10 m × 10 m) divided in step b1 is used as a node, and the pollutant concentration and elevation data of each node are extracted. For example, the node feature where the No. 3 reactor is located is PM2.5 = 200 μg / m 3, Elevation = 25m. If a diffusion path exists between two grids determined by the reverse particle tracking in step b4, an edge is established. For example, an edge is established between the node of the 3rd reactor and the adjacent grid node on the east side. Calculate the path weight coefficient according to the diffusion direction (such as southeast wind at 135°) and concentration gradient. For example, the concentration difference ΔC = 20 μg / m³, the distance d = 10m, and the weight coefficient w = 20 / 10 = 2.0.

[0079] Exemplarily, the graph structure constructed by the chemical industrial park contains 10,000 nodes (100×100 grids), among which the node of the 3rd reactor (node identifier 5000) establishes edges with 3 nodes on the east, south, and north sides. Among them, the weight coefficient of the east side edge is 2.0, and the direction angle is 135°; the weight coefficient of the south side edge is 1.5, and the direction angle is 180°; the weight coefficient of the north side edge is 0.8, and the direction angle is 90°.

[0080] Step 132: Input the graph structure into a pre-trained graph convolutional neural network, and iteratively update the node features through the multi-layer graph convolutional layers of the pre-trained graph convolutional neural network. The weight matrix of the convolutional kernel corresponding to each layer of the graph convolutional layer is dynamically adjusted according to the dynamic diffusion direction identifier of the pollutant diffusion path.

[0081] Among them, the node features are iteratively updated through 3 layers of graph convolution, and the formula for each layer is: ; Among them, is the feature vector of node i at the l +1 layer, representing the feature of the i th spatial grid unit in the target area after passing through the l +1 layer of graph convolution, including information such as pollutant diffusion risk and geospatial correlation. Initial layer ( l =0) features: original data such as pollutant concentration, elevation, and land use type (such as PM2.5 concentration of 200 μg / m³, elevation of 25m). Deep features: spatio-temporal correlation risk features learned through multiple rounds of graph convolution (such as risk probability of 0.73). σ is the ReLU activation function, which is used to capture the non-linear correlation between the pollutant diffusion path and geospatial data, such as a sudden increase in concentration caused by terrain mutation. Sum over the neighbor nodes j of node i, and aggregate the information of the neighbor grids associated with the diffusion paths directly connected to the current grid. N(i) includes the grids connected by the pollutant diffusion path to node i (determined by reverse particle tracking). Exemplarily: The neighbors of the node of the 3rd reactor in the chemical industrial park ( i ) include the grids on the east, south, and north sides (j1, j2, j3), corresponding to the diffusion paths respectively. α ij is the node i and the neighbor jThe attention coefficient between, W (l) : The learnable weight matrix of the l-th layer, which linearly transforms the node features and extracts high-order spatial association patterns. Exemplarily, in the chemical industrial park model, W (l) Maps the 5-dimensional input features (concentration, elevation, etc.) to 32-dimensional hidden features to capture the coupling effect of diffusion-topography.

[0082] Exemplarily, the initial features 200, 25 of node 5000 are updated to 0.72, 0.35 (normalized risk potential features) after 3 layers of GCN, and the attention coefficient 0.85 of the east side edge dominates the feature propagation.

[0083] Step 133: Map the node features output by the last graph convolutional layer to a preset pollution risk probability space to generate a second pollution risk distribution map including a spatio-temporal association risk heat map and the confidence of the pollutant diffusion path.

[0084] Among them, the pollution risk probability space can map the node features to the interval [0, 1] to represent the risk probability. For example, 0.7 means a 70% probability of high risk. The confidence of the diffusion path is based on the edge attention coefficient output by the graph convolutional neural network to calculate the credibility of the path (e.g., α = 0.85 corresponds to a confidence of 85%).

[0085] In this step, the node features output by the graph convolutional neural network (such as 0.72, 0.35) are input into the fully connected layer and Softmax to generate the risk probability. For example, the risk probability of node 5000 is 0.72 (high risk). Map the risk probability to a color gradient (red, yellow, green) to generate a spatio-temporal association risk heat map. Mark the confidence of the diffusion path according to the edge attention coefficient (e.g., 85% confidence for the east side edge). The second risk distribution map of the chemical industrial park shows that the area around reactor 3 is red (risk probability 0.72), which is consistent with the results of the traditional model; an orange area (risk probability 0.68) appears in the southeast wetland, which is uniquely detected by the graph convolutional neural network (not found in the traditional model), and the path confidence is 78%.

[0086] By performing Step 131 to Step 133, the embodiment of the present application models the complex association between the geographical space and the diffusion path through a graph structure, combines dynamic direction encoding and the attention mechanism, and identifies hidden risks (such as wetland adsorption of pollution) missed by the traditional model. In the case of the chemical industrial park, three new high-risk areas (confidence > 70%) are added to the second risk distribution map, and the evaluation result error is reduced by 32% compared with a single physical model, supporting the formulation of precise control measures.

[0087] In the above embodiment, in Step 132, dynamically adjusting the weight matrix of the convolutional kernel corresponding to the multi-layer graph convolutional layer according to the dynamic diffusion direction identifier of the pollutant diffusion path includes: Step c1: According to the dynamic diffusion direction identification of the pollutant diffusion path, decompose the corresponding diffusion direction into a horizontal direction angle and a vertical direction angle, and generate a direction vector through sine position encoding.

[0088] In this step, the graph convolutional neural network can decompose the diffusion direction into a horizontal angle θ and a vertical angle ϕ through direction encoding, and generate a 3D direction vector [sin θ , cos θ , sin ϕ . Exemplarily, the southeast direction is decomposed into a horizontal angle of 135° and a vertical angle of 0°, and then a vector is generated. The graph convolutional neural network can dynamically adjust the convolution kernel weights through the introduced attention mechanism to capture the non-linear associations of multi-source data.

[0089] For example, through the graph convolutional neural network, it is analyzed that the underground anti-seepage layer of a chemical industrial park is damaged, resulting in the diffusion of pollutants along a hidden path to the wetland area. The traditional model did not detect this risk, while the risk probability output by the graph convolutional neural network reaches 0.73, and an orange warning can be issued. Exemplarily, the dominant direction of the diffusion path in a chemical industrial park is the southeast wind ( θ = 135°, ϕ = 0°), and a direction vector [0.707, -0.707, 0] is encoded for subsequent attention calculation.

[0090] Step c2: Calculate the attention coefficient of the edge according to the direction vector and the node feature.

[0091] Calculate the attention weight of the edge according to the direction vector and the node feature. The formula is α ij = softmax(W × [h i || d ij ), where h i is the node feature, d ij is the direction vector, and || represents the concatenation operation.

[0092] Specifically, concatenate the node feature (such as PM2.5 = 200 μg / m 3 , elevation = 25 m) with the direction vector 0.707, -0.707, 0 to form an extended feature vector. Map the concatenated feature to a scalar through the learnable weight matrix W: eij = W × [h i || d ij , and use softmax to normalize the attention scores of adjacent edges. Exemplarily, the extended feature of node 5000 (the 3rd reactor) and the east side edge is 200, 25, 0.707, -0.707, 0, and after calculation by the fully connected layer, e ij=1.2, α after Softmax normalization ij =0.85, indicating that the contribution weight of this edge to node update is 85%.

[0093] Step c3: Combine the attention coefficient with the baseline weight matrix to generate a dynamic weight matrix. The baseline weight matrix W0 is the initial learnable parameter of the graph convolution layer and is used for feature transformation. dyn This approach enhances the propagation of directionally sensitive features through weights adjusted by the attention coefficient. The attention coefficient is combined with the baseline weight matrix according to the channel dimension to achieve attention weighting. Dynamic weights are scaled based on the network depth (number of layers) to ensure that each layer independently learns directional patterns. The baseline weights for the first layer are 5-dimensional input features and 32-dimensional output features. After dynamic weight adjustment, the feature propagation weight for the east edge is increased, while the weights for the other edges (south and north) are decreased.

[0094] Step c4: Update the weight matrix of the convolution kernel corresponding to each graph convolution layer based on the dynamic weight matrix.

[0095] By executing steps c1 to c4, this embodiment of the present application uses directional encoding and dynamic weight adjustment to enable the graph convolutional neural network to capture the directionally sensitive characteristics of pollutant diffusion. In the chemical park case, a southeast wind direction attention coefficient of 0.85 drove the model to identify wetland leakage paths missed by traditional physical models (with a confidence level of 78%). A risk probability of 0.73 triggered an orange alert, supporting the precise investigation of damage to underground anti-seepage layers. The dynamic weighting mechanism enabled the graph convolutional neural network to reduce modeling errors in complex diffusion patterns, validating the technical advantages of directionally adaptive modeling.

[0096] In one possible embodiment, S14, performing a fusion analysis based on the first pollution risk distribution map and the second pollution risk distribution map to obtain an environmental pollution risk assessment result for the target area includes: Step 141: Based on the pollutant concentration fluctuation value, dynamically adjust the fusion weight of the first pollution risk distribution map and the second pollution risk distribution map to obtain comprehensive risk assessment distribution data of the target area.

[0097] It should be understood that the pollutant concentration fluctuation value σ² refers to the variance of pollutant concentration within the target area, reflecting its degree of temporal or spatial fluctuation. A larger σ² indicates more dramatic concentration fluctuations and higher data uncertainty. The pollutant concentration fluctuation value σ² is obtained by calculating the variance of the concentration time series data (e.g., 24-hour monitoring values) for each grid cell.

[0098] In this step, dynamic fusion weights can be introduced during the fusion analysis process. The dynamic fusion weights include the concentration fluctuation variance σ 2The contribution weights of the adjusted diffusion model and the graph convolutional neural network. The contribution weight of the diffusion model, i.e., the weight w1 of the first pollution risk distribution map, is 1 / (1 + σ 2 ), and the contribution weight of the graph convolutional neural network, i.e., the weight w2 of the second pollution risk distribution map, is σ 2 / (1 + σ 2 ).

[0099] Exemplarily, for the area around the No. 3 reactor, σ 2 = 0.18. Calculate w1 = 0.45 and w2 = 0.55. The fused concentration C final = 0.45×0.85 + 0.55×0.73 = 0.79, which is determined to be a high risk.

[0100] Step 142: According to the comprehensive risk assessment distribution data and the corresponding relationship between the preset risk levels and the risk assessment intervals, divide the target area into multiple sub-areas with different risk levels.

[0101] Exemplarily, the risk levels are classified as low risk, medium risk, and high risk. Among them, low risk (<0.3) corresponds to green and no intervention is required; medium risk (0.3 - 0.6) corresponds to yellow and monitoring is recommended; high risk (>0.6) corresponds to red and immediate disposal is required. Optionally, use a clustering algorithm to merge adjacent grids with the same risk level into continuous sub-areas. Convert the clustering result into a vector polygon.

[0102] Exemplarily, in a chemical industrial park, for the area around the No. 3 reactor, C final = 0.79 (red), with an area of 1.2 km²; for the coastal wetland, C final = 0.68 (red), with an area of 0.5 km²; for other areas, C final < 0.3 (green). Generate 3 risk sub-areas and mark the geographical boundaries.

[0103] Step 143: Generate a dynamic report including the geographical boundaries, risk levels, and corresponding treatment plans of each sub-area, and perform trusted storage and traceability of the dynamic report through blockchain technology.

[0104] In step 143, calculate the report hash value, write the hash value and the timestamp into the blockchain, and generate a unique data fingerprint through the hash algorithm to ensure that the report cannot be tampered with.

[0105] By executing steps 141 to 143, the embodiments of the present application reduce the error of the fusion model, and the blockchain storage ensures that the evaluation results cannot be tampered with, supporting accountability and auditing; providing efficient and reliable technical support for risk assessment in complex environments.

[0106] Figure 2Schematic structural diagram of an environmental pollution risk assessment system provided by an embodiment of the present application, as Figure 2 shown. The system includes: An acquisition module 21, configured to acquire geospatial data and environmental pollution data of a target area.

[0107] A determination module 22, configured to determine a first pollution risk distribution map and a pollutant diffusion path by using a target pollutant diffusion model according to the environmental pollution data.

[0108] A generation module 23, configured to generate a second pollution risk distribution map by using a pre-trained graph convolutional neural network in combination with the geospatial data and the pollutant diffusion path.

[0109] A fusion analysis module 24, configured to perform fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain an environmental pollution risk assessment result of the target area.

[0110] Figure 2 The described environmental pollution risk assessment system can execute Figure 1 the environmental pollution risk assessment method described in the illustrated embodiment, and its implementation principle and technical effects will not be elaborated. For the environmental pollution risk assessment system in the above embodiment, the specific manners in which each module and unit perform operations have been described in detail in the embodiment related to the method, and will not be elaborated here.

[0111] In a possible design, Figure 2 the environmental pollution risk assessment system of the illustrated embodiment can be implemented as a computing device, as Figure 3 shown. The computing device may include a storage component 31 and a processing component 32.

[0112] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0113] The processing component 32 is configured to: acquire geospatial data and environmental pollution data of a target area; determine a first pollution risk distribution map and a pollutant diffusion path by using a target pollutant diffusion model according to the environmental pollution data; generate a second pollution risk distribution map by using a pre-trained graph convolutional neural network in combination with the geospatial data and the pollutant diffusion path; perform fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain an environmental pollution risk assessment result of the target area.

[0114] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above methods. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above methods.

[0115] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as random access memory (RAM), static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disks or optical discs.

[0116] Of course, the computing device may also necessarily include other components, such as input / output interfaces, display components, communication components, etc.

[0117] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module can be an output device, an input device, etc.

[0118] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0119] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device can refer to a cloud server, and the above processing component, storage component, etc. can be basic server resources leased or purchased from a cloud computing platform.

[0120] The embodiments of the present application also provide a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above-mentioned Figure 1 environmental pollution risk assessment method of the embodiments shown.

[0121] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described systems and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0122] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0123] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not intended to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. An environmental pollution risk assessment method, characterized in that, Including: Obtain the geospatial data and environmental pollution data of the target area; According to the environmental pollution data, use the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path; Combining the geospatial data and the pollutant diffusion path, use the pre-trained graph convolutional neural network to generate the second pollution risk distribution map; Conduct a fusion analysis based on the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

2. The method according to claim 1, wherein The step of using the target pollutant diffusion model to determine the first pollution risk distribution map and the pollutant diffusion path according to the environmental pollution data includes: Preprocess the environmental pollution data to obtain the preprocessed environmental pollution data; Select an initial pollutant diffusion model that matches both the geographical environment information and the pollutant type of the target area from the preset diffusion model library, and adjust the parameters of the initial pollutant diffusion model according to the meteorological data in the preprocessed environmental pollution data to obtain the target pollutant diffusion model; According to the pollutant concentration distribution data in the preprocessed environmental pollution data, use the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area, and generate the first pollution risk distribution map and the pollutant diffusion path.

3. The method according to claim 2, wherein The step of using the target pollutant diffusion model to simulate the diffusion process of pollutants in the target area according to the pollutant concentration distribution data in the preprocessed environmental pollution data and generate the first pollution risk distribution map and the pollutant diffusion path includes: Based on the geospatial data, divide the target area into multiple spatial grid units, and obtain the initial pollutant concentration values corresponding to each spatial grid unit from the pollutant concentration distribution data; According to the initial pollutant concentration values, iteratively solve the target pollutant diffusion model at a preset time step, calculate the pollutant concentration change amount of each spatial grid unit at multiple time nodes, and update the pollutant concentration value according to the pollutant concentration change amount; Map the updated pollutant concentration values to the corresponding preset risk level color scales to generate the first pollution risk distribution map; Based on the spatial gradient distribution of the updated pollutant concentration values, use the reverse particle tracking algorithm to determine the pollutant diffusion path, and the diffusion path includes the pollution source location, dynamic diffusion direction identifier, concentration contour line and diffusion range boundary.

4. The method according to claim 2, wherein The step of adjusting the parameters of the initial pollutant diffusion model according to the meteorological data in the preprocessed environmental pollution data to obtain the target pollutant diffusion model includes: Combine the pollutant diffusion equation and the sedimentation equation into the initial pollutant diffusion model, and the parameters of the initial pollutant diffusion model include the reference diffusion coefficient and the reference sedimentation rate; According to the ratio of the wind speed data in the meteorological data to the reference wind speed, combine the wind speed influence factor and the reference diffusion coefficient to generate the adjusted diffusion coefficient; According to the ratio of the humidity data to the temperature data in the meteorological data, combine the temperature and humidity coupling coefficient and the reference sedimentation rate to generate the adjusted sedimentation rate; Combine the adjusted diffusion coefficient with the pollutant diffusion equation, and combine the adjusted sedimentation rate with the sedimentation equation to obtain a target pollutant diffusion model.

5. The method according to claim 1, wherein Combining the geospatial data and the pollutant diffusion path, and using a pre-trained graph convolutional neural network to generate a second pollution risk distribution map, including: According to the geospatial data and the pollutant diffusion path, construct a graph structure including nodes and edges. One node represents a spatial grid unit of the target area, and one edge is a pollutant diffusion path. The node feature of the node includes the pollutant concentration value of the corresponding grid unit, and the edge feature of the edge includes the path weight coefficient; Input the graph structure into the pre-trained graph convolutional neural network, and iteratively update the node features through multiple graph convolutional layers of the pre-trained graph convolutional neural network. The weight matrix of the convolutional kernel corresponding to each graph convolutional layer is dynamically adjusted according to the dynamic diffusion direction identifier of the pollutant diffusion path; Map the node features output by the last graph convolutional layer to a preset pollution risk probability space to generate a second pollution risk distribution map including a spatio-temporal correlation risk heat map and the confidence of the pollutant diffusion path.

6. The method according to claim 5, wherein Dynamically adjusting the weight matrix of the convolutional kernel corresponding to the multiple graph convolutional layers according to the dynamic diffusion direction identifier of the pollutant diffusion path includes: According to the dynamic diffusion direction identifier of the pollutant diffusion path, decompose the corresponding diffusion direction into a horizontal direction angle and a vertical direction angle, and generate a direction vector through sine position encoding; Calculate the attention coefficient of the edge according to the direction vector and the node feature; Combine the attention coefficient with the benchmark weight matrix term to generate a dynamic weight matrix; Update the weight matrix of the convolutional kernel corresponding to each graph convolutional layer based on the dynamic weight matrix.

7. The method according to claim 1, characterized in that The fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area includes: Based on the pollutant concentration fluctuation value, dynamically adjust the fusion weight of the first pollution risk distribution map and the second pollution risk distribution map to obtain the comprehensive risk assessment distribution data of the target area; According to the comprehensive risk assessment distribution data and the corresponding relationship between the preset risk levels and the risk assessment intervals, divide the target area into multiple sub-areas with different risk levels; Generate a dynamic report including the geographical boundaries, risk levels and corresponding treatment plans of each sub-area, and perform trustworthy storage and traceability of the dynamic report through blockchain technology.

8. An environmental pollution risk assessment system, characterized in that, Including: An acquisition module for acquiring the geospatial data and environmental pollution data of the target area; A determination module for determining the first pollution risk distribution map and the pollutant diffusion path by using the target pollutant diffusion model according to the environmental pollution data; A generation module for combining the geospatial data and the pollutant diffusion path, and using a pre-trained graph convolutional neural network to generate a second pollution risk distribution map; A fusion analysis module for performing fusion analysis according to the first pollution risk distribution map and the second pollution risk distribution map to obtain the environmental pollution risk assessment result of the target area.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an environmental pollution risk assessment method according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that, There is a computer program stored, and when the computer program is executed by a computer, it implements an environmental pollution risk assessment method according to any one of claims 1 to 7.

Images