CN-121981399-A - Reduction simulation method, system and equipment for air quality emission reduction scene

CN121981399ACN 121981399 ACN121981399 ACN 121981399ACN-121981399-A

Abstract

The invention discloses a reduction simulation method, a system and equipment for air quality emission reduction scenes, belonging to the technical field of air pollution control, wherein the method comprises the steps of determining the dimension of a control factor and emission reduction proportion sampling points required by constructing a response curved surface model; designing and simulating a group of single-factor basic emission reduction scenes, calculating and deducting a predicted value of the multi-factor collaborative emission reduction scenes through combination based on a single-factor basic emission reduction scene simulation result, performing simulation verification on the total design quantity and two random multi-factor verification scenes, iteratively optimizing sampling design according to the verification result until the accuracy requirement is met, and finally constructing a response curved surface model by utilizing basic scene data and combined predicted data. According to the method, through a strategy of dimension reduction design, combination generation and verification iteration, a very small number of numerical models are used for simulating loads, a high-dimension and high-precision response relation model is constructed efficiently, and the application efficiency of the response curved surface model in rapid simulation and decision support of air pollution is improved remarkably.

Inventors

Jiang Changtan
PU QIAN
XIONG GUIHONG
ZHAO JIE
LIU QIANG
ZHANG LEI
LIU JIAOJIAO

Assignees

重庆市生态环境监测中心

Dates

Publication Date: 20260505
Application Date: 20260210

Claims (9)

1. The reduction simulation method for the air quality emission reduction scene is characterized by comprising the following steps of: S1, determining a plurality of control factors for constructing a response curved surface model and the variation range of each control factor, wherein the variation range is determined according to the emission reduction ratio; S2, generating a group of single-factor basic emission reduction scenes with only single control factors changed based on the control factors and the emission reduction proportion, and simulating the single-factor basic emission reduction scenes by using an air quality model to obtain basic simulation data; S3, calculating and generating a pollutant concentration predicted value corresponding to a multi-factor collaborative emission reduction scene with a plurality of control factors changed simultaneously through a preset combination rule based on the basic simulation scene; S4, designing a verification scenario set, wherein the verification scenario set comprises a total emission reduction verification scenario of control source factors and a multi-factor random emission reduction verification scenario comprising random combination of a plurality of control factors; s5, comparing the pollutant concentration predicted value with verification simulation data, if the verification result does not reach a preset standard, returning to the step S1 to adjust the control factor and/or the emission reduction ratio, and repeating the steps for iteration until the verification is passed; S6, after verification, constructing a response curved surface model of the regional source and species by utilizing the basic simulation data and the pollutant concentration predicted value.
2. The method of claim 1, wherein the control factor is a multi-dimensional variable, and the dimensions of the variable include at least a spatial region, a pollution source type, and a pollution factor species.
3. The method for reducing and simulating the air quality emission reduction scene according to claim 1, wherein the number of single factor base emission reduction scenes=the number of control factors, the number of control factor change samples is the number of types of emission reduction ratios.
4. The method for reducing and simulating the air quality emission reduction scene according to claim 3, wherein the calculating and generating the pollutant concentration predicted value corresponding to the multi-factor collaborative emission reduction scene in which the plurality of control factors change simultaneously according to the preset combination rule comprises: Cross-combining the single-factor basic emission reduction scenes by using an arrangement and combination method to obtain a multi-factor cooperative emission reduction scene; In each multi-factor collaborative emission reduction scene, calculating a pollutant concentration predicted value corresponding to the multi-factor collaborative emission reduction scene from basic simulation data corresponding to each single-factor basic emission reduction scene based on a linear superposition principle or a preset nonlinear weighting function.
5. The method for reducing simulation of air quality emission reduction scenes according to claim 4, wherein the number of the multi-factor collaborative emission reduction scenes is = Where a is the number of control factor change samples and t is the number of control factors.
6. The method for reducing simulation of air quality emission reduction situations according to claim 5, wherein the total emission reduction verification situation number=pollution factor-control factor variation sampling number, and the multi-factor random emission reduction verification situation number is equal to or greater than-1.
7. The method for reducing simulation of an air quality emission reduction scenario according to claim 1, wherein the constructing a response surface model of a zoning, source-classifying and species-classifying type by using the basic simulation data and the pollutant concentration predicted value comprises: Wherein Y (x, Y,) represents a response surface prediction result, Represents the R n region precursor x The emissions of the source of the type of pollution vary, Represents the R n region precursor y The emission change of the pollution source is classified by n in different areas, i is the different pollution source class, and m is the upper limit of the pollution source class.
8. A abatement simulation system for an air quality abatement scenario, comprising: The control factor screening module is used for determining a plurality of control factors for constructing the response curved surface model and the change range of each control factor, and the change range is determined according to the emission reduction ratio; The basic emission reduction scene construction module is used for generating a group of single-factor basic emission reduction scenes with only single control factors changed based on the control factors and the emission reduction proportion, and simulating the single-factor basic emission reduction scenes by using an air quality model to obtain basic simulation data; The collaborative emission reduction prediction module is used for calculating and generating a pollutant concentration predicted value corresponding to a multi-factor collaborative emission reduction scene with a plurality of control factors changed simultaneously through a preset combination rule based on the basic simulation scene; the verification scenario set construction module is used for designing a verification scenario set, wherein the verification scenario set comprises a total emission reduction verification scenario of control source factors and a multi-factor random emission reduction verification scenario comprising random combination of a plurality of control factors; The comparison verification module is used for comparing the pollutant concentration predicted value with verification simulation data, returning to the control factor screening module to adjust the control factor and/or the emission reduction ratio if the verification result does not reach the preset standard, and repeating the steps for iteration until the verification is passed; And the response surface model construction module is used for constructing a response surface model of the regional source and species by utilizing the basic simulation data and the pollutant concentration predicted value after verification is passed.
9. An electronic device comprising a memory and a processor, the memory having stored thereon computer instructions executable on the processor, wherein the processor executes a method of reducing simulation of an air quality emission reduction scenario according to any one of claims 1-7 when the computer instructions are executed by the processor.

Description

Reduction simulation method, system and equipment for air quality emission reduction scene Technical Field The invention relates to the technical field of air pollution control, in particular to a reduction simulation method, a reduction simulation system and reduction simulation equipment for air quality emission reduction situations. Background Currently, the atmospheric pollution in China is changed from single PM 2.5 pollution to composite pollution mainly comprising PM 2.5 and O 3, and modern pollution treatment is proposed to strengthen the quality standard management of the urban atmospheric environment and promote the cooperative control of PM 2.5 and O 3. There is a complex coupling relationship between PM 2.5 and O 3, which presents some difficulty in cooperative control. There are common precursor VOCs and NOx, and a great deal of research consensus shows that the synergistic emission reduction of NOx and VOCs is a key for realizing the synergistic treatment of PM 2.5 and O 3, but due to the complex nonlinear response relationship between PM 2.5 and O 3 and the precursor, the unscientific emission reduction ratio can lead to no reduction and no increase of the concentration of pollutants. In order to quantitatively describe the nonlinear relation between PM 2.5 and O 3 and the precursor, a numerical model is one of the main tools for analyzing secondary pollution generation mechanisms in common use, the atmospheric physics and atmospheric gasification processes are parameterized through a mathematical method and a computer language, and the concentration of the secondary pollutant can be accurately simulated through meteorological field driving on the basis of an atmospheric pollutant emission list, so that the nonlinear relation between PM 2.5 secondary components and O 3 and the precursor is quantified. However, the method has high requirement on computing resources, the model running period is long, a simulation result cannot be obtained in real time, the influence of the change of the environmental air quality concentration cannot be obtained in real time according to the given pollution source emission control scene, and the requirement of rapid auxiliary atmospheric environment decision support cannot be met. In order to achieve rapid simulation and prediction of atmospheric pollutants and apply them to atmospheric compounding pollution control, an atmospheric pollutant emission-air quality response surface model (RSM, response Surface Model) was established. The RSM is a simplified version of prediction model, the core of the RSM is mathematical modeling, the relation between the input parameters and the output results of the air quality model is established through a mathematical statistical method, complex data processing and physicochemical processes in the air quality model simulation are avoided, and the common modeling method comprises algorithms such as multidimensional Kriging interpolation (Multidimensional Kriging), support vector regression (Support Vector Regression), artificial neural network (ARTIFICIAL NEURAL NETWORK) and the like. The multidimensional Kriging interpolation method can realize accurate interpolation according to the simulation result of the air quality model, and simultaneously gives a confidence interval of a predicted value, which has great value for environmental management decision support. The support vector regression method is insensitive to noise and abnormal values in sample data by defining an epsilon-insensitive loss function, so that a model fitting result is more robust, the support vector regression method is very suitable for processing high-dimensional problems, and a plurality of meteorological factors and emission factors can be considered simultaneously. The artificial neural network algorithm can approach any complex nonlinear function with arbitrary precision in theory, and can capture the most subtle factor interaction. However, in order to obtain a relatively reliable nonlinear fitting result, the above methods are all required to be built on the basis of a large number of scene samples, scene data in the atmospheric environment field is derived from air quality model simulation, and the scene number also determines the simulation load of the air quality model. The number of scenes mainly depends on the number of control factors, namely the dimension of a sampling space, which is the most important key parameter for constructing a response surface. Studies have shown that as the number of control factors increases, the number of scenes increases exponentially. When the control factor number is 8, at least 113 scenes are needed to construct the nonlinear response relationship, and even more than 500 scene samples are needed to construct the higher-order stable nonlinear relationship to further optimize the fitting effect. An increase in the number of control factors will expand the sampling space, thus dramatically increasing the number